[bugfix]: address verify-pass findings on the 1546 fix branch

- recipe_fingerprint also excludes attention.resolved_backend (a silent
  runtime fallback must fail the gate in its cohort, not open a fresh
  ungated one); requested_backend stays in the hash; regression test
  extended to both runtime-resolved values.
- Legacy v1 configs no longer crash after the GPU measurement: identity
  fields are emitted only for config_schema_version=2 configs, and the
  comparator falls back to the legacy (model_id, gpu_type) cohort when a
  record carries no identity fields (partial identity still raises).
- 'recipe' removed from V2_OPTIONAL_METADATA_FIELDS: the generated recipe
  owns the key now; declaring it in a config fails validation loudly
  instead of being silently overwritten.
- Docs: stale 'until those land' paragraph reconciled with what this
  branch actually lands; baseline_status documented.

.buildkite/performance-benchmarks/tests/wan-t2v-1.3b.json

@@ -1,5 +1,9 @@
 {
   "benchmark_id": "wan-t2v-1.3b-2gpu",
+  "config_schema_version": 2,
+  "workload_id": "wan-t2v-1.3b",
+  "variant_id": "canonical",
+  "benchmark_version": 1,
   "description": "Wan2.1 T2V 1.3B inference performance",
   "model": {
     "model_path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",

.buildkite/pipeline.yml

@@ -114,6 +114,17 @@ steps:
             limit: 2
       agents:
         queue: "default"
+    - label: ":test_tube: LoRA Extraction Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "lora_extraction"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
     - label: ":test_tube: Training Tests"
       if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "training"
       command: "timeout 90m .buildkite/scripts/pr_test.sh"
@@ -371,6 +382,21 @@ steps:
                   - TEST_TYPE=inference_lora
                 agents:
                   queue: "default"
+            - path:
+                - "scripts/lora_extraction/**"
+                - "fastvideo/tests/lora_extraction/**"
+                - "fastvideo/models/loader/**"
+                - "fastvideo/training/training_utils.py"
+                - "fastvideo/layers/lora/**"
+                - "pyproject.toml"
+                - "docker/Dockerfile"
+              config:
+                command: "timeout 90m .buildkite/scripts/pr_test.sh"
+                label: ":test_tube: LoRA Extraction Tests"
+                env:
+                  - TEST_TYPE=lora_extraction
+                agents:
+                  queue: "default"
             - path:
                 - "fastvideo/**"
                 - "pyproject.toml"

.github/workflows/ci-slash-commands.yml

@@ -125,7 +125,7 @@ jobs:
           set -euo pipefail
           TEST_NAME=$(echo "$COMMENT" | grep -oP '(?<=/test\s)\S+' | head -1 || true)
 
-          VALID="encoder vae transformer kernel unit dreamverse ssim training lora-inference lora-training distillation self-forcing vsa vmoba performance api train-framework eval full fastcheck pre-commit"
+          VALID="encoder vae transformer kernel unit dreamverse ssim training lora-inference lora-training lora-extraction distillation self-forcing vsa vmoba performance api train-framework eval full fastcheck pre-commit"
           if [ -z "$TEST_NAME" ] || ! echo "$VALID" | grep -qw "$TEST_NAME"; then
             echo "Unknown test: '$TEST_NAME'. Valid: $VALID"
             exit 1
@@ -136,6 +136,7 @@ jobs:
             [kernel]=kernel_tests [unit]=unit_test [dreamverse]=dreamverse_app
             [ssim]=ssim [training]=training
             [lora-inference]=inference_lora [lora-training]=training_lora
+            [lora-extraction]=lora_extraction
             [distillation]=distillation_dmd [self-forcing]=self_forcing
             [vsa]=training_vsa [vmoba]=inference_vmoba
             [performance]=performance [api]=api_server

.github/workflows/infra-build-image.yml

@@ -13,12 +13,33 @@ on:
         required: false
         default: false
         type: boolean
+  # Auto-rebuild the CUDA images when their Dockerfile changes on main. The CUDA
+  # matrix is the only lane that builds from docker/Dockerfile, so a path-scoped
+  # push trigger is a sufficient change detector on its own -- no separate
+  # detect-changes/paths-filter job is needed now that there is a single
+  # in-scope Dockerfile. Dreamverse (apps/dreamverse/docker/Dockerfile) and the
+  # rocm Dockerfile stay manual-dispatch only.
+  push:
+    branches: [main]
+    paths:
+      - 'docker/Dockerfile'
 
 
 permissions:
   contents: read
   packages: write
 
+# One static group, no cancellation: every run of this workflow writes the same
+# mutable registry tags (latest, py3.12-latest, ...), so runs must serialize —
+# concurrent push/dispatch runs would race on those tags, and cancelling a run
+# mid-publish can strand the cu126/cu130 tag families at different commits. An
+# in-flight superseded build wastes its runner time, but its tags are then
+# overwritten by the newer queued run. GitHub keeps a single pending run per
+# group: the newest queued run replaces any older queued one.
+concurrency:
+  group: infra-build-image
+  cancel-in-progress: false
+
 jobs:
   # CUDA matrix: Python 3.12 x {12.6.3, 13.0.0} x {amd64, arm64}. Each architecture
   # builds natively and pushes only by digest; publish-cuda-manifests is the sole
@@ -28,7 +49,11 @@ jobs:
   # aliases; 13.0.0/cu130 is published under explicit versioned tags. Flash-attn
   # 2.8.3 comes from the architecture-specific prebuilt releases.
   build-cuda-images:
-    if: ${{ github.event.inputs.build_cuda_matrix == 'true' }}
+    # Runs on a manual dispatch when build_cuda_matrix is set, or automatically
+    # on a push that changed docker/Dockerfile (inputs are null on push). The
+    # repository guard keeps fork syncs from auto-building; manual dispatch
+    # still works in forks.
+    if: ${{ (github.event_name == 'push' && github.repository == 'hao-ai-lab/FastVideo') || github.event.inputs.build_cuda_matrix == 'true' }}
     strategy:
       fail-fast: false
       matrix:
@@ -75,10 +100,10 @@ jobs:
     secrets: inherit
 
   publish-cuda-manifests:
-    # !cancelled(): a failed sibling build leg must not skip the manifests for a
-    # CUDA lane whose own digests all exist; the digest-count check below fails
-    # the incomplete lane loudly instead.
-    if: ${{ !cancelled() && github.event.inputs.build_cuda_matrix == 'true' }}
+    # !cancelled(): publish lanes whose digests exist even if a sibling build
+    # leg failed (the digest-count check fails incomplete lanes); it also
+    # bypasses skipped-needs propagation, hence the explicit skipped check.
+    if: ${{ !cancelled() && needs.build-cuda-images.result != 'skipped' }}
     needs: build-cuda-images
     runs-on: ubuntu-latest
     permissions:

apps/dreamverse/docker/Dockerfile

@@ -84,9 +84,12 @@ RUN source /opt/venv/bin/activate \
        FFMPEG_NATIVE_CXX=/usr/bin/g++ \
        bash /opt/FastVideo/apps/dreamverse/scripts/install_native_ffmpeg.sh
 
+# FASTVIDEO_FA4: FA4 (flash_attn.cute) is opt-in; this image installs it via
+# the dreamverse extra and is validated with it, so enable it here.
 ENV FASTVIDEO_DREAMVERSE_HOME=/var/lib/dreamverse \
     STREAM_MODE=av_fmp4 \
     FASTVIDEO_ENABLE_PROMPT_SAFETY=0 \
+    FASTVIDEO_FA4=1 \
     HF_HOME=/root/.cache/huggingface
 
 RUN mkdir -p /var/lib/dreamverse

apps/performance_dashboard/README.md

@@ -12,13 +12,17 @@ Defaults:
 
 - `HF_REPO_ID=FastVideo/performance-tracking`
 - `PERFORMANCE_TRACKING_ROOT=/tmp/fastvideo-perf-dashboard`
-- `PERF_MAX_REGRESSION=0.05`
 
-Records can include source metadata:
+Records can include source metadata and rolling-baseline policy context:
 
 - `run_source`: `pr`, `local`, `scheduled_main`, or `unknown`
 - `baseline_eligible`: only successful scheduled-main records should be true
 - Buildkite metadata such as branch, PR number, build URL, build ID, and job ID
+- `regression_thresholds`: per-metric rolling-baseline percent and absolute
+  floors used for recomputed status context
+
+Dashboard/API metric payloads expose `threshold_exceeded` for raw threshold
+crossings; `regressed` remains the gated CI-failure signal.
 
 Set one of `HF_API_KEY`, `HUGGINGFACE_HUB_TOKEN`, or `HF_TOKEN` if the
 configured dataset repo requires authenticated access:
@@ -89,7 +93,8 @@ Trend charts show metric-specific axes and exact point details on hover/focus:
 - PR number, branch, and Buildkite URL when present
 
 The latest status table uses the stored JSON `success` value. Recomputed
-baseline context is shown separately and does not override stored status.
+baseline context applies each metric's percent and absolute regression floors
+and does not override stored status.
 
 ## API
 

apps/performance_dashboard/frontend/src/App.tsx

@@ -1,6 +1,7 @@
 import { useEffect, useMemo, useState } from "react";
 
-import { fetchSummary, fetchTrends, refreshData, RunSource, SummaryResponse, TrendGroup, TrendPoint } from "./api";
+import { fetchSummary, fetchTrends, refreshData } from "./api";
+import type { CohortValue, RunSource, SummaryResponse, TrendGroup, TrendPoint } from "./api";
 
 const METRIC_KEYS = ["latency", "throughput", "memory", "text_encoder_time_s", "dit_time_s", "vae_decode_time_s"];
 const RUN_SOURCES: Array<{ value: "" | RunSource; label: string }> = [
@@ -109,6 +110,61 @@ function metricLabel(metricKey: string) {
   return METRIC_DEFINITIONS[metricKey]?.label ?? metricKey;
 }
 
+type CohortFields = {
+  model_id: string;
+  gpu_type: string;
+  workload_id: CohortValue;
+  variant_id: CohortValue;
+  benchmark_version: CohortValue;
+  recipe_fingerprint: CohortValue;
+  hardware_profile_id: CohortValue;
+  software_profile_id: CohortValue;
+};
+
+function cohortValue(value: CohortValue) {
+  if (value === null || value === undefined || value === "") {
+    return "legacy";
+  }
+  return String(value);
+}
+
+function shortCohortValue(value: CohortValue) {
+  const text = cohortValue(value);
+  if (text === "legacy" || text.length <= 14) {
+    return text;
+  }
+  return text.slice(0, 12);
+}
+
+function cohortKey(cohort: CohortFields) {
+  return [
+    cohort.model_id,
+    cohort.gpu_type,
+    cohortValue(cohort.workload_id),
+    cohortValue(cohort.variant_id),
+    cohortValue(cohort.benchmark_version),
+    cohortValue(cohort.recipe_fingerprint),
+    cohortValue(cohort.hardware_profile_id),
+    cohortValue(cohort.software_profile_id)
+  ].join("|");
+}
+
+function cohortTitle(cohort: CohortFields) {
+  const workload = cohortValue(cohort.workload_id);
+  const variant = cohortValue(cohort.variant_id);
+  const version = cohortValue(cohort.benchmark_version);
+  const versionLabel = version === "legacy" ? version : `v${version}`;
+  return `${workload} / ${variant} / ${versionLabel}`;
+}
+
+function cohortDetail(cohort: CohortFields) {
+  return [
+    `recipe ${shortCohortValue(cohort.recipe_fingerprint)}`,
+    shortCohortValue(cohort.hardware_profile_id),
+    shortCohortValue(cohort.software_profile_id)
+  ].join(" | ");
+}
+
 function formatMetricValue(metricKey: string, value: number | null | undefined, tooltip = false) {
   const definition = METRIC_DEFINITIONS[metricKey];
   if (!definition) {
@@ -171,7 +227,9 @@ function TrendChart({ group, metricKey }: { group: TrendGroup; metricKey: string
         top: `${(activePoint.y / height) * 100}%`
       }
     : undefined;
-  const ariaLabel = `${metricLabel(metricKey)} trend for ${group.model_id} on ${group.gpu_type}`;
+  const ariaLabel = `${metricLabel(metricKey)} trend for ${group.model_id} on ${group.gpu_type}, ${cohortTitle(
+    group
+  )}`;
 
   return (
     <div className="chart-shell">
@@ -419,7 +477,7 @@ export default function App() {
       <section className="panel">
         <div className="panel-header">
           <h2>Latest Status</h2>
-          <span>{latestRows.length} model/GPU groups</span>
+          <span>{latestRows.length} comparison cohorts</span>
         </div>
         {latestRows.length === 0 ? (
           <div className="empty">No records match the selected filters.</div>
@@ -432,6 +490,7 @@ export default function App() {
                   <th>Recomputed</th>
                   <th>Model</th>
                   <th>GPU</th>
+                  <th>Cohort</th>
                   <th>Commit</th>
                   <th>Source</th>
                   <th>Baseline</th>
@@ -440,11 +499,13 @@ export default function App() {
                   <th>Throughput</th>
                   <th>Memory</th>
                   <th>Worst</th>
+                  <th>Exceeded</th>
+                  <th>Failing</th>
                 </tr>
               </thead>
               <tbody>
                 {latestRows.map((row) => (
-                  <tr key={`${row.model_id}-${row.gpu_type}`}>
+                  <tr key={cohortKey(row)}>
                     <td>
                       <span className={`badge ${row.status}`}>{row.status}</span>
                     </td>
@@ -455,6 +516,12 @@ export default function App() {
                     </td>
                     <td>{row.model_id}</td>
                     <td>{row.gpu_type}</td>
+                    <td>
+                      <div className="cohort-cell">
+                        <strong>{cohortTitle(row)}</strong>
+                        <span>{cohortDetail(row)}</span>
+                      </div>
+                    </td>
                     <td>{shortSha(row.commit_sha)}</td>
                     <td>
                       <span className={`source-badge source-${row.run_source}`}>{runSourceLabel(row.run_source)}</span>
@@ -465,6 +532,12 @@ export default function App() {
                     <td>{formatNumber(row.metrics.throughput?.current, 3)}</td>
                     <td>{formatNumber(row.metrics.memory?.current, 1)}</td>
                     <td>{formatNumber(row.worst_regression_pct, 1)}%</td>
+                    <td>
+                      {row.threshold_exceeded_metrics.length
+                        ? row.threshold_exceeded_metrics.join(", ")
+                        : "none"}
+                    </td>
+                    <td>{row.failing_metrics.length ? row.failing_metrics.join(", ") : "none"}</td>
                   </tr>
                 ))}
               </tbody>
@@ -487,11 +560,13 @@ export default function App() {
           ) : (
             trends.map((group) =>
               METRIC_KEYS.map((metricKey) => (
-                <article className="trend-card" key={`${group.model_id}-${group.gpu_type}-${metricKey}`}>
+                <article className="trend-card" key={`${cohortKey(group)}-${metricKey}`}>
                   <div>
                     <h3>{metricLabel(metricKey)}</h3>
                     <p>
                       {group.model_id} | {group.gpu_type}
+                      <span>{cohortTitle(group)}</span>
+                      <span>{cohortDetail(group)}</span>
                     </p>
                   </div>
                   <TrendChart group={group} metricKey={metricKey} />

apps/performance_dashboard/frontend/src/api.ts

@@ -2,11 +2,28 @@ export type MetricValue = {
   current: number | null;
   baseline: number | null;
   regression_pct: number | null;
+  absolute_delta: number | null;
+  threshold_percent: number;
+  threshold_absolute: number;
+  gated: boolean;
+  threshold_exceeded: boolean;
+  regressed: boolean;
   label: string;
   lower_is_better: boolean;
   precision: number;
 };
 
+export type CohortValue = string | number | null;
+
+export type ComparisonCohort = {
+  workload_id: CohortValue;
+  variant_id: CohortValue;
+  benchmark_version: CohortValue;
+  recipe_fingerprint: CohortValue;
+  hardware_profile_id: CohortValue;
+  software_profile_id: CohortValue;
+};
+
 export type SummaryRow = {
   model_id: string;
   gpu_type: string;
@@ -15,7 +32,8 @@ export type SummaryRow = {
   success: boolean;
   baseline_n: number;
   worst_regression_pct: number | null;
-  regression_threshold_pct: number;
+  threshold_exceeded_metrics: string[];
+  failing_metrics: string[];
   computed_regression_status: "pass" | "fail";
   status: "pass" | "fail";
   run_source: RunSource;
@@ -27,7 +45,7 @@ export type SummaryRow = {
   build_id: string;
   job_id: string;
   metrics: Record<string, MetricValue>;
-};
+} & ComparisonCohort;
 
 export type RunSource = "pr" | "local" | "scheduled_main" | "unknown";
 
@@ -61,13 +79,13 @@ export type TrendPoint = {
   build_id: string;
   job_id: string;
   metrics: Record<string, number | null>;
-};
+} & ComparisonCohort;
 
 export type TrendGroup = {
   model_id: string;
   gpu_type: string;
   points: TrendPoint[];
-};
+} & ComparisonCohort;
 
 export type TrendsResponse = {
   groups: TrendGroup[];

apps/performance_dashboard/frontend/src/styles.css

@@ -149,6 +149,11 @@ h3 {
   font-size: 0.82rem;
 }
 
+.trend-card p {
+  display: grid;
+  gap: 2px;
+}
+
 .stat strong {
   display: block;
   margin-top: 8px;
@@ -186,7 +191,7 @@ h3 {
 
 table {
   width: 100%;
-  min-width: 1120px;
+  min-width: 1260px;
   border-collapse: collapse;
 }
 
@@ -209,6 +214,25 @@ td {
   font-size: 0.9rem;
 }
 
+.cohort-cell {
+  display: grid;
+  gap: 2px;
+}
+
+.cohort-cell strong,
+.trend-card p span {
+  color: #1b2836;
+  font-size: 0.78rem;
+  font-weight: 700;
+}
+
+.cohort-cell span,
+.trend-card p span + span {
+  color: #607080;
+  font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, "Liberation Mono", monospace;
+  font-size: 0.72rem;
+}
+
 .badge {
   display: inline-flex;
   align-items: center;

docker/Dockerfile

@@ -170,12 +170,14 @@ RUN --mount=type=cache,target=/opt/uv/cache \
 # rmtree clears the wheel's stale cute files first to avoid an install conflict.
 # Then verify both survive so a broken overlay fails the build instead of shipping
 # an FA2-less image. x86 only: the FA4 stack (quack-kernels etc.) is unvalidated on
-# arm64 / GB10 (sm_121), so there we skip the overlay and FA4 falls back to FA2.
+# arm64 / GB10 (sm_121), so there we skip the overlay; FA4 is opt-in
+# (FASTVIDEO_FA4=1) and errors if set without the overlay, so leave it unset on
+# arm64 and the image runs FA3/FA2 as usual.
 RUN --mount=type=cache,target=/opt/uv/cache \
     source $HOME/.local/bin/env && \
     source /opt/venv/bin/activate && \
     if [ "${TARGETARCH}" = "arm64" ]; then \
-        echo "Skipping FA4 cute overlay on arm64 (FA4 stack unvalidated there; FA4 falls back to FA2)"; \
+        echo "Skipping FA4 cute overlay on arm64 (FA4 stack unvalidated there; do not set FASTVIDEO_FA4)"; \
     else \
         python -c "import glob, shutil; [shutil.rmtree(d, ignore_errors=True) for d in glob.glob('/opt/venv/lib/python*/site-packages/flash_attn/cute')]" && \
         uv pip install "flash-attn-4 @ git+https://github.com/Dao-AILab/flash-attention.git@${FA4_CUTE_REF}#subdirectory=flash_attn/cute" && \

docs/contributing/ci_architecture.md

@@ -103,6 +103,7 @@ can merge a PR.
 |---|---|---|
 | SSIM Tests | `ssim` | `fastvideo/**/*.py`, `pyproject.toml`, `docker/Dockerfile` |
 | LoRA Inference Tests | `inference_lora` | LoRA tests, loader, transformer tests, pipelines, LoRA layers |
+| LoRA Extraction Tests | `lora_extraction` | LoRA extraction scripts/tests, loader, training utilities, LoRA layers |
 | Training Tests | `training` | `fastvideo/**`, `pyproject.toml`, `docker/Dockerfile` |
 | Distillation DMD Tests | `distillation_dmd` | `fastvideo/training/*distillation_pipeline.py` |
 | Self-Forcing Tests | `self_forcing` | self-forcing distillation pipeline and tests |
@@ -144,6 +145,7 @@ Valid direct test names:
 | `/test training` | `training` |
 | `/test lora-inference` | `inference_lora` |
 | `/test lora-training` | `training_lora` |
+| `/test lora-extraction` | `lora_extraction` |
 | `/test distillation` | `distillation_dmd` |
 | `/test self-forcing` | `self_forcing` |
 | `/test vsa` | `training_vsa` |

docs/contributing/performance_benchmarks.md

@@ -72,7 +72,10 @@ fastvideo/tests/performance/
     │           writes Markdown summary + (optionally) uploads new records
     ├── dashboard.py
     │       └── builds time-series Plotly HTML from HF history
-    └── hf_store.py               # shared HF I/O + DataFrame helpers
+
+fastvideo/performance/
+    ├── hf_store.py               # shared HF I/O + DataFrame helpers
+    └── metric_policy.py          # shared rolling-baseline threshold policy
 ```
 
 The HF dataset (`FastVideo/performance-tracking` by default) holds one
@@ -92,16 +95,18 @@ and recipe changes instead of treating all records for a model as equivalent.
 
 ## Metrics
 
-Each benchmark records six metrics:
+Each benchmark records six metrics. The rolling-baseline comparator also has a
+per-metric policy with direction, percent threshold, absolute threshold, and a
+`gated` flag.
 
-| Metric | Raw key | Normalized key | Direction |
-|---|---|---|---|
-| End-to-end generation latency | `avg_generation_time_s` | `latency` | Lower is better |
-| Video throughput | `throughput_fps` | `throughput` | Higher is better |
-| Peak GPU memory | `max_peak_memory_mb` | `memory` | Lower is better |
-| Text encoder time | `text_encoder_time_s` | `text_encoder_time_s` | Lower is better |
-| DiT denoising time | `dit_time_s` | `dit_time_s` | Lower is better |
-| VAE decode time | `vae_decode_time_s` | `vae_decode_time_s` | Lower is better |
+| Metric | Raw key | Normalized key | Direction | Default rolling policy |
+|---|---|---|---|---|
+| End-to-end generation latency | `avg_generation_time_s` | `latency` | Lower is better | 8% and 0.5 s |
+| Video throughput | `throughput_fps` | `throughput` | Higher is better | 8% and 0.05 FPS |
+| Peak GPU memory | `max_peak_memory_mb` | `memory` | Lower is better | 5% and 256 MB |
+| Text encoder time | `text_encoder_time_s` | `text_encoder_time_s` | Lower is better | 5% and 0.25 s |
+| DiT denoising time | `dit_time_s` | `dit_time_s` | Lower is better | 5% and 0.25 s |
+| VAE decode time | `vae_decode_time_s` | `vae_decode_time_s` | Lower is better | 5% and 0.25 s |
 
 `test_inference_performance.py` temporarily sets `FASTVIDEO_STAGE_LOGGING=1`
 while it runs so pipeline stage execution times are available in
@@ -156,9 +161,22 @@ headroom and almost never need touching.
 
 `compare_baseline.py` loads the last 5 successful, baseline-eligible records
 for the same `(model_id, gpu_type)` from the HF dataset, computes the median
-for each available metric, and fails if the current run regresses by more than
-`PERF_MAX_REGRESSION` (default 5%). For latency, memory, and component times,
-higher values are regressions. For throughput, lower values are regressions.
+for each available metric, and evaluates the current run with the metric's
+rolling regression policy. For latency, memory, and component times, higher
+values are regressions. For throughput, lower values are regressions.
+
+A metric exceeds its rolling threshold when both of these are true:
+
+```text
+percent_delta > threshold_percent
+absolute_delta > threshold_absolute
+```
+
+Gated metrics fail CI when that threshold crossing happens. Set `gated: false`
+for metrics that should remain visible in reports and the dashboard without
+failing CI. Dashboard/API payloads expose `threshold_exceeded` separately from
+`regressed`, where `regressed` means a gated CI failure. Missing or `null`
+metrics are skipped.
 
 This is the **drift detector** — it catches sub-threshold regressions that
 slowly add up. Only scheduled-main successful records are baseline eligible.
@@ -172,6 +190,49 @@ agent skill to advance the rolling median.
 
 ## Schemas
 
+### Benchmark config (`.buildkite/performance-benchmarks/tests/*.json`)
+
+Benchmark configs without `config_schema_version` are treated as legacy v1
+configs and remain loadable. New or migrated configs should use
+`config_schema_version: 2` and include explicit comparable identity fields:
+
+```jsonc
+{
+  "benchmark_id": "wan-t2v-1.3b-2gpu",
+  "config_schema_version": 2,
+  "workload_id": "wan-t2v-1.3b",
+  "variant_id": "canonical",
+  "benchmark_version": 1
+}
+```
+
+`benchmark_id` is still required in this phase because raw artifact names,
+generated-video directories, normalized record paths, and the current rolling
+baseline comparator still depend on it. The v2 identity fields are config
+metadata that make the measured workload explicit:
+
+| Field | Purpose |
+|---|---|
+| `workload_id` | Stable benchmark family, such as `wan-t2v-1.3b`. |
+| `variant_id` | Intentional recipe family, such as `canonical`. |
+| `benchmark_version` | Version of the measurement protocol and comparison policy. |
+
+If a config declares `config_schema_version: 2`, loading fails clearly when any
+required v2 identity field is missing. If v2 identity or metadata fields are
+added without `config_schema_version: 2`, loading also fails so partial
+migrations do not silently run as v1 configs. Optional v2 metadata fields
+reserved for follow-up work, such as `metric_threshold_policy` and
+`quality_metadata`, must be JSON objects when present. (`recipe` is emitted
+by the harness and is not config-declarable.)
+
+Recipe fingerprinting, hardware/software profile IDs, exact-identity
+comparison, and dashboard cohort grouping land with this change: v2 records
+compare only within their identity cohort, and a record that opens a NEW
+cohort is marked `baseline_status: "initialized_new_cohort"` (regression
+gating starts once that cohort accumulates history). Legacy v1 configs keep
+the `(model_id, gpu_type)` rolling-baseline comparison. Metric-specific
+threshold policies and promoted baselines remain separate follow-ups.
+
 ### Raw record (`results/perf_*.json`)
 
 Written by `test_inference_performance.py`. One file per benchmark run.
@@ -179,6 +240,10 @@ Written by `test_inference_performance.py`. One file per benchmark run.
 ```jsonc
 {
   "benchmark_id": "wan-t2v-1.3b-2gpu",
+  "config_schema_version": 2,
+  "workload_id": "wan-t2v-1.3b",
+  "variant_id": "canonical",
+  "benchmark_version": 1,
   "model_short_name": "Wan2.1-T2V-1.3B-Diffusers",
   "device": "NVIDIA L40S",
   "num_gpus": 2,
@@ -196,12 +261,55 @@ Written by `test_inference_performance.py`. One file per benchmark run.
     "max_dit_time_s": 10.0,
     "max_vae_decode_time_s": 10.0
   },
+  "regression_thresholds": {
+    "latency": {
+      "threshold_percent": 0.10,
+      "threshold_absolute": 1.0,
+      "gated": true
+    }
+  },
   "commit": "<full sha>",
   "pr_number": "1234",
   "timestamp": "2026-05-08T22:00:00+00:00",
   "text_encoder_time_s": 2.141,
   "dit_time_s": 8.437,
-  "vae_decode_time_s": 3.208
+  "vae_decode_time_s": 3.208,
+  "recipe": {
+    "recipe_schema_version": 1,
+    "benchmark": {
+      "benchmark_id": "wan-t2v-1.3b-2gpu",
+      "workload_id": "wan-t2v-1.3b",
+      "variant_id": "canonical",
+      "benchmark_version": 1
+    },
+    "model": { "model_path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers" },
+    "init_kwargs": { "num_gpus": 2, "sp_size": 2, "tp_size": 1 },
+    "generation_kwargs": { "height": 480, "width": 832, "num_frames": 45 },
+    "inputs": { "prompt_count": 1, "prompt_sha256": ["<measured-prompt-sha256>"] },
+    "attention": { "requested_backend": "FLASH_ATTN", "resolved_backend": "FLASH_ATTN" }
+  },
+  "recipe_fingerprint": "<sha256>",
+  "hardware_profile": {
+    "device_type": "cuda",
+    "gpu_count": 2,
+    "gpus": [{ "name": "NVIDIA L40S", "memory_gb": 48, "compute_capability": "8.9" }],
+    "interconnect": "none_or_partial"
+  },
+  "hardware_profile_id": "hw-<sha256-prefix>",
+  "software_profile": {
+    "python": "3.12",
+    "pytorch": "2.12",
+    "cuda": "13.0",
+    "packages": {
+      "fastvideo_kernel": "0.3.2",
+      "flashinfer": "0.2.11",
+      "nvidia_cutlass_dsl": "4.5.0",
+      "triton": "3.4.1"
+    }
+  },
+  "software_profile_id": "sw-<sha256-prefix>",
+  "environment_metadata": { "env": { "IMAGE_VERSION": "py3.12-cuda13.0.0" } },
+  "environment_fingerprint": "env-<sha256-prefix>"
 }
 ```
 
@@ -213,6 +321,9 @@ result, used as the rolling-baseline source of truth.
 ```jsonc
 {
   "model_id": "wan-t2v-1.3b-2gpu",
+  "workload_id": "wan-t2v-1.3b",
+  "variant_id": "canonical",
+  "benchmark_version": 1,
   "timestamp": "2026-05-08T22:00:00+00:00",
   "commit_sha": "<full sha>",
   "gpu_type": "NVIDIA L40S",
@@ -222,6 +333,17 @@ result, used as the rolling-baseline source of truth.
   "text_encoder_time_s": 2.141,
   "dit_time_s": 8.437,
   "vae_decode_time_s": 3.208,
+  "regression_thresholds": {
+    "latency": {
+      "threshold_percent": 0.08,
+      "threshold_absolute": 0.5,
+      "gated": true
+    }
+  },
+  "recipe_fingerprint": "<sha256>",
+  "hardware_profile_id": "hw-<sha256-prefix>",
+  "software_profile_id": "sw-<sha256-prefix>",
+  "environment_fingerprint": "env-<sha256-prefix>",
   "success": true
 }
 ```
@@ -234,22 +356,33 @@ dashboard lists skipped plots for metric series that have no non-null values.
 Records missing both `run_source` and `baseline_eligible` are treated as legacy
 successful main/full-suite uploads and remain eligible for rolling baselines.
 
+New records compare only against the same `model_id`, `gpu_type`,
+`workload_id`, `variant_id`, `benchmark_version`, `recipe_fingerprint`,
+`hardware_profile_id`, and `software_profile_id` cohort.
+`environment_metadata` and `environment_fingerprint` are audit data and are not
+part of the comparison key.
+The recipe prompt digests describe the prompts actually measured by the
+benchmark run; extra configured prompts are ignored unless the benchmark runner
+executes them.
+Software profile package cohorts keep exact versions for relevant
+attention/kernel packages, including FastVideo kernels, FlashAttention,
+FlashInfer, Cutlass DSL, SageAttention, Triton, and xFormers when installed.
+
 ## Environment variable reference
 
 | Variable | Default | Used by | Purpose |
 |---|---|---|---|
-| `PERF_MAX_REGRESSION` | `0.05` | `compare_baseline.py` | Per-metric regression fraction that fails the build. |
 | `PERFORMANCE_TRACKING_ROOT` | `/tmp/perf-tracking` | `compare_baseline.py`, `dashboard.py` | Local directory the HF dataset is synced to. |
 | `PERF_REPORTS_DIR` | `/root/data/perf_reports` | `compare_baseline.py`, `dashboard.py` | Where the Markdown summary and Plotly HTML get written for Buildkite to pick up. |
-| `HF_REPO_ID` | `FastVideo/performance-tracking` | `hf_store.py` | HF dataset repo holding rolling-baseline records. |
-| `HF_API_KEY`, `HUGGINGFACE_HUB_TOKEN`, `HF_TOKEN` | unset | `hf_store.py` | Required for upload or private dataset reads. |
+| `HF_REPO_ID` | `FastVideo/performance-tracking` | `fastvideo/performance/hf_store.py` | HF dataset repo holding rolling-baseline records. |
+| `HF_API_KEY`, `HUGGINGFACE_HUB_TOKEN`, `HF_TOKEN` | unset | `fastvideo/performance/hf_store.py` | Required for upload or private dataset reads. |
 | `PERF_RUN_SOURCE` | inferred | `compare_baseline.py` | Source metadata for uploaded records: `pr`, `local`, `scheduled_main`, or `unknown`. |
 | `PERF_UPLOAD_POLICY` | `never` | `compare_baseline.py` | Upload policy: `never`, `pass`, or `always`. |
 | `PERF_PYTEST_RC` | unset | `compare_baseline.py` | Static-threshold pytest exit code, used so scheduled-main failures can be uploaded with `success=false`. |
 | `TEST_SCOPE` | unset | `compare_baseline.py` | CI context used to infer scheduled-main runs together with `BUILDKITE_BRANCH=main`. |
 | `BUILDKITE_BRANCH`, `BUILDKITE_COMMIT`, `BUILDKITE_PULL_REQUEST` | unset | `compare_baseline.py`, `test_inference_performance.py` | CI metadata stamped into records. |
 | `DASHBOARD_DAYS` | `30` | `dashboard.py` | Lookback window for the Plotly trend pages. |
-| `PERFORMANCE_TRACKING_SYNC_REUSE_TTL_SECONDS` | `3600` | `hf_store.py` | Freshness window for reusing an existing HF sync when requested by dashboard consumers. |
+| `PERFORMANCE_TRACKING_SYNC_REUSE_TTL_SECONDS` | `3600` | `fastvideo/performance/hf_store.py` | Freshness window for reusing an existing HF sync when requested by dashboard consumers. |
 | `FASTVIDEO_STAGE_LOGGING` | set by the pytest test | `test_inference_performance.py` | Enables pipeline stage timing capture for component metrics during benchmark runs. |
 
 ## CI integration
@@ -261,9 +394,11 @@ Buildkite artifact upload is in
 `.buildkite/scripts/pr_test.sh:upload_performance_artifacts`.
 
 Each performance build runs pytest first. If that fixed-threshold phase fails,
-`compare_baseline.py` is skipped, so Markdown summaries and normalized JSON
-artifacts are not emitted. The dashboard still runs best-effort for
-observability. When pytest passes, the rolling-baseline phase emits:
+PR/direct runs skip `compare_baseline.py` because they only upload passing
+records. Scheduled-main runs still execute `compare_baseline.py` with
+`PERF_PYTEST_RC` set so the failed canonical attempt is visible in normalized
+JSON and dashboard history. The dashboard runs best-effort for observability.
+When the rolling-baseline phase runs, it emits:
 
 * **Markdown summary** — appended to `$GITHUB_STEP_SUMMARY` when that variable
   is set, and written as `perf_<sha>_<ts>.md` for Buildkite upload. Contains a
@@ -279,11 +414,16 @@ observability. When pytest passes, the rolling-baseline phase emits:
 ## Adding a new benchmark
 
 1. Drop a new JSON config into
-   `.buildkite/performance-benchmarks/tests/<name>.json`. Required keys:
+   `.buildkite/performance-benchmarks/tests/<name>.json`. New configs should
+   use v2 identity fields:
 
    ```json
    {
      "benchmark_id": "<unique-id>",
+     "config_schema_version": 2,
+     "workload_id": "<stable-workload-id>",
+     "variant_id": "canonical",
+     "benchmark_version": 1,
      "model": { "model_path": "...", "model_short_name": "..." },
      "init_kwargs": { "num_gpus": 1, ... },
      "generation_kwargs": { "num_frames": 45, ... },
@@ -299,9 +439,19 @@ observability. When pytest passes, the rolling-baseline phase emits:
          "max_vae_decode_time_s": 10.0
        },
        "default": { "max_generation_time_s": 120.0, "max_peak_memory_mb": 30000.0 }
+     },
+     "regression_thresholds": {
+       "latency": { "threshold_percent": 0.10, "threshold_absolute": 1.0, "gated": true }
      }
-   }
-   ```
+  }
+
+  ```
+
+   Legacy v1 configs without `config_schema_version` still load, but should not
+   gain v2 identity or metadata fields until they are migrated to
+   `config_schema_version: 2`. For v2 configs, `workload_id`, `variant_id`,
+   and `benchmark_version` are part of the comparison key; benchmark runs
+   fail if any of these identity fields are missing.
 
 2. The pytest test auto-discovers all configs — no test code needed. CI
    picks it up on the next `/test performance` run.
@@ -320,6 +470,13 @@ observability. When pytest passes, the rolling-baseline phase emits:
    a useful fixed gate. The rolling baseline will still track component times
    when static component thresholds are omitted.
 
+6. Omit `regression_thresholds` to use the default rolling-baseline policy, or
+   include only benchmark-specific deviations. Tune these independently from
+   the fixed thresholds when a metric is noisy or should be informational. The
+   fixed `thresholds` block is an absolute pytest ceiling. The
+   `regression_thresholds` block controls rolling-baseline comparisons against
+   recent scheduled-main records.
+
 ## Troubleshooting
 
 **"No baseline for ... Initializing"** — first run for this `(model_id,

docs/inference/optimizations.md

@@ -74,6 +74,23 @@ uv pip install ninja
 python setup.py install
 ```
 
+### Flash Attention 4 (opt-in)
+
+FastVideo never auto-selects FlashAttention-4 (`flash_attn.cute`) just because it
+is installed: its CuTeDSL kernels JIT-compile per shape family and can fail at
+runtime on some GPU/shape combinations. To use FA4, install the pinned
+`flash-attn-4` build (see the `flash-attn-4` source in `pyproject.toml`) and set:
+
+```bash
+export FASTVIDEO_FA4=1
+```
+
+On GPUs below sm90 a capability gate routes to FlashAttention-2 the calls FA4
+cannot serve there: grad-enabled (training) attention (FA4's backward requires
+sm90+) and GQA attention (FA4's `pack_gqa` fails to JIT-compile below sm90).
+On sm90+ both run on FA4. If FA4 is unusable while `FASTVIDEO_FA4=1` is set,
+FastVideo fails loudly instead of silently falling back.
+
 ### FP4 Flash Attention 4 (Blackwell only)
 
 **`FLASH_ATTN`** with **`--nvfp4_fa4`**

fastvideo-kernel/CMakeLists.txt

@@ -12,6 +12,9 @@ set(_FASTVIDEO_USER_CUDA_ARCH "${CMAKE_CUDA_ARCHITECTURES}")
 if(NOT DEFINED GPU_BACKEND AND DEFINED ENV{GPU_BACKEND})
     set(GPU_BACKEND "$ENV{GPU_BACKEND}")
 endif()
+if(NOT GPU_BACKEND)
+    set(GPU_BACKEND "CUDA")
+endif()
 
 if(GPU_BACKEND STREQUAL "ROCM")
     enable_language(HIP)
@@ -50,7 +53,16 @@ if(NOT GPU_BACKEND STREQUAL "ROCM")
         if(_FASTVIDEO_USER_CUDA_ARCH)
             # Caller pinned -DCMAKE_CUDA_ARCHITECTURES (which torch ignores); translate it
             # to the TORCH_CUDA_ARCH_LIST spelling: "121" -> "12.1", "90a" -> "9.0a".
+            # Only numeric spellings translate; keywords like "native"/"all" would
+            # otherwise be mangled into nonsense ("nativ.e").
             foreach(_fv_arch IN LISTS _FASTVIDEO_USER_CUDA_ARCH)
+                if(NOT _fv_arch MATCHES "^[0-9]+[af]?$")
+                    message(FATAL_ERROR
+                        "fastvideo-kernel: CMAKE_CUDA_ARCHITECTURES='${_fv_arch}' is not "
+                        "supported. Use a numeric arch (e.g. 90a, 121), set "
+                        "TORCH_CUDA_ARCH_LIST directly (e.g. 9.0a), or unset both to "
+                        "auto-detect from the visible GPU.")
+                endif()
                 string(REGEX MATCH "[af]$" _fv_suffix "${_fv_arch}")
                 string(REGEX REPLACE "[af]$" "" _fv_num "${_fv_arch}")
                 string(REGEX REPLACE "(.)$" ".\\1" _fv_num "${_fv_num}")    # dot before the last digit
@@ -173,6 +185,14 @@ else()
     endif()
 endif()
 
+# ThunderKittens headers don't compile on aarch64 hosts: plain char is unsigned
+# there, and tk's base_types.cuh brace-initializes signed-char vector members
+# from char (narrowing error). Skip TK until upstream is aarch64-clean.
+if(ENABLE_TK_KERNELS AND CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch64|arm64)$")
+    message(STATUS "ThunderKittens kernels: forced OFF on ${CMAKE_SYSTEM_PROCESSOR} (tk headers are not aarch64-clean)")
+    set(ENABLE_TK_KERNELS OFF)
+endif()
+
 if(ENABLE_TK_KERNELS)
     message(STATUS "ThunderKittens kernels: ENABLED")
 else()
@@ -263,11 +283,6 @@ set(CUDA_FLAGS
     "--expt-relaxed-constexpr"
     "-Xcompiler=-fno-strict-aliasing"
     "-Xcompiler=-fPIC"
-    # ARM/aarch64 defaults `char` to unsigned, but ThunderKittens headers assume the
-    # x86 signed-char behavior (else base_types.cuh hits "narrowing conversion from
-    # char to signed char"). Force signed char so TK compiles on Grace Hopper; this
-    # is a no-op on x86_64, where char is already signed.
-    "-Xcompiler=-fsigned-char"
     "-DTORCH_COMPILE"
     "-Xnvlink=--verbose"
     "-Xptxas=--verbose"
@@ -426,3 +441,14 @@ if(ENABLE_ATTN_QAT_INFER)
     install(TARGETS fp4attn_cuda LIBRARY DESTINATION .)
     install(TARGETS fp4quant_cuda LIBRARY DESTINATION .)
 endif()
+
+# One-look answer to "what is this build producing?" — kept last so it is the
+# final thing configure prints. The per-kernel matrix lives in README.md.
+message(STATUS "============== fastvideo-kernel build summary ==============")
+message(STATUS "host / backend:       ${CMAKE_SYSTEM_PROCESSOR} / ${GPU_BACKEND}")
+message(STATUS "TORCH_CUDA_ARCH_LIST: ${TORCH_CUDA_ARCH_LIST}")
+message(STATUS "fastvideo_kernel_ops: ON  (turbodiffusion int8-gemm/quant/rmsnorm/layernorm, all listed archs)")
+message(STATUS "  + TK sta/block_sparse (sm_90a only):          ${ENABLE_TK_KERNELS}")
+message(STATUS "fp4attn/fp4quant (sm_120a only, CUDA >= 12.8):  ${ENABLE_ATTN_QAT_INFER}")
+message(STATUS "Triton fallbacks ship in python/fastvideo_kernel/triton_kernels regardless.")
+message(STATUS "============================================================")

fastvideo-kernel/README.md

@@ -2,6 +2,42 @@
 
 CUDA kernels for FastVideo video generation.
 
+## Kernel inventory
+
+Compiled CUDA extensions (CMake, see the build summary printed at the end of every configure):
+
+| Extension | Kernels | Sources | GPU arch | Build gate |
+|---|---|---|---|---|
+| `fastvideo_kernel._C.fastvideo_kernel_ops` | TurboDiffusion INT8 GEMM, quant, RMSNorm, LayerNorm | `csrc/turbodiffusion/` | every arch in `TORCH_CUDA_ARCH_LIST` | always built |
+| same extension, optional part | ThunderKittens sliding-tile attention (`sta_fwd`) and VSA block-sparse (`block_sparse_fwd/bwd`) | `csrc/attention/*_h100.cu` | Hopper `sm_90a` only | `FASTVIDEO_KERNEL_BUILD_TK` (AUTO = ON iff `9.0a` is in the arch list; always OFF on aarch64 hosts — TK headers don't compile there) |
+| `fp4attn_cuda`, `fp4quant_cuda` | FP4 attention + quantization ("attn_qat_infer", modified SageAttention3) | `attn_qat_infer/` | consumer Blackwell `sm_120a` only, CUDA ≥ 12.8 | `FASTVIDEO_KERNEL_BUILD_ATTN_QAT_INFER` (AUTO = ON iff `12.0a` is in the arch list) |
+
+Runtime-JIT kernels (no build step, ship in every wheel/image):
+
+| Kernels | Where | Used when |
+|---|---|---|
+| Triton: STA, VSA block-sparse, SLA, fused compress+topk, FP4 QAT training, quant/norm utils | `python/fastvideo_kernel/triton_kernels/` | automatic fallback when the matching C++ op is absent (`ops.py`, `turbodiffusion_ops.py`) |
+| FA4 CuTe-DSL block-sparse (VSA-256 fastpath on `sm_100`) | `block_sparse_attn_cute_fwd.py` | optional `flash_attn.cute` dependency, see below |
+| VMoBA `moba_attn_varlen` | `vmoba.py` | wraps flash-attn varlen |
+
+## What gets built where, and when
+
+| Surface | Trigger | Leg | `TORCH_CUDA_ARCH_LIST` | TK | FP4 |
+|---|---|---|---|---|---|
+| PyPI wheels (`.github/workflows/publish-kernel.yml`) | version bump in `fastvideo-kernel/pyproject.toml` on main, or manual dispatch | x86_64 cu126 | `9.0a` | ON | — (CUDA < 12.8) |
+| | | x86_64 cu130 | `9.0a;12.0a` | ON | ON |
+| | | aarch64 cu130 | `10.0a;12.0a` | — | ON |
+| Docker images `ghcr.io/hao-ai-lab/fastvideo/fastvideo-dev` (`.github/workflows/infra-build-image.yml`) | `docker/Dockerfile` changes on main, or manual dispatch | amd64 cuda12.6.3 + cuda13.0.0 | `9.0a` | ON | — |
+| | | arm64 cuda12.6.3 (GH200) | `9.0a` | — (aarch64) | — |
+| | | arm64 cuda13.0.0 (GB10 / DGX Spark) | `12.1` | — | — |
+| Local `./build.sh` | manual | probes the visible GPU via torch | detected | ON iff sm_90 (non-aarch64 host) | ON iff sm_120 |
+
+Notes:
+
+- No Docker image ships the FP4 kernels; only the x86_64/aarch64 cu130 wheels do.
+- On arm64 images (GH200 included) STA/VSA run on the Triton fallbacks, since TK never builds on aarch64.
+- Kernel tests run on Buildkite GPU CI for PRs touching `fastvideo-kernel/**` (see `.buildkite/pipeline.yml`).
+
 ## Installation
 
 ### Standard Installation (Local Development)

fastvideo-kernel/build.sh

@@ -46,6 +46,23 @@ fi
 if git rev-parse --git-dir >/dev/null 2>&1; then
     git submodule update --init --recursive include/cutlass include/tk
 fi
+# Fail fast with a clear message if the headers are still missing (e.g. a
+# Docker context that excluded .git AND the submodule contents) instead of
+# dying later in a wall of nvcc include errors. CUTLASS is consumed by the
+# always-built turbodiffusion sources, so it is a hard error; ThunderKittens
+# only feeds the TK-gated Hopper kernels, so a missing tree just warns (the
+# TK gate resolves later, and non-SM90/ROCm builds never touch it).
+if [ ! -d include/cutlass/include ]; then
+    echo "ERROR: include/cutlass/include is missing. Outside a git checkout the" >&2
+    echo "       CUTLASS sources must already be present (run" >&2
+    echo "       'git submodule update --init --recursive include/cutlass include/tk'" >&2
+    echo "       in the source checkout, or include them in the build context)." >&2
+    exit 1
+fi
+if [ ! -d include/tk/include ]; then
+    echo "WARNING: include/tk/include is missing; ThunderKittens (Hopper sm_90a)" >&2
+    echo "         kernels cannot be built. Fine for non-SM90/ROCm targets." >&2
+fi
 
 # Install build dependencies
 uv pip install scikit-build-core cmake ninja

fastvideo/attention/backends/flash_attn.py

@@ -1,15 +1,39 @@
 # SPDX-License-Identifier: Apache-2.0
 
+import importlib.util
 import os
+
 import torch
 import torch.nn.functional as F
 from dataclasses import dataclass
 
-try:
-    from fastvideo.attention.utils.flash_attn_cute import flash_attn_func
+from fastvideo import envs
+from fastvideo.attention.backends.abstract import (
+    AttentionBackend,
+    AttentionImpl,
+    AttentionMetadata,
+    AttentionMetadataBuilder,
+)
+from fastvideo.logger import init_logger
 
+logger = init_logger(__name__)
+
+# FA4 (flash_attn.cute) is explicit opt-in via FASTVIDEO_FA4=1, mirroring the
+# kernel package's FASTVIDEO_VSA_CUTEDSL: its CuTeDSL kernels JIT-compile per
+# shape family and can fail at runtime on some arch/shape combinations, so it
+# is never auto-selected just because it is installed. Below sm90 a capability
+# gate in flash_attn_cute routes to FA2 the calls FA4 cannot serve there:
+# grad-enabled (its backward asserts sm90+) and GQA (pack_gqa fails CuTeDSL
+# JIT, observed on sm_89).
+if envs.FASTVIDEO_FA4:
+    try:
+        from fastvideo.attention.utils.flash_attn_cute import flash_attn_func
+    except ImportError as e:
+        raise RuntimeError(f"FASTVIDEO_FA4=1 but flash_attn.cute (FA4) is not usable ({e}); "
+                           "fix the FA4 install (see the flash-attn-4 pin in pyproject.toml) "
+                           "or unset FASTVIDEO_FA4.") from e
     fa_version = "4"
-except ImportError:
+else:
     try:
         from flash_attn_interface import flash_attn_func as flash_attn_3_func
 
@@ -21,6 +45,12 @@ except ImportError:
         from flash_attn import flash_attn_func as flash_attn_2_func
         flash_attn_func = flash_attn_2_func
         fa_version = "2"
+    try:
+        if importlib.util.find_spec("flash_attn.cute") is not None:
+            logger.info("flash_attn.cute (FA4) is installed but not enabled; "
+                        "set FASTVIDEO_FA4=1 to use it for inference.")
+    except ImportError:
+        pass
 
 # torch.compile traceability: the FA4/cute path (fa_version=="4") is
 # already a registered torch.library custom op, so dynamo treats it as a
@@ -87,9 +117,10 @@ if fa_version in ("2", "3"):
             return _fa_default(q, k, v, softmax_scale=softmax_scale, causal=causal)
         return torch.ops.fastvideo._flash_attn_default_forward(q, k, v, softmax_scale, causal)
 elif fa_version == "4":
-    # FA4 path: `flash_attn_func` is already a torch.library custom op
-    # (registered in `fastvideo.attention.utils.flash_attn_cute`), so a
-    # passthrough is enough — no extra registration needed.
+    # FA4 path: `flash_attn_func` (from `flash_attn_cute`) goes through a
+    # registered torch.library custom op (with an FA4 backward on sm90+;
+    # grad-enabled and GQA calls below sm90 route to FA2), so a passthrough
+    # is enough — no extra registration needed.
     def flash_attn_func_compilable(q, k, v, softmax_scale=None, causal=False):
         return flash_attn_func(q, k, v, softmax_scale=softmax_scale, causal=causal)
 else:
@@ -99,17 +130,6 @@ else:
     raise RuntimeError(f"Unsupported FlashAttention version: {fa_version!r} — expected "
                        f"'2', '3', or '4' from the import probe above.")
 
-from fastvideo.attention.backends.abstract import (
-    AttentionBackend,
-    AttentionImpl,
-    AttentionMetadata,
-    AttentionMetadataBuilder,
-)
-from fastvideo.logger import init_logger
-
-logger = init_logger(__name__)
-
-_WARNED_NON_FA_DTYPE = False
 logger.info("Using FlashAttention-%s backend", fa_version)
 
 # FP4 FA4 support: quantize Q/K to NVFP4 E2M1 for block-scaled MMA on Blackwell.
@@ -271,12 +291,8 @@ class FlashAttentionImpl(AttentionImpl):
         # SP). Cast through bf16 and restore, matching TORCH_SDPA's tolerance.
         orig_dtype = query.dtype
         if orig_dtype not in (torch.float16, torch.bfloat16):
-            global _WARNED_NON_FA_DTYPE
-            if not _WARNED_NON_FA_DTYPE:
-                _WARNED_NON_FA_DTYPE = True
-                logger.warning(
-                    "FLASH_ATTN received %s inputs; casting to bfloat16 for the "
-                    "kernel and restoring on output.", orig_dtype)
+            logger.warning_once(f"FLASH_ATTN received {orig_dtype} inputs; casting to "
+                                f"bfloat16 for the kernel and restoring on output.")
             query = query.to(torch.bfloat16)
             key = key.to(torch.bfloat16)
             value = value.to(torch.bfloat16)

fastvideo/attention/utils/flash_attn_cute.py

@@ -4,10 +4,9 @@ import functools
 from collections.abc import Callable
 
 import torch
-from flash_attn import flash_attn_func as _flash_attn_2_func
-from flash_attn import flash_attn_varlen_func as _flash_attn_2_varlen_func
 
 from fastvideo.logger import init_logger
+from fastvideo.platforms import current_platform
 
 logger = init_logger(__name__)
 
@@ -16,7 +15,8 @@ if torch.cuda.is_available():
         from flash_attn.cute.interface import _flash_attn_bwd, _flash_attn_fwd
     except ImportError:
         # flash_attn.cute (FA4) is simply not installed -- expected on builds
-        # without it; callers fall back to FA3/FA2 quietly.
+        # without it; callers handle the ImportError (the FASTVIDEO_FA4 gate in
+        # flash_attn.py raises, the FP4 probe treats FA4 as unavailable).
         raise
     except Exception as e:
         # flash_attn.cute IS installed but failed to import -- almost always an
@@ -24,23 +24,57 @@ if torch.cuda.is_available():
         # 'cutlass.cute.core' has no attribute 'ThrMma'" (an AttributeError, not
         # ImportError). This is fixable by pinning a compatible
         # nvidia-cutlass-dsl, so warn loudly, then re-raise as ImportError so
-        # callers fall back to FA3/FA2 instead of crashing worker init.
+        # callers can handle it uniformly.
         logger.warning(
-            "flash_attn.cute (FA4) is installed but failed to import (%r); "
-            "falling back to FA3/FA2. This is usually an nvidia-cutlass-dsl "
-            "version mismatch -- pin a compatible nvidia-cutlass-dsl to "
-            "restore FA4.", e)
+            "flash_attn.cute (FA4) is installed but failed to import (%r). "
+            "This is usually an nvidia-cutlass-dsl version mismatch -- pin a "
+            "compatible nvidia-cutlass-dsl to restore FA4.", e)
         raise ImportError(f"flash_attn.cute (FA4) import failed: {e!r}") from e
 else:
     # This error will be caught in flash_attn.py or flash_attn_no_pad.py
     raise ImportError("flash_attn.cute is only available on CUDA devices; this error must be handled internally")
 
+try:
+    # FA2 serves the calls FA4 cute cannot on pre-sm90 GPUs (backward, GQA).
+    # Optional so FA4-only installs can still import this module.
+    from flash_attn import flash_attn_func as _flash_attn_2_func
+    from flash_attn import flash_attn_varlen_func as _flash_attn_2_varlen_func
+except ImportError:
+    _flash_attn_2_func = None
+    _flash_attn_2_varlen_func = None
+
 
 def _check_dropout(dropout_p: float) -> None:
     if dropout_p != 0.0:
         raise NotImplementedError(f"flash_attn.cute does not support dropout (got dropout_p={dropout_p})")
 
 
+@functools.cache
+def _sm90_or_newer() -> bool:
+    return current_platform.has_device_capability(90)
+
+
+def _use_fa2(q: torch.Tensor, k: torch.Tensor, v: torch.Tensor) -> bool:
+    if _sm90_or_newer():
+        return False
+    # Pre-sm90 FA4 cute limitations, both served by FA2 (deterministic
+    # capability gate, not a runtime fallback):
+    #   * the backward asserts sm90+ (L40S/sm_89 dies on its arch check);
+    #   * GQA fails CuTeDSL JIT in pack_gqa ("ValueError: Operation creation
+    #     failed", observed on sm_89 with HunyuanGameCraft/LTX2).
+    if q.shape[-2] != k.shape[-2]:
+        return True
+    return torch.is_grad_enabled() and any(t.requires_grad for t in (q, k, v))
+
+
+def _fa2_or_raise(fa2_func: Callable | None) -> Callable:
+    if fa2_func is None:
+        raise RuntimeError("this attention call cannot run on FA4 cute below sm90 (its backward and "
+                           "GQA support require sm90+) and flash-attn 2, which serves it there, is "
+                           "not installed.")
+    return fa2_func
+
+
 @torch.library.custom_op(
     "fastvideo::_flash_attn_cute_forward",
     mutates_args=(),
@@ -243,70 +277,6 @@ torch.library.register_autograd(
 )
 
 
-# FA4's CuTeDSL kernels JIT-compile per shape family, and some configurations
-# fail MLIR op creation at runtime even though the import succeeded (observed:
-# GQA models on sm_89 dying in pack_gqa with "ValueError: Operation creation
-# failed"). Degrade to FA2 once, process-wide, instead of crashing inference.
-class _FA4Policy:
-    """Per-call gate for the FA4 cute fast path, with FA2 as the fallback.
-
-    FA4 is skipped when:
-      * a previous call failed at runtime -- CuTeDSL JIT compilation is
-        shape-dependent, so the first failure disables FA4 for the rest of
-        the process instead of retrying a broken JIT on every call; or
-      * the call needs autograd -- FA4's backward asserts sm90+ (L40S/sm_89
-        dies on its arch check) and is unvalidated for training in this repo
-        (its lse is not even allocated through our inference-shaped custom
-        op), so training keeps the pre-FA4 behavior: FA2 on every device.
-    """
-
-    def __init__(self) -> None:
-        self.broken = False
-
-    def use_fa4(self, *tensors: torch.Tensor) -> bool:
-        if self.broken:
-            return False
-        return not (torch.is_grad_enabled() and any(t.requires_grad for t in tensors))
-
-    def mark_broken(self, error: Exception) -> None:
-        if not self.broken:
-            self.broken = True
-            logger.warning(
-                "flash_attn.cute (FA4) failed at runtime (%r); falling back "
-                "to FA2 for the rest of this process.", error)
-
-
-_FA4 = _FA4Policy()
-
-
-def _with_fa2_fallback(fa2_func: Callable) -> Callable:
-    """Pair an FA4 cute wrapper with its FA2 twin of the same signature.
-
-    The decorated body runs only when ``_FA4`` allows it; otherwise (or after
-    the first FA4 runtime failure) the call is served by ``fa2_func``.
-    ``NotImplementedError`` is a contract error (e.g. dropout), not a JIT
-    failure, so it propagates without disabling FA4.
-    """
-
-    def decorator(fa4_func: Callable) -> Callable:
-
-        @functools.wraps(fa4_func)
-        def wrapper(q: torch.Tensor, k: torch.Tensor, v: torch.Tensor, *args, **kwargs) -> torch.Tensor:
-            if _FA4.use_fa4(q, k, v):
-                try:
-                    return fa4_func(q, k, v, *args, **kwargs)
-                except NotImplementedError:
-                    raise
-                except Exception as e:  # CuTeDSL compile errors surface as ValueError
-                    _FA4.mark_broken(e)
-            return fa2_func(q, k, v, *args, **kwargs)
-
-        return wrapper
-
-    return decorator
-
-
-@_with_fa2_fallback(_flash_attn_2_func)
 def flash_attn_func(
     q: torch.Tensor,
     k: torch.Tensor,
@@ -317,6 +287,16 @@ def flash_attn_func(
     deterministic: bool = False,
 ) -> torch.Tensor:
     """Only returns the output, not the lse."""
+    if _use_fa2(q, k, v):
+        return _fa2_or_raise(_flash_attn_2_func)(
+            q,
+            k,
+            v,
+            dropout_p=dropout_p,
+            softmax_scale=softmax_scale,
+            causal=causal,
+            deterministic=deterministic,
+        )
     _check_dropout(dropout_p)
     out, _ = torch.ops.fastvideo._flash_attn_cute_forward(q, k, v, softmax_scale, causal, deterministic)
     return out
@@ -392,7 +372,6 @@ def flash_attn_fp4_func(
     return torch.ops.fastvideo._flash_attn_cute_fp4_forward(q, k, v, sfq, sfk, softmax_scale, causal)
 
 
-@_with_fa2_fallback(_flash_attn_2_varlen_func)
 def flash_attn_varlen_func(
     q: torch.Tensor,
     k: torch.Tensor,
@@ -407,6 +386,20 @@ def flash_attn_varlen_func(
     deterministic: bool = False,
 ) -> torch.Tensor:
     """Only returns the output, not the lse."""
+    if _use_fa2(q, k, v):
+        return _fa2_or_raise(_flash_attn_2_varlen_func)(
+            q,
+            k,
+            v,
+            cu_seqlens_q,
+            cu_seqlens_k,
+            max_seqlen_q,
+            max_seqlen_k,
+            dropout_p=dropout_p,
+            softmax_scale=softmax_scale,
+            causal=causal,
+            deterministic=deterministic,
+        )
     _check_dropout(dropout_p)
     out, _ = torch.ops.fastvideo._flash_attn_cute_varlen_forward(
         q,

fastvideo/attention/utils/flash_attn_no_pad.py

@@ -21,24 +21,35 @@ from einops import rearrange
 from flash_attn import flash_attn_varlen_qkvpacked_func
 from flash_attn.bert_padding import pad_input, unpad_input
 
+from fastvideo import envs
+
 
 def _resolve_flash_attn_varlen_func() -> Any:
-    try:
-        from fastvideo.attention.utils.flash_attn_cute import (
-            flash_attn_varlen_func as flash_attn_varlen_func_cute, )
+    if envs.FASTVIDEO_FA4:
+        # FA4 cute is explicit opt-in (see fastvideo/attention/backends/
+        # flash_attn.py); with FASTVIDEO_FA4=1 an unimportable FA4 build must
+        # fail loudly here rather than fall through to FA3/FA2. RuntimeError,
+        # not ImportError: importers like bsa_attn.py treat ImportError as
+        # "flash-attn not installed" and silently degrade to reference kernels.
+        try:
+            from fastvideo.attention.utils.flash_attn_cute import (
+                flash_attn_varlen_func as flash_attn_varlen_func_cute, )
+        except ImportError as e:
+            raise RuntimeError(f"FASTVIDEO_FA4=1 but flash_attn.cute (FA4) is not usable ({e}); "
+                               "fix the FA4 install (see the flash-attn-4 pin in pyproject.toml) "
+                               "or unset FASTVIDEO_FA4.") from e
 
         return flash_attn_varlen_func_cute
+    try:
+        from flash_attn_interface import (
+            flash_attn_varlen_func as flash_attn_varlen_func_interface, )
+
+        return flash_attn_varlen_func_interface
     except ImportError:
-        try:
-            from flash_attn_interface import (
-                flash_attn_varlen_func as flash_attn_varlen_func_interface, )
+        from flash_attn import (
+            flash_attn_varlen_func as flash_attn_varlen_func_flash, )
 
-            return flash_attn_varlen_func_interface
-        except ImportError:
-            from flash_attn import (
-                flash_attn_varlen_func as flash_attn_varlen_func_flash, )
-
-            return flash_attn_varlen_func_flash
+        return flash_attn_varlen_func_flash
 
 
 flash_attn_varlen_func_impl = _resolve_flash_attn_varlen_func()

fastvideo/envs.py

@@ -20,6 +20,7 @@ if TYPE_CHECKING:
     FASTVIDEO_LOGGING_CONFIG_PATH: str | None = None
     FASTVIDEO_TRACE_FUNCTION: int = 0
     FASTVIDEO_ATTENTION_BACKEND: str | None = None
+    FASTVIDEO_FA4: bool = False
     FASTVIDEO_WORKER_MULTIPROC_METHOD: str = "spawn"
     FASTVIDEO_TARGET_DEVICE: str = "cuda"
     MAX_JOBS: str | None = None
@@ -207,9 +208,18 @@ environment_variables: dict[str, Callable[[], Any]] = {
     # - "VIDEO_SPARSE_ATTN": use Video Sparse Attention
     # - "SAGE_ATTN": use Sage Attention
     # - "SAGE_ATTN_THREE": use Sage Attention 3
+    # FLASH_ATTN uses FlashAttention-3/2; to run FlashAttention-4 set
+    # FASTVIDEO_FA4=1 as well (see below).
     "FASTVIDEO_ATTENTION_BACKEND":
     lambda: os.getenv("FASTVIDEO_ATTENTION_BACKEND", None),
 
+    # If set (=1), the FLASH_ATTN backend uses FlashAttention-4
+    # (flash_attn.cute). FA4 is opt-in and never auto-selected just because it
+    # is installed. Below sm90, grad-enabled and GQA calls are routed to FA2
+    # (FA4's backward asserts sm90+ and its pack_gqa fails to JIT there).
+    "FASTVIDEO_FA4":
+    lambda: os.getenv("FASTVIDEO_FA4", "0") != "0",
+
     # Use dedicated multiprocess context for workers.
     "FASTVIDEO_WORKER_MULTIPROC_METHOD":
     lambda: os.getenv("FASTVIDEO_WORKER_MULTIPROC_METHOD", "spawn"),

fastvideo/models/dits/causal_wanvideo.py

@@ -323,9 +323,9 @@ class CausalWanTransformerBlock(nn.Module):
         value, _ = self.to_v(norm_hidden_states)
 
         if self.norm_q is not None:
-            query = self.norm_q.forward_native(query)
+            query = self.norm_q(query)
         if self.norm_k is not None:
-            key = self.norm_k.forward_native(key)
+            key = self.norm_k(key)
 
         query = query.squeeze(1).unflatten(2, (self.num_attention_heads, -1))
         key = key.squeeze(1).unflatten(2, (self.num_attention_heads, -1))

fastvideo/models/dits/matrixgame2/causal_model.py

@@ -537,9 +537,9 @@ class CausalMatrixGame2TransformerBlock(nn.Module):
         value, _ = self.to_v(norm_hidden_states)
 
         if self.norm_q is not None:
-            query = self.norm_q.forward_native(query)
+            query = self.norm_q(query)
         if self.norm_k is not None:
-            key = self.norm_k.forward_native(key)
+            key = self.norm_k(key)
 
         query = query.squeeze(1).unflatten(2, (self.num_attention_heads, -1))
         key = key.squeeze(1).unflatten(2, (self.num_attention_heads, -1))

fastvideo/performance/__init__.py

@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Performance benchmark and dashboard utilities."""

fastvideo/performance/hf_store.py

@@ -57,9 +57,7 @@ def is_baseline_eligible_record(record: dict[str, Any]) -> bool:
     """
     if record.get("baseline_eligible") is True:
         return True
-    if "baseline_eligible" not in record and "run_source" not in record:
-        return True
-    return False
+    return "baseline_eligible" not in record and "run_source" not in record
 
 
 def resolve_hf_token() -> str | None:
@@ -270,16 +268,31 @@ def load_records_for_model(
     model_id: str,
     gpu_type: str | None = None,
     *,
+    workload_id: str | None = None,
+    variant_id: str | None = None,
+    benchmark_version: str | None = None,
+    recipe_fingerprint: str | None = None,
+    hardware_profile_id: str | None = None,
+    software_profile_id: str | None = None,
     last_n: int | None = None,
     successful_only: bool = True,
     baseline_eligible_only: bool = False,
 ) -> list[dict[str, Any]]:
-    """Return records for a specific *model_id*, optionally filtered by GPU.
+    """Return records for a specific *model_id*, optionally filtered by cohort.
 
     Args:
         local_dir: Root directory previously populated by :func:`sync_from_hf`.
         model_id: Matches the ``model_id`` field inside each JSON record.
         gpu_type: When set, only records whose ``gpu_type`` matches are returned.
+        workload_id: When set, only records from the same workload are returned.
+        variant_id: When set, only records from the same workload variant are returned.
+        benchmark_version: When set, only records from the same benchmark version are returned.
+        recipe_fingerprint: When set, only records from the same benchmark
+            recipe are returned.
+        hardware_profile_id: When set, only records from the same hardware
+            cohort are returned.
+        software_profile_id: When set, only records from the same software
+            cohort are returned.
         last_n: When set, return only the most recent *n* records (after all
             other filters). Useful for sliding-window baseline calculations.
         successful_only: Passed through to :func:`load_records`.
@@ -301,6 +314,18 @@ def load_records_for_model(
     if gpu_type is not None:
         records = [r for r in records if r.get("gpu_type") == gpu_type]
 
+    identity_filters = {
+        "workload_id": workload_id,
+        "variant_id": variant_id,
+        "benchmark_version": benchmark_version,
+        "recipe_fingerprint": recipe_fingerprint,
+        "hardware_profile_id": hardware_profile_id,
+        "software_profile_id": software_profile_id,
+    }
+    for key, expected in identity_filters.items():
+        if expected is not None:
+            records = [r for r in records if str(r.get(key)) == str(expected)]
+
     if last_n is not None:
         records = records[-last_n:]
 

fastvideo/performance/metric_policy.py

@@ -0,0 +1,130 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Metric policy for rolling performance baseline comparisons."""
+
+from __future__ import annotations
+
+from collections.abc import Mapping
+from dataclasses import dataclass
+from typing import Any
+
+
+@dataclass(frozen=True)
+class MetricPolicy:
+    key: str
+    label: str
+    precision: int
+    lower_is_better: bool
+    threshold_percent: float
+    threshold_absolute: float
+    gated: bool = True
+
+
+@dataclass(frozen=True)
+class MetricDelta:
+    absolute: float
+    percent: float
+    threshold_exceeded: bool
+    regressed: bool
+
+
+DEFAULT_METRIC_POLICIES: tuple[MetricPolicy, ...] = (
+    MetricPolicy("latency", "Latency", 3, True, 0.08, 0.5),
+    MetricPolicy("throughput", "Throughput", 3, False, 0.08, 0.05),
+    MetricPolicy("memory", "Memory", 1, True, 0.05, 256.0),
+    MetricPolicy("text_encoder_time_s", "Text Enc", 3, True, 0.05, 0.25),
+    MetricPolicy("dit_time_s", "DiT", 3, True, 0.05, 0.25),
+    MetricPolicy("vae_decode_time_s", "VAE Decode", 3, True, 0.05, 0.25),
+)
+
+def _optional_float(value: Any) -> float | None:
+    if value is None or isinstance(value, bool):
+        return None
+    try:
+        return float(value)
+    except (TypeError, ValueError):
+        return None
+
+
+def _optional_bool(value: Any) -> bool | None:
+    if isinstance(value, bool):
+        return value
+    if isinstance(value, str):
+        normalized = value.strip().lower()
+        if normalized in {"1", "true", "yes", "on"}:
+            return True
+        if normalized in {"0", "false", "no", "off"}:
+            return False
+    return None
+
+
+def resolve_metric_policies(
+    threshold_overrides: Mapping[str, Any] | None,
+) -> tuple[MetricPolicy, ...]:
+    """Return default metric policies with optional per-metric overrides."""
+
+    if not isinstance(threshold_overrides, Mapping):
+        threshold_overrides = {}
+    policies: list[MetricPolicy] = []
+    for base_policy in DEFAULT_METRIC_POLICIES:
+        raw_override = threshold_overrides.get(base_policy.key, {})
+        if not isinstance(raw_override, Mapping):
+            raw_override = {}
+
+        threshold_percent = _optional_float(raw_override.get("threshold_percent"))
+        threshold_absolute = _optional_float(raw_override.get("threshold_absolute"))
+        gated = _optional_bool(raw_override.get("gated"))
+
+        policies.append(
+            MetricPolicy(
+                key=base_policy.key,
+                label=base_policy.label,
+                precision=base_policy.precision,
+                lower_is_better=base_policy.lower_is_better,
+                threshold_percent=(
+                    base_policy.threshold_percent
+                    if threshold_percent is None
+                    else threshold_percent
+                ),
+                threshold_absolute=(
+                    base_policy.threshold_absolute
+                    if threshold_absolute is None
+                    else threshold_absolute
+                ),
+                gated=base_policy.gated if gated is None else gated,
+            )
+        )
+    return tuple(policies)
+
+
+def serialize_metric_thresholds(
+    policies: tuple[MetricPolicy, ...],
+) -> dict[str, dict[str, float | bool]]:
+    return {
+        policy.key: {
+            "threshold_percent": policy.threshold_percent,
+            "threshold_absolute": policy.threshold_absolute,
+            "gated": policy.gated,
+        }
+        for policy in policies
+    }
+
+
+def regression_delta(
+    policy: MetricPolicy,
+    current: float,
+    baseline: float,
+) -> MetricDelta | None:
+    if baseline <= 0:
+        return None
+    absolute_delta = current - baseline if policy.lower_is_better else baseline - current
+    percent_delta = absolute_delta / baseline
+    threshold_exceeded = (
+        percent_delta > policy.threshold_percent
+        and absolute_delta > policy.threshold_absolute
+    )
+    return MetricDelta(
+        absolute=absolute_delta,
+        percent=percent_delta,
+        threshold_exceeded=threshold_exceeded,
+        regressed=policy.gated and threshold_exceeded,
+    )

fastvideo/performance_dashboard/api.py

@@ -13,7 +13,7 @@ from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import FileResponse
 from fastapi.staticfiles import StaticFiles
 
-from fastvideo.tests.performance import hf_store
+from fastvideo.performance import hf_store
 
 from .service import build_latest_summary, build_trends, filter_records
 
@@ -150,7 +150,6 @@ def create_app(store: PerformanceDataStore | None = None) -> FastAPI:
         filtered = filter_records(loaded, model_id=model_id, gpu_type=gpu_type)
         rows = build_latest_summary(
             filtered,
-            max_regression=float(os.environ.get("PERF_MAX_REGRESSION", "0.05")),
             run_source=run_source,
         )
         return {

fastvideo/performance_dashboard/metrics.py

@@ -1,26 +1,8 @@
 # SPDX-License-Identifier: Apache-2.0
 """Metric definitions shared by the performance dashboard backend."""
 
-from __future__ import annotations
+from fastvideo.performance.metric_policy import DEFAULT_METRIC_POLICIES
 
-from dataclasses import dataclass
-
-
-@dataclass(frozen=True)
-class MetricDefinition:
-    key: str
-    label: str
-    precision: int
-    lower_is_better: bool
-
-
-METRICS: tuple[MetricDefinition, ...] = (
-    MetricDefinition("latency", "Latency", 3, True),
-    MetricDefinition("throughput", "Throughput", 3, False),
-    MetricDefinition("memory", "Memory", 1, True),
-    MetricDefinition("text_encoder_time_s", "Text Encoder", 3, True),
-    MetricDefinition("dit_time_s", "DiT", 3, True),
-    MetricDefinition("vae_decode_time_s", "VAE Decode", 3, True),
-)
+METRICS = DEFAULT_METRIC_POLICIES
 
 METRIC_BY_KEY = {metric.key: metric for metric in METRICS}

fastvideo/performance_dashboard/service.py

@@ -13,11 +13,19 @@ from collections import defaultdict
 from datetime import datetime, timezone
 from typing import Any
 
-from fastvideo.tests.performance.hf_store import is_baseline_eligible_record, safe_float
-
-from .metrics import METRICS
+from fastvideo.performance.hf_store import is_baseline_eligible_record, safe_float
+from fastvideo.performance.metric_policy import regression_delta, resolve_metric_policies
 
 Record = dict[str, Any]
+CohortKey = tuple[str, str, str, str, str, str, str, str]
+COMPARISON_COHORT_KEYS = (
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+    "recipe_fingerprint",
+    "hardware_profile_id",
+    "software_profile_id",
+)
 
 
 def parse_timestamp(value: Any) -> datetime | None:
@@ -78,15 +86,44 @@ def record_metadata(record: Record) -> Record:
     }
 
 
-def group_by_model_gpu(records: list[Record]) -> dict[tuple[str, str], list[Record]]:
-    groups: dict[tuple[str, str], list[Record]] = defaultdict(list)
+def record_comparison_metadata(record: Record) -> Record:
+    return {key: record.get(key) or "" for key in COMPARISON_COHORT_KEYS}
+
+
+def comparison_cohort_key(record: Record) -> CohortKey:
+    return (
+        str(record.get("model_id") or "unknown"),
+        str(record.get("gpu_type") or "unknown"),
+        str(record.get("workload_id") or ""),
+        str(record.get("variant_id") or ""),
+        str(record.get("benchmark_version") or ""),
+        str(record.get("recipe_fingerprint") or ""),
+        str(record.get("hardware_profile_id") or ""),
+        str(record.get("software_profile_id") or ""),
+    )
+
+
+def group_by_comparison_cohort(records: list[Record]) -> dict[CohortKey, list[Record]]:
+    groups: dict[CohortKey, list[Record]] = defaultdict(list)
     for record in records:
-        model_id = str(record.get("model_id") or "unknown")
-        gpu_type = str(record.get("gpu_type") or "unknown")
-        groups[(model_id, gpu_type)].append(record)
+        groups[comparison_cohort_key(record)].append(record)
     return {key: sorted(value, key=record_sort_key) for key, value in groups.items()}
 
 
+def comparison_sort_key(record: Record) -> CohortKey:
+    return comparison_cohort_key(record)
+
+
+def latest_row_sort_key(row: Record) -> tuple[Any, ...]:
+    return (row["status"] != "fail", *comparison_sort_key(row))
+
+
+def group_identity(key: CohortKey) -> tuple[str, str]:
+    model_id = key[0]
+    gpu_type = key[1]
+    return model_id, gpu_type
+
+
 def baseline_value(records: list[Record], metric_key: str) -> float | None:
     values = [safe_float(record.get(metric_key)) for record in records]
     values = [value for value in values if value is not None]
@@ -95,22 +132,13 @@ def baseline_value(records: list[Record], metric_key: str) -> float | None:
     return float(statistics.median(values))
 
 
-def regression_percent(metric_key: str, current: float | None, baseline: float | None) -> float | None:
-    if current is None or baseline is None or baseline <= 0:
-        return None
-    metric = next(metric for metric in METRICS if metric.key == metric_key)
-    if metric.lower_is_better:
-        return (current - baseline) / baseline * 100.0
-    return (baseline - current) / baseline * 100.0
-
-
 def build_latest_summary(records: list[Record],
                          *,
                          baseline_window: int = 5,
-                         max_regression: float = 0.05,
                          run_source: str | None = None) -> list[Record]:
     rows: list[Record] = []
-    for (model_id, gpu_type), group in group_by_model_gpu(records).items():
+    for key, group in group_by_comparison_cohort(records).items():
+        model_id, gpu_type = group_identity(key)
         latest_candidates = group
         if run_source:
             latest_candidates = [record for record in group if record_run_source(record) == run_source]
@@ -123,76 +151,87 @@ def build_latest_summary(records: list[Record],
             if record is not latest and record.get("success", True) and is_baseline_eligible_record(record)
         ]
         baseline_records = baseline_pool[-baseline_window:]
+        metric_policies = resolve_metric_policies(latest.get("regression_thresholds"))
 
         metrics: dict[str, Record] = {}
         regressions: list[float] = []
-        for metric in METRICS:
-            current = safe_float(latest.get(metric.key))
-            baseline = baseline_value(baseline_records, metric.key)
-            regression = regression_percent(metric.key, current, baseline)
-            metrics[metric.key] = {
+        failing_metrics: list[str] = []
+        threshold_exceeded_metrics: list[str] = []
+        for policy in metric_policies:
+            current = safe_float(latest.get(policy.key))
+            baseline = baseline_value(baseline_records, policy.key)
+            delta = None
+            if current is not None and baseline is not None:
+                delta = regression_delta(policy, current, baseline)
+            regression = None if delta is None else delta.percent * 100.0
+            metrics[policy.key] = {
                 "current": current,
                 "baseline": baseline,
                 "regression_pct": regression,
-                "label": metric.label,
-                "lower_is_better": metric.lower_is_better,
-                "precision": metric.precision,
+                "absolute_delta": None if delta is None else delta.absolute,
+                "threshold_percent": policy.threshold_percent * 100.0,
+                "threshold_absolute": policy.threshold_absolute,
+                "gated": policy.gated,
+                "threshold_exceeded": False if delta is None else delta.threshold_exceeded,
+                "regressed": False if delta is None else delta.regressed,
+                "label": policy.label,
+                "lower_is_better": policy.lower_is_better,
+                "precision": policy.precision,
             }
             if regression is not None:
                 regressions.append(regression)
+            if delta is not None and delta.threshold_exceeded:
+                threshold_exceeded_metrics.append(policy.key)
+            if delta is not None and delta.regressed:
+                failing_metrics.append(policy.key)
 
         worst_regression = max(regressions) if regressions else None
         success = bool(latest.get("success", True))
         status = "pass" if success else "fail"
 
         rows.append({
-            "model_id":
-            model_id,
-            "gpu_type":
-            gpu_type,
-            "timestamp":
-            latest.get("timestamp"),
-            "commit_sha":
-            latest.get("commit_sha"),
+            "model_id": model_id,
+            "gpu_type": gpu_type,
+            "timestamp": latest.get("timestamp"),
+            "commit_sha": latest.get("commit_sha"),
             **record_metadata(latest),
-            "success":
-            success,
-            "baseline_n":
-            len(baseline_records),
-            "worst_regression_pct":
-            worst_regression,
-            "regression_threshold_pct":
-            max_regression * 100.0,
-            "computed_regression_status":
-            "fail" if worst_regression is not None and worst_regression > max_regression * 100.0 else "pass",
-            "status":
-            status,
-            "metrics":
-            metrics,
+            **record_comparison_metadata(latest),
+            "success": success,
+            "baseline_n": len(baseline_records),
+            "worst_regression_pct": worst_regression,
+            "threshold_exceeded_metrics": threshold_exceeded_metrics,
+            "failing_metrics": failing_metrics,
+            "computed_regression_status": "fail" if failing_metrics else "pass",
+            "status": status,
+            "metrics": metrics,
         })
 
-    return sorted(rows, key=lambda row: (row["status"] != "fail", row["model_id"], row["gpu_type"]))
+    return sorted(rows, key=latest_row_sort_key)
 
 
 def build_trends(records: list[Record]) -> list[Record]:
     trends: list[Record] = []
-    for (model_id, gpu_type), group in group_by_model_gpu(records).items():
+    for key, group in group_by_comparison_cohort(records).items():
+        model_id, gpu_type = group_identity(key)
         points = []
         for record in group:
+            metric_policies = resolve_metric_policies(record.get("regression_thresholds"))
             point = {
                 "timestamp": record.get("timestamp"),
                 "commit_sha": record.get("commit_sha"),
                 **record_metadata(record),
+                **record_comparison_metadata(record),
                 "success": bool(record.get("success", True)),
                 "metrics": {
-                    metric.key: safe_float(record.get(metric.key))
-                    for metric in METRICS
+                    policy.key: safe_float(record.get(policy.key))
+                    for policy in metric_policies
                 },
             }
             points.append(point)
         trends.append({
             "model_id": model_id,
             "gpu_type": gpu_type,
+            **record_comparison_metadata(group[-1]),
             "points": points,
         })
-    return sorted(trends, key=lambda trend: (trend["model_id"], trend["gpu_type"]))
+    return sorted(trends, key=comparison_sort_key)

fastvideo/tests/attention/test_flash_attn_no_pad_resolver.py

@@ -3,15 +3,16 @@
 
 Landed in PR #1225 slice 5 (Attn-QAT 5/12). The resolver centralises the
 varlen-flash-attn import-fallback logic that several backends
-(``bsa_attn.py``, ``video_sparse_attn.py``) used to duplicate. The fallback
+(``bsa_attn.py``, ``video_sparse_attn.py``) used to duplicate. The resolution
 order is:
 
-    1. ``fastvideo.attention.utils.flash_attn_cute``
+    1. ``fastvideo.attention.utils.flash_attn_cute`` -- only when
+       ``FASTVIDEO_FA4=1`` (explicit opt-in), and then it must import or the
+       resolver raises RuntimeError instead of falling through
     2. ``flash_attn_interface``
     3. ``flash_attn``
 
-These tests verify that the resolver picks the highest-priority impl
-available and falls through cleanly on ``ImportError``. CPU-only, no
+These tests verify the opt-in gate and the FA3/FA2 fallthrough. CPU-only, no
 flash-attn install required.
 """
 
@@ -35,8 +36,36 @@ def _reload_resolver_module():
     return importlib.import_module("fastvideo.attention.utils.flash_attn_no_pad")
 
 
-def test_resolver_falls_back_when_cute_unavailable(monkeypatch) -> None:
-    """When ``flash_attn_cute`` is unimportable, resolver tries the next impl."""
+def test_resolver_skips_cute_without_opt_in(monkeypatch) -> None:
+    """Without ``FASTVIDEO_FA4=1`` the resolver must not even attempt the cute
+    import."""
+    monkeypatch.delenv("FASTVIDEO_FA4", raising=False)
+    attempted: list[str] = []
+    real_import = builtins.__import__
+
+    def spying_import(name, globals=None, locals=None, fromlist=(), level=0):
+        attempted.append(name)
+        return real_import(name, globals, locals, fromlist, level)
+
+    monkeypatch.setattr(builtins, "__import__", spying_import)
+
+    mod = _reload_resolver_module()
+    resolved = mod._resolve_flash_attn_varlen_func()
+    assert resolved is not None
+    assert resolved.__name__ == "flash_attn_varlen_func"
+    assert "fastvideo.attention.utils.flash_attn_cute" not in attempted
+
+
+def test_resolver_raises_when_opted_in_but_cute_unavailable(monkeypatch) -> None:
+    """With ``FASTVIDEO_FA4=1`` an unimportable cute build fails loudly instead
+    of silently falling through to FA3/FA2.
+
+    The resolver runs at module import time, so the reload itself must raise.
+    It raises RuntimeError (not ImportError) so importers that treat
+    ImportError as "flash-attn not installed" (``bsa_attn.py``) cannot swallow
+    the opted-in failure.
+    """
+    monkeypatch.setenv("FASTVIDEO_FA4", "1")
     real_import = builtins.__import__
 
     def patched_import(name, globals=None, locals=None, fromlist=(), level=0):
@@ -46,21 +75,17 @@ def test_resolver_falls_back_when_cute_unavailable(monkeypatch) -> None:
 
     monkeypatch.setattr(builtins, "__import__", patched_import)
 
-    mod = _reload_resolver_module()
-    resolved = mod._resolve_flash_attn_varlen_func()
-    assert resolved is not None
-    assert resolved.__name__ == "flash_attn_varlen_func"
+    with pytest.raises(RuntimeError, match="cute disabled for test"):
+        _reload_resolver_module()
 
 
-def test_resolver_returns_flash_attn_when_cute_and_interface_unavailable(monkeypatch) -> None:
+def test_resolver_returns_flash_attn_when_interface_unavailable(monkeypatch) -> None:
     """The terminal fallback is the plain ``flash_attn`` import."""
+    monkeypatch.delenv("FASTVIDEO_FA4", raising=False)
     real_import = builtins.__import__
 
     def patched_import(name, globals=None, locals=None, fromlist=(), level=0):
-        if name in {
-                "fastvideo.attention.utils.flash_attn_cute",
-                "flash_attn_interface",
-        }:
+        if name == "flash_attn_interface":
             raise ImportError(f"{name} disabled for test")
         return real_import(name, globals, locals, fromlist, level)
 

fastvideo/tests/layers/test_rmsnorm_forward_dispatch.py

@@ -0,0 +1,203 @@
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import subprocess
+from pathlib import Path
+from typing import Any
+
+import pytest
+import torch
+import torch.distributed as dist
+from torch.distributed import init_device_mesh
+from torch.distributed.fsdp import CPUOffloadPolicy, fully_shard
+from torch.distributed.tensor import DTensor
+
+from fastvideo.layers.layernorm import RMSNorm
+
+WORLD_SIZE = 2
+HIDDEN_SIZE = 8
+SEED = 1379
+REPO_ROOT = Path(__file__).resolve().parents[3]
+
+
+def _run_torchrun(script_path: Path, mode: str, output_path: Path) -> None:
+    # --standalone binds the rendezvous port atomically, avoiding the
+    # free-port-probe race a hand-picked --master_port would have.
+    cmd = [
+        "torchrun",
+        "--standalone",
+        "--nproc_per_node",
+        str(WORLD_SIZE),
+        str(script_path),
+        "--rmsnorm-fsdp-worker",
+        "--mode",
+        mode,
+        "--output",
+        str(output_path),
+    ]
+    env = os.environ.copy()
+    env.setdefault("TORCHDYNAMO_DISABLE", "1")
+    try:
+        process = subprocess.run(
+            cmd,
+            capture_output=True,
+            text=True,
+            env=env,
+            timeout=120,
+        )
+    except subprocess.TimeoutExpired as error:
+        raise RuntimeError(
+            f"{mode} worker timed out after 120 seconds\n"
+            f"STDOUT:\n{error.stdout}\n"
+            f"STDERR:\n{error.stderr}"
+        ) from error
+    if process.returncode != 0:
+        raise RuntimeError(
+            f"{mode} worker failed with code {process.returncode}\n"
+            f"STDOUT:\n{process.stdout}\n"
+            f"STDERR:\n{process.stderr}"
+        )
+
+
+def _summarize_tensor(tensor: torch.Tensor | Any) -> dict[str, Any]:
+    return {
+        "type": type(tensor).__name__,
+        "is_dtensor": isinstance(tensor, DTensor),
+        "shape": list(tensor.shape) if hasattr(tensor, "shape") else None,
+        "device": str(tensor.device) if hasattr(tensor, "device") else None,
+        "dtype": str(tensor.dtype) if hasattr(tensor, "dtype") else None,
+    }
+
+
+def _run_worker(mode: str, output_path: Path) -> None:
+    if mode not in {
+        "module_no_offload",
+        "direct_no_offload",
+        "module_cpu_offload",
+        "direct_cpu_offload",
+    }:
+        raise ValueError(f"Unsupported mode: {mode}")
+
+    dist.init_process_group("nccl")
+    rank = dist.get_rank()
+    world_size = dist.get_world_size()
+    local_rank = int(os.environ.get("LOCAL_RANK", "0"))
+    device = torch.device(f"cuda:{local_rank}")
+    torch.cuda.set_device(device)
+    torch.manual_seed(SEED + rank)
+
+    try:
+        mesh = init_device_mesh("cuda", (world_size,))
+        norm = RMSNorm(HIDDEN_SIZE, eps=1e-6, has_weight=True).to(device)
+        with torch.no_grad():
+            norm.weight.fill_(1.0)
+
+        fsdp_kwargs: dict[str, Any] = {"mesh": mesh}
+        if mode.endswith("cpu_offload"):
+            fsdp_kwargs["offload_policy"] = CPUOffloadPolicy(pin_memory=False)
+        # fully_shard is applied to the bare RMSNorm to make the hook bypass
+        # observable. Production sharding (fsdp_load.shard_model) only wraps
+        # whole transformer blocks, whose pre-forward all-gather localizes norm
+        # weights before the qk-norm call sites run, so this pins the dispatch
+        # invariant rather than reproducing a production topology.
+        fully_shard(norm, **fsdp_kwargs)
+
+        x = torch.randn(2, 3, HIDDEN_SIZE, device=device, dtype=torch.bfloat16)
+        call_kind = "direct" if mode.startswith("direct") else "module"
+
+        try:
+            if call_kind == "direct":
+                output = norm.forward_native(x)
+            else:
+                output = norm(x)
+            torch.cuda.synchronize(device)
+            result = {
+                "rank": rank,
+                "ok": True,
+                "mode": mode,
+                "weight": _summarize_tensor(norm.weight),
+                "output": _summarize_tensor(output),
+            }
+        except Exception as exc:
+            result = {
+                "rank": rank,
+                "ok": False,
+                "mode": mode,
+                "error_type": type(exc).__name__,
+                "error": str(exc),
+                "weight": _summarize_tensor(norm.weight),
+            }
+
+        gathered = [None for _ in range(world_size)] if rank == 0 else None
+        dist.gather_object(result, object_gather_list=gathered, dst=0)
+        if rank == 0:
+            output_path.write_text(json.dumps(gathered, indent=2), encoding="utf-8")
+        dist.barrier()
+    finally:
+        dist.destroy_process_group()
+
+
+@pytest.mark.parametrize(
+    ("mode", "expect_ok"),
+    [
+        ("module_no_offload", True),
+        ("direct_no_offload", False),
+        ("module_cpu_offload", True),
+        ("direct_cpu_offload", False),
+    ],
+)
+def test_rmsnorm_forward_native_bypasses_fsdp_hooks(mode: str, expect_ok: bool, tmp_path: Path) -> None:
+    if not torch.cuda.is_available():
+        pytest.skip("This test requires CUDA.")
+    if torch.cuda.device_count() < WORLD_SIZE:
+        pytest.skip(f"This test requires at least {WORLD_SIZE} CUDA devices.")
+
+    output_path = tmp_path / f"{mode}.json"
+    _run_torchrun(Path(__file__).resolve(), mode, output_path)
+    results = json.loads(output_path.read_text(encoding="utf-8"))
+    print(f"\n{mode} results:\n{json.dumps(results, indent=2)}")
+
+    if expect_ok:
+        failures = [result for result in results if not result["ok"]]
+        assert not failures, json.dumps(results, indent=2)
+        return
+
+    successes = [result for result in results if result["ok"]]
+    assert not successes, json.dumps(results, indent=2)
+    error_text = "\n".join(result.get("error", "") for result in results)
+    # Pin the specific bypassed-hook failure: "got mixed torch.Tensor and
+    # DTensor" ("Tensor" alone is a substring of "DTensor", so it adds nothing).
+    assert "mixed" in error_text and "DTensor" in error_text, json.dumps(results, indent=2)
+
+
+def test_no_direct_forward_native_calls_in_models() -> None:
+    """Direct .forward_native(...) calls bypass nn.Module.__call__ and FSDP
+    hooks (issue #1379); model code must use module dispatch instead."""
+    models_dir = REPO_ROOT / "fastvideo" / "models"
+    offenders = [
+        str(path.relative_to(REPO_ROOT))
+        for path in sorted(models_dir.rglob("*.py"))
+        if ".forward_native(" in path.read_text(encoding="utf-8")
+    ]
+    assert not offenders, f"Replace .forward_native(...) with module dispatch in: {offenders}"
+
+
+def _parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--rmsnorm-fsdp-worker", action="store_true")
+    parser.add_argument("--mode", type=str, default=None)
+    parser.add_argument("--output", type=str, default=None)
+    return parser.parse_args()
+
+
+if __name__ == "__main__":
+    args = _parse_args()
+    if not args.rmsnorm_fsdp_worker:
+        raise SystemExit("This module is intended to be run by pytest.")
+    if args.mode is None or args.output is None:
+        raise SystemExit("--mode and --output are required in worker mode.")
+    _run_worker(mode=args.mode, output_path=Path(args.output))

fastvideo/tests/modal/launch_l40s_job.py

@@ -32,7 +32,8 @@ import modal
 
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 try:
-    from modal_image_utils import resolve_image_ref  # noqa: E402
+    from modal_image_utils import (  # noqa: E402
+        resolve_image_ref, resolve_uv_torch_backend)
 except ModuleNotFoundError:
     # Remote Modal containers re-import this module but mount only the
     # entrypoint file; the digest resolution already happened at local
@@ -40,6 +41,9 @@ except ModuleNotFoundError:
     def resolve_image_ref(image_ref: str) -> str:
         return image_ref
 
+    def resolve_uv_torch_backend(image_tag: str) -> str | None:
+        return os.environ.get("UV_TORCH_BACKEND")
+
 app = modal.App("fastvideo-gpu-job")
 
 REPO_DIR = "/FastVideo"
@@ -72,12 +76,7 @@ local_secrets = modal.Secret.from_dict({
 # Mutable tags inherit the registry image's baked backend, including custom
 # FASTVIDEO_MODAL_IMAGE overrides. Explicit CUDA tags also work with older
 # images that predate the baked setting, and a caller override always wins.
-uv_torch_backend_override = os.environ.get("UV_TORCH_BACKEND")
-if not uv_torch_backend_override:
-    if "cuda13" in IMAGE_TAG.lower():
-        uv_torch_backend_override = "cu130"
-    elif "cuda12.6" in IMAGE_TAG.lower():
-        uv_torch_backend_override = "cu126"
+uv_torch_backend_override = resolve_uv_torch_backend(IMAGE_TAG)
 
 image = (
     modal.Image.from_registry(IMAGE_REF, add_python="3.12")
@@ -98,6 +97,9 @@ image = (
         "TOKENIZERS_PARALLELISM": "false",
         **({"UV_TORCH_BACKEND": uv_torch_backend_override} if uv_torch_backend_override else {}),
         "FASTVIDEO_ATTENTION_BACKEND": os.environ.get("FASTVIDEO_ATTENTION_BACKEND", "FLASH_ATTN"),
+        # FA4 is opt-in (FASTVIDEO_FA4); keep CI parity with the seeded
+        # references. Caller override wins.
+        "FASTVIDEO_FA4": os.environ.get("FASTVIDEO_FA4", "1"),
     })
 )
 

fastvideo/tests/modal/modal_image_utils.py

@@ -5,6 +5,7 @@ through the ``fastvideo`` package, so they keep working on thin CI hosts that
 have ``modal`` but not torch.
 """
 import json
+import os
 import urllib.request
 
 _REGISTRY = "ghcr.io"
@@ -55,3 +56,23 @@ def resolve_image_ref(image_ref: str) -> str:
         print(f"WARNING: could not resolve {image_ref} to a digest ({error}); "
               "Modal may reuse a stale cached image for this tag.")
     return image_ref
+
+
+def resolve_uv_torch_backend(image_tag: str) -> str | None:
+    """UV_TORCH_BACKEND for a launcher image tag.
+
+    A caller-set UV_TORCH_BACKEND always wins. Otherwise sniff the CUDA
+    version from an explicit image tag (cuda13 -> cu130, cuda12.6 -> cu126)
+    so uv resolves torch against the image's toolkit. Mutable tags (e.g.
+    py3.12-latest) return None and inherit the registry image's baked
+    backend, which keeps a latest-tag CUDA transition safe.
+    """
+    override = os.environ.get("UV_TORCH_BACKEND")
+    if override:
+        return override
+    tag = image_tag.lower()
+    if "cuda13" in tag:
+        return "cu130"
+    if "cuda12.6" in tag:
+        return "cu126"
+    return None

fastvideo/tests/modal/pr_test.py

@@ -5,7 +5,8 @@ import modal
 
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 try:
-    from modal_image_utils import resolve_image_ref  # noqa: E402
+    from modal_image_utils import (  # noqa: E402
+        resolve_image_ref, resolve_uv_torch_backend)
 except ModuleNotFoundError:
     # Remote Modal containers re-import this module but mount only the
     # entrypoint file; the digest resolution already happened at local
@@ -13,6 +14,9 @@ except ModuleNotFoundError:
     def resolve_image_ref(image_ref: str) -> str:
         return image_ref
 
+    def resolve_uv_torch_backend(image_tag: str) -> str | None:
+        return os.environ.get("UV_TORCH_BACKEND")
+
 app = modal.App()
 
 model_vol = modal.Volume.from_name("hf-model-weights")
@@ -24,12 +28,7 @@ print(f"Using image: {image_ref}")
 # Mutable tags inherit the registry image's baked backend, keeping a latest-tag
 # transition safe. Explicit CUDA tags also work with older images that predate
 # the baked setting, and a caller override always wins.
-uv_torch_backend_override = os.environ.get("UV_TORCH_BACKEND")
-if not uv_torch_backend_override:
-    if "cuda13" in image_tag.lower():
-        uv_torch_backend_override = "cu130"
-    elif "cuda12.6" in image_tag.lower():
-        uv_torch_backend_override = "cu126"
+uv_torch_backend_override = resolve_uv_torch_backend(image_tag)
 
 image = (modal.Image.from_registry(
     image_ref, add_python="3.12"
@@ -61,6 +60,10 @@ image = (modal.Image.from_registry(
     **({
         "UV_TORCH_BACKEND": uv_torch_backend_override
     } if uv_torch_backend_override else {}),
+    # FA4 is opt-in (FASTVIDEO_FA4); CI lanes keep it enabled to match the
+    # SSIM/perf baselines. Caller override wins.
+    "FASTVIDEO_FA4":
+    os.environ.get("FASTVIDEO_FA4", "1"),
     "HF_REPO_ID":
     "FastVideo/performance-tracking",
 }))
@@ -273,7 +276,7 @@ def run_self_forcing_tests():
 @app.function(gpu="L40S:1", image=image, timeout=900)
 def run_unit_test():
     run_test(
-        "pytest ./fastvideo/tests/api/ ./fastvideo/tests/contract/ ./fastvideo/tests/dataset/ ./fastvideo/tests/workflow/ ./fastvideo/tests/entrypoints/ ./fastvideo/tests/train/ --ignore=./fastvideo/tests/entrypoints/test_openai_api_integration.py --ignore=./fastvideo/tests/train/models --ignore=./fastvideo/tests/train/methods -vs"
+        "pytest ./fastvideo/tests/api/ ./fastvideo/tests/contract/ ./fastvideo/tests/dataset/ ./fastvideo/tests/workflow/ ./fastvideo/tests/entrypoints/ ./fastvideo/tests/train/ ./fastvideo/tests/stages/ --ignore=./fastvideo/tests/entrypoints/test_openai_api_integration.py --ignore=./fastvideo/tests/train/models --ignore=./fastvideo/tests/train/methods -vs"
     )
 
 

fastvideo/tests/modal/ssim_test.py

@@ -13,7 +13,8 @@ import modal
 
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 try:
-    from modal_image_utils import resolve_image_ref  # noqa: E402
+    from modal_image_utils import (  # noqa: E402
+        resolve_image_ref, resolve_uv_torch_backend)
 except ModuleNotFoundError:
     # Remote Modal containers re-import this module but mount only the
     # entrypoint file; the digest resolution already happened at local
@@ -21,6 +22,9 @@ except ModuleNotFoundError:
     def resolve_image_ref(image_ref: str) -> str:
         return image_ref
 
+    def resolve_uv_torch_backend(image_tag: str) -> str | None:
+        return os.environ.get("UV_TORCH_BACKEND")
+
 app = modal.App()
 
 model_vol = modal.Volume.from_name("hf-model-weights")
@@ -32,12 +36,7 @@ print(f"Using image: {image_ref}")
 # Mutable tags inherit the registry image's baked backend, keeping a latest-tag
 # transition safe. Explicit CUDA tags also work with older images that predate
 # the baked setting, and a caller override always wins.
-uv_torch_backend_override = os.environ.get("UV_TORCH_BACKEND")
-if not uv_torch_backend_override:
-    if "cuda13" in image_tag.lower():
-        uv_torch_backend_override = "cu130"
-    elif "cuda12.6" in image_tag.lower():
-        uv_torch_backend_override = "cu126"
+uv_torch_backend_override = resolve_uv_torch_backend(image_tag)
 
 image = (
     modal.Image.from_registry(image_ref, add_python="3.12")
@@ -64,6 +63,9 @@ image = (
             "BUILDKITE_PULL_REQUEST": os.environ.get("BUILDKITE_PULL_REQUEST", ""),
             "IMAGE_VERSION": image_version,
             **({"UV_TORCH_BACKEND": uv_torch_backend_override} if uv_torch_backend_override else {}),
+            # FA4 is opt-in (FASTVIDEO_FA4); the SSIM references were seeded
+            # with FA4 inference, so keep it enabled in CI. Caller override wins.
+            "FASTVIDEO_FA4": os.environ.get("FASTVIDEO_FA4", "1"),
         }
     )
 )

fastvideo/tests/performance/compare_baseline.py

@@ -5,11 +5,12 @@ This script:
 1) reads current benchmark results from fastvideo/tests/performance/results,
 2) syncs the canonical baseline from the configured HF dataset repo,
 3) compares each current record against the median of up to 5 prior
-   baseline-eligible successful records (filtered by gpu_type),
+   baseline-eligible successful records in the same workload/variant/version,
+   GPU, recipe, hardware, and software cohort,
 4) writes normalized records back to the HF dataset repo according to
    PERF_UPLOAD_POLICY,
-5) exits non-zero if any metric regresses by more than PERF_MAX_REGRESSION
-   (default 5%).
+5) exits non-zero if any gated metric exceeds both its percent and absolute
+   regression floors.
 """
 
 import glob
@@ -21,21 +22,36 @@ from datetime import datetime, timezone
 from typing import Any
 
 try:
-    from .hf_store import (
+    from fastvideo.performance.hf_store import (
         load_records_for_model,
         safe_float,
         sanitize,
         sync_from_hf,
         upload_record,
     )
+    from fastvideo.performance.metric_policy import (
+        MetricPolicy,
+        regression_delta,
+        resolve_metric_policies,
+        serialize_metric_thresholds,
+    )
 except ImportError:
-    from hf_store import (
+    repo_root = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
+    if repo_root not in sys.path:
+        sys.path.insert(0, repo_root)
+    from fastvideo.performance.hf_store import (
         load_records_for_model,
         safe_float,
         sanitize,
         sync_from_hf,
         upload_record,
     )
+    from fastvideo.performance.metric_policy import (
+        MetricPolicy,
+        regression_delta,
+        resolve_metric_policies,
+        serialize_metric_thresholds,
+    )
 
 RESULTS_DIR = os.path.join(
     os.path.dirname(os.path.abspath(__file__)),
@@ -46,25 +62,30 @@ TRACKING_ROOT = os.environ.get(
     "/tmp/perf-tracking",
 )
 PERF_REPORTS_DIR = os.environ.get("PERF_REPORTS_DIR", "/root/data/perf_reports")
-MAX_REGRESSION = float(os.environ.get("PERF_MAX_REGRESSION", "0.05"))
 UPLOAD_POLICY = os.environ.get("PERF_UPLOAD_POLICY", "never").strip().lower()
 VALID_UPLOAD_POLICIES = {"never", "pass", "always"}
 VALID_RUN_SOURCES = {"pr", "local", "scheduled_main", "unknown"}
-METRICS = (
-    ("latency", "Latency", 3),
-    ("throughput", "Throughput", 3),
-    ("memory", "Memory", 1),
-    ("text_encoder_time_s", "Text Enc", 3),
-    ("dit_time_s", "DiT", 3),
-    ("vae_decode_time_s", "VAE Decode", 3),
+IDENTITY_KEYS = (
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+    "recipe",
+    "recipe_fingerprint",
+    "hardware_profile",
+    "hardware_profile_id",
+    "software_profile",
+    "software_profile_id",
+    "environment_metadata",
+    "environment_fingerprint",
+)
+COMPARISON_IDENTITY_KEYS = (
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+    "recipe_fingerprint",
+    "hardware_profile_id",
+    "software_profile_id",
 )
-LOWER_IS_BETTER_METRICS = {
-    "latency",
-    "memory",
-    "text_encoder_time_s",
-    "dit_time_s",
-    "vae_decode_time_s",
-}
 
 
 def _should_persist_tracking() -> bool:
@@ -137,6 +158,46 @@ def _record_metadata(run_source: str, result: dict[str, Any]) -> dict[str, Any]:
     }
 
 
+def _identity_metadata(result: dict[str, Any]) -> dict[str, Any]:
+    metadata = {
+        key: result[key]
+        for key in IDENTITY_KEYS
+        if key in result and result[key] is not None
+    }
+    recipe = result.get("recipe")
+    benchmark = recipe.get("benchmark") if isinstance(recipe, dict) else None
+    if isinstance(benchmark, dict):
+        for key in ("workload_id", "variant_id", "benchmark_version"):
+            if key not in metadata and benchmark.get(key) is not None:
+                metadata[key] = benchmark[key]
+    return metadata
+
+
+def _comparison_identity_filters(record: dict[str, Any]) -> dict[str, str]:
+    missing = [
+        key for key in COMPARISON_IDENTITY_KEYS
+        if key not in record or record[key] is None or (isinstance(record[key], str) and not record[key].strip())
+    ]
+    if len(missing) == len(COMPARISON_IDENTITY_KEYS):
+        # Legacy v1 record: no identity fields at all. Keep the pre-v2
+        # (model_id, gpu_type) rolling-baseline comparison instead of
+        # crashing — v1 configs are documented as loadable.
+        return {}
+    if missing:
+        raise ValueError("Performance record missing required comparison identity fields: " + ", ".join(missing))
+    return {
+        key: str(record[key])
+        for key in COMPARISON_IDENTITY_KEYS
+    }
+
+
+def _format_identity_filters(filters: dict[str, str]) -> str:
+    if not filters:
+        return ""
+    compact = ", ".join(f"{key}={value}" for key, value in filters.items())
+    return f" ({compact})"
+
+
 def _load_current_results() -> list[dict[str, Any]]:
     pattern = os.path.join(RESULTS_DIR, "perf_*.json")
     records: list[dict[str, Any]] = []
@@ -168,8 +229,9 @@ def normalize_performance_result(result: dict[str, Any]) -> dict[str, Any]:
     text_encoder_time = safe_float(result.get("text_encoder_time_s"))
     dit_time = safe_float(result.get("dit_time_s"))
     vae_decode_time = safe_float(result.get("vae_decode_time_s"))
+    metric_policies = resolve_metric_policies(result.get("regression_thresholds"))
 
-    return {
+    record = {
         "model_id": model_id,
         "timestamp": timestamp,
         "commit_sha": commit_sha,
@@ -180,9 +242,12 @@ def normalize_performance_result(result: dict[str, Any]) -> dict[str, Any]:
         "text_encoder_time_s": text_encoder_time,
         "dit_time_s": dit_time,
         "vae_decode_time_s": vae_decode_time,
+        "regression_thresholds": serialize_metric_thresholds(metric_policies),
         "success": True,
         **_record_metadata(_detect_run_source(), result),
     }
+    record.update(_identity_metadata(result))
+    return record
 
 
 def _normalize_record(result: dict[str, Any]) -> dict[str, Any]:
@@ -229,55 +294,41 @@ def _baseline_metric(records: list[dict[str, Any]], key: str) -> float | None:
     return statistics.median(values)
 
 
+def _metric_policy_summary(policy: MetricPolicy) -> str:
+    gated = "gated" if policy.gated else "info"
+    return (
+        f"{gated}, >{policy.threshold_percent * 100:.1f}% "
+        f"and >{policy.threshold_absolute:.{policy.precision}f}"
+    )
+
+
 def _check_regressions(
     current: dict[str, Any],
     baseline_records: list[dict[str, Any]],
-    max_regression: float,
+    metric_policies: tuple[MetricPolicy, ...],
 ) -> list[str]:
     failures: list[str] = []
 
-    for metric, _label, _precision in METRICS:
-        if metric not in LOWER_IS_BETTER_METRICS:
+    for policy in metric_policies:
+        baseline = _baseline_metric(baseline_records, policy.key)
+        curr = safe_float(current.get(policy.key))
+        if baseline is None or curr is None:
             continue
-        baseline = _baseline_metric(baseline_records, metric)
-        curr = safe_float(current.get(metric))
-        if baseline is None or curr is None or baseline <= 0:
+        delta = regression_delta(policy, curr, baseline)
+        if delta is None or not delta.regressed:
             continue
-        regression = (curr - baseline) / baseline
-        if regression > max_regression:
-            failures.append(f"{current['model_id']} {metric} regressed by "
-                            f"{regression * 100:.1f}% "
-                            f"(current={curr:.3f}, baseline_median={baseline:.3f})")
-
-    baseline_tp = _baseline_metric(baseline_records, "throughput")
-    curr_tp = safe_float(current.get("throughput"))
-    if baseline_tp is not None and curr_tp is not None and baseline_tp > 0:
-        regression = (baseline_tp - curr_tp) / baseline_tp
-        if regression > max_regression:
-            failures.append(f"{current['model_id']} throughput regressed by "
-                            f"{regression * 100:.1f}% "
-                            f"(current={curr_tp:.3f}, baseline_median={baseline_tp:.3f})")
+        failures.append(
+            f"{current['model_id']} {policy.key} regressed by "
+            f"{delta.percent * 100:.1f}% and "
+            f"{delta.absolute:.{policy.precision}f} "
+            f"(current={curr:.{policy.precision}f}, "
+            f"baseline_median={baseline:.{policy.precision}f}, "
+            f"threshold={_metric_policy_summary(policy)})"
+        )
 
     return failures
 
 
-def _metric_delta_percent(
-    metric: str,
-    current: dict[str, Any],
-    baseline_records: list[dict[str, Any]],
-) -> float | None:
-    curr = safe_float(current.get(metric))
-    baseline = _baseline_metric(baseline_records, metric)
-    if curr is None or baseline is None or baseline <= 0:
-        return None
-
-    if metric in LOWER_IS_BETTER_METRICS:
-        return (curr - baseline) / baseline * 100.0
-    if metric == "throughput":
-        return (baseline - curr) / baseline * 100.0
-    return None
-
-
 def _compact_value(value: float | None, precision: int = 3) -> str:
     if value is None:
         return "n/a"
@@ -287,23 +338,42 @@ def _compact_value(value: float | None, precision: int = 3) -> str:
 def _build_summary_row(
     record: dict[str, Any],
     baseline_records: list[dict[str, Any]],
+    metric_policies: tuple[MetricPolicy, ...],
     has_failed: bool,
 ) -> dict[str, Any]:
     """Format a single benchmark result as a row for the Markdown table."""
 
-    metric_values: dict[str, dict[str, float | None]] = {}
+    metric_values: dict[str, dict[str, Any]] = {}
     regressions: list[float] = []
-    for metric, _label, _precision in METRICS:
-        curr = safe_float(record.get(metric))
-        baseline = _baseline_metric(baseline_records, metric)
-        regression = _metric_delta_percent(metric, record, baseline_records)
-        metric_values[metric] = {
+    failing_metrics: list[str] = []
+    threshold_exceeded_metrics: list[str] = []
+    for policy in metric_policies:
+        curr = safe_float(record.get(policy.key))
+        baseline = _baseline_metric(baseline_records, policy.key)
+        delta = (
+            regression_delta(policy, curr, baseline)
+            if curr is not None and baseline is not None
+            else None
+        )
+        regression = None if delta is None else delta.percent * 100.0
+        absolute_delta = None if delta is None else delta.absolute
+        metric_values[policy.key] = {
             "curr": curr,
             "base": baseline,
             "regression_pct": regression,
+            "absolute_delta": absolute_delta,
+            "threshold_percent": policy.threshold_percent * 100.0,
+            "threshold_absolute": policy.threshold_absolute,
+            "gated": policy.gated,
+            "threshold_exceeded": False if delta is None else delta.threshold_exceeded,
+            "regressed": False if delta is None else delta.regressed,
         }
         if regression is not None:
             regressions.append(regression)
+        if delta is not None and delta.threshold_exceeded:
+            threshold_exceeded_metrics.append(policy.key)
+        if delta is not None and delta.regressed:
+            failing_metrics.append(policy.key)
 
     worst_regression_pct = max(regressions) if regressions else None
 
@@ -313,40 +383,50 @@ def _build_summary_row(
         "baseline_n": len(baseline_records),
         "metrics": metric_values,
         "worst_regression_pct": worst_regression_pct,
+        "threshold_exceeded_metrics": threshold_exceeded_metrics,
+        "failing_metrics": failing_metrics,
         "failed": has_failed,
     }
 
 
 def _build_markdown_summary(
     summary_rows: list[dict[str, Any]],
-    max_regression: float,
+    metric_policies: tuple[MetricPolicy, ...],
 ) -> str:
     lines = [
         "## Performance Baseline Comparison",
         "",
-        f"Threshold: regressions greater than {max_regression * 100:.1f}% fail",
+        "Threshold: gated metrics fail only when both percent and absolute "
+        "regression floors are exceeded.",
         "",
         ("| Model | GPU | Baseline N | Latency (curr/base) | "
          "Throughput (curr/base) | Memory (curr/base) | "
          "Text Enc (curr/base) | DiT (curr/base) | "
-         "VAE Decode (curr/base) | Worst Regression | Status |"),
-        "|---|---|---:|---|---|---|---|---|---|---:|---|",
+         "VAE Decode (curr/base) | Worst Regression | Exceeded Metrics | "
+         "Failing Metrics | Status |"),
+        "|---|---|---:|---|---|---|---|---|---|---:|---|---|---|",
     ]
 
     for row in summary_rows:
         metric_cells = []
-        for metric, _label, precision in METRICS:
-            values = row["metrics"][metric]
-            metric_cells.append(f"{_compact_value(values['curr'], precision)} / "
-                                f"{_compact_value(values['base'], precision)}")
+        for policy in metric_policies:
+            values = row["metrics"][policy.key]
+            metric_cells.append(f"{_compact_value(values['curr'], policy.precision)} / "
+                                f"{_compact_value(values['base'], policy.precision)}")
 
         worst_reg = ("n/a" if row["worst_regression_pct"] is None else f"{row['worst_regression_pct']:.1f}%")
+        exceeded_metrics = (
+            ", ".join(row["threshold_exceeded_metrics"])
+            if row["threshold_exceeded_metrics"]
+            else "none"
+        )
+        failing_metrics = ", ".join(row["failing_metrics"]) if row["failing_metrics"] else "none"
         status = "FAIL" if row["failed"] else "PASS"
 
         lines.append(f"| {row['model_id']} | {row['gpu_type']} | "
                      f"{row['baseline_n']} | "
                      f"{' | '.join(metric_cells)} | "
-                     f"{worst_reg} | {status} |")
+                     f"{worst_reg} | {exceeded_metrics} | {failing_metrics} | {status} |")
 
     return "\n".join(lines) + "\n"
 
@@ -400,23 +480,38 @@ def main() -> int:
 
     for raw in current_results:
         record = _normalize_record(raw)
+        metric_policies = resolve_metric_policies(record.get("regression_thresholds"))
 
         baseline_records = load_records_for_model(
             TRACKING_ROOT,
             record["model_id"],
             record["gpu_type"],
+            **_comparison_identity_filters(record),
             last_n=5,
             successful_only=True,
             baseline_eligible_only=True,
         )
 
         if not baseline_records:
-            print(f"No baseline for {record['model_id']} on "
-                  f"{record['gpu_type']}. Initializing...")
+            # A brand-new cohort has NO regression gating until history
+            # accumulates — make that loud and machine-readable instead of
+            # an indistinguishable pass, so a cohort shift (intended or
+            # accidental, e.g. an identity-field change) never silently
+            # blinds the comparison.
+            record["baseline_status"] = "initialized_new_cohort"
+            print("=" * 72)
+            print(f"WARNING: NO BASELINE — initializing a NEW cohort for "
+                  f"{record['model_id']} on {record['gpu_type']}"
+                  f"{_format_identity_filters(_comparison_identity_filters(record))}")
+            print("Regression gating is INACTIVE for this cohort until "
+                  "baseline history accumulates. If this cohort shift is "
+                  "unexpected, check the identity fields above.")
+            print("=" * 72)
             failures: list[str] = []
             record["success"] = True
         else:
-            failures = _check_regressions(record, baseline_records, MAX_REGRESSION)
+            record["baseline_status"] = "compared"
+            failures = _check_regressions(record, baseline_records, metric_policies)
         if static_threshold_failed:
             failures.append(f"{record['model_id']} fixed-threshold phase failed "
                             f"(PERF_PYTEST_RC={os.environ.get('PERF_PYTEST_RC')})")
@@ -434,10 +529,10 @@ def main() -> int:
             print("Tracking upload skipped for "
                   f"{record['model_id']} ({record['run_source']}, success={record['success']})")
 
-        summary_rows.append(_build_summary_row(record, baseline_records, bool(failures)))
+        summary_rows.append(_build_summary_row(record, baseline_records, metric_policies, bool(failures)))
 
     commit_sha = os.environ.get("BUILDKITE_COMMIT", "unknown")[:7]
-    markdown = _build_markdown_summary(summary_rows, MAX_REGRESSION)
+    markdown = _build_markdown_summary(summary_rows, resolve_metric_policies(None))
     _emit_markdown_summary(markdown, commit_sha)
 
     if all_failures:

fastvideo/tests/performance/dashboard.py

@@ -1,12 +1,19 @@
 # SPDX-License-Identifier: Apache-2.0
 import os
+import sys
 from html import escape
 from datetime import datetime
 
 import plotly.express as px
 import pandas as pd
 
-from hf_store import sync_from_hf, load_as_dataframe
+try:
+    from fastvideo.performance.hf_store import load_as_dataframe, sync_from_hf
+except ImportError:
+    repo_root = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
+    if repo_root not in sys.path:
+        sys.path.insert(0, repo_root)
+    from fastvideo.performance.hf_store import load_as_dataframe, sync_from_hf
 
 TRACKING_ROOT = os.environ.get(
     "PERFORMANCE_TRACKING_ROOT",
@@ -22,15 +29,76 @@ METRICS = (
     "dit_time_s",
     "vae_decode_time_s",
 )
+COMPARISON_COHORT_KEYS = (
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+    "recipe_fingerprint",
+    "hardware_profile_id",
+    "software_profile_id",
+)
+GROUP_KEYS = ("model_id", "gpu_type", *COMPARISON_COHORT_KEYS)
+
+
+def _cohort_value(value: object) -> str:
+    if value is None:
+        return ""
+    try:
+        if pd.isna(value):
+            return ""
+    except (TypeError, ValueError):
+        pass
+    return str(value)
+
+
+def _display_value(value: object) -> str:
+    return _cohort_value(value) or "legacy"
+
+
+def _short_value(value: object) -> str:
+    text = _display_value(value)
+    if text == "legacy" or len(text) <= 14:
+        return text
+    return text[:12]
+
+
+def _cohort_title(group_key: tuple[object, ...]) -> str:
+    workload = _display_value(group_key[2])
+    variant = _display_value(group_key[3])
+    version = _display_value(group_key[4])
+    version_label = version if version == "legacy" else f"v{version}"
+    return f"{workload} / {variant} / {version_label}"
+
+
+def _cohort_detail(group_key: tuple[object, ...]) -> str:
+    return " | ".join((
+        f"recipe {_short_value(group_key[5])}",
+        _short_value(group_key[6]),
+        _short_value(group_key[7]),
+    ))
+
+
+def _group_record(group_key: tuple[object, ...]) -> dict[str, str]:
+    return {
+        key: _cohort_value(value)
+        for key, value in zip(GROUP_KEYS, group_key)
+    }
+
+
+def _dashboard_frame(df: pd.DataFrame) -> pd.DataFrame:
+    dashboard_df = df.copy()
+    for key in GROUP_KEYS:
+        if key not in dashboard_df.columns:
+            dashboard_df[key] = ""
+        dashboard_df[key] = dashboard_df[key].map(_cohort_value)
+    return dashboard_df
 
 # -----------------------------
 # 1. Grouping
 # -----------------------------
 def group_data(df: pd.DataFrame):
-    # Group only by model+GPU so each group produces a time-series line.
-    # config_id (commit SHA) is carried as a column for hover/color use.
-    keys = ["model_id", "gpu_type"]
-    return df.groupby(keys, dropna=False)
+    # Group by the same comparison cohort used by baseline gating.
+    return _dashboard_frame(df).groupby(list(GROUP_KEYS), dropna=False)
 
 # -----------------------------
 # 2. Plot builder
@@ -39,15 +107,20 @@ def build_plots(df: pd.DataFrame) -> tuple[list, list[dict[str, object]]]:
     figs = []
     skipped_metrics: list[dict[str, object]] = []
 
-    for (model_id, gpu_type), g in group_data(df):
+    for group_key, g in group_data(df):
+        model_id, gpu_type = group_key[:2]
+        group_record = _group_record(group_key)
+        cohort_title = _cohort_title(group_key)
+        cohort_detail = _cohort_detail(group_key)
         g = g.sort_values("timestamp")
 
         # One chart per metric so the y-axes aren't on wildly different scales
         for metric in METRICS:
             if metric not in g.columns:
                 skipped_metrics.append({
-                    "model_id": model_id,
-                    "gpu_type": gpu_type,
+                    **group_record,
+                    "cohort": cohort_title,
+                    "cohort_detail": cohort_detail,
                     "metric": metric,
                     "reason": "column missing from loaded records",
                     "records": len(g),
@@ -58,8 +131,9 @@ def build_plots(df: pd.DataFrame) -> tuple[list, list[dict[str, object]]]:
             non_null = int(g[metric].notna().sum())
             if non_null == 0:
                 skipped_metrics.append({
-                    "model_id": model_id,
-                    "gpu_type": gpu_type,
+                    **group_record,
+                    "cohort": cohort_title,
+                    "cohort_detail": cohort_detail,
                     "metric": metric,
                     "reason": "no non-null values in loaded records",
                     "records": len(g),
@@ -72,8 +146,8 @@ def build_plots(df: pd.DataFrame) -> tuple[list, list[dict[str, object]]]:
                 x="timestamp",
                 y=metric,
                 markers=True,
-                hover_data=["config_id", "commit_sha"],
-                title=f"{model_id} | {gpu_type} | {metric}",
+                hover_data=["config_id", "commit_sha", *COMPARISON_COHORT_KEYS],
+                title=f"{model_id} | {gpu_type} | {cohort_title} | {cohort_detail} | {metric}",
                 labels={"timestamp": "Time", metric: metric},
             )
             figs.append(fig)
@@ -88,7 +162,7 @@ def render_skipped_metrics(skipped_metrics: list[dict[str, object]]) -> str:
     rows = [
         "<h3>Skipped Metric Plots</h3>",
         "<table>",
-        ("<thead><tr><th>Model</th><th>GPU</th><th>Metric</th>"
+        ("<thead><tr><th>Model</th><th>GPU</th><th>Cohort</th><th>Metric</th>"
          "<th>Records</th><th>Non-null</th><th>Reason</th></tr></thead>"),
         "<tbody>",
     ]
@@ -97,6 +171,7 @@ def render_skipped_metrics(skipped_metrics: list[dict[str, object]]) -> str:
             "<tr>"
             f"<td>{escape(str(item['model_id']))}</td>"
             f"<td>{escape(str(item['gpu_type']))}</td>"
+            f"<td>{escape(str(item['cohort']))}<br><code>{escape(str(item['cohort_detail']))}</code></td>"
             f"<td>{escape(str(item['metric']))}</td>"
             f"<td>{item['records']}</td>"
             f"<td>{item['non_null']}</td>"
@@ -158,6 +233,7 @@ def main() -> None:
         print("Skipped metric plots:")
         for item in skipped_metrics:
             print(f"  - {item['model_id']} | {item['gpu_type']} | "
+                  f"{item['cohort']} | {item['cohort_detail']} | "
                   f"{item['metric']}: {item['reason']} "
                   f"({item['non_null']}/{item['records']} non-null)")
     print(f"Generated {len(figs)} metric plot(s)")

fastvideo/tests/performance/identity.py

@@ -0,0 +1,435 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Comparable identity helpers for performance benchmark records."""
+
+from __future__ import annotations
+
+import dataclasses
+import enum
+import hashlib
+import importlib.metadata
+import json
+import os
+import platform
+import re
+from collections.abc import Mapping, Sequence
+from pathlib import Path
+from typing import Any
+
+import torch
+
+RECIPE_SCHEMA_VERSION = 1
+PROFILE_ID_LENGTH = 16
+_PATH_EXCLUDED_GENERATION_KEYS = {"output_path", "output_video_name"}
+_PROMPT_KEYS = {"prompt", "negative_prompt", "neg_prompt"}
+REQUIRED_BENCHMARK_IDENTITY_KEYS = (
+    "benchmark_id",
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+)
+_PROFILE_ENV_VARS = (
+    "CUDA_VISIBLE_DEVICES",
+    "FASTVIDEO_ATTENTION_BACKEND",
+    "IMAGE_VERSION",
+    "UV_TORCH_BACKEND",
+)
+_PACKAGE_DISTRIBUTIONS = {
+    "fastvideo_kernel": ("fastvideo-kernel", "fastvideo_kernel"),
+    "flash_attn": ("flash-attn", "flash_attn"),
+    "flash_attn_4": ("flash-attn-4",),
+    "flash_attention_fp4": ("flash-attention-fp4",),
+    "flashinfer": ("flashinfer-python", "flashinfer"),
+    "nvidia_cutlass_dsl": ("nvidia-cutlass-dsl", "cutlass-dsl"),
+    "sageattention": ("sageattention",),
+    "sageattn3": ("sageattn3",),
+    "triton": ("triton",),
+    "xformers": ("xformers",),
+}
+
+
+def canonical_json(payload: Any) -> str:
+    """Return deterministic, compact JSON for a recipe/profile payload."""
+
+    return json.dumps(
+        _canonicalize(payload),
+        sort_keys=True,
+        separators=(",", ":"),
+        ensure_ascii=True,
+    )
+
+
+def sha256_hexdigest(payload: str | bytes) -> str:
+    if isinstance(payload, str):
+        payload = payload.encode("utf-8")
+    return hashlib.sha256(payload).hexdigest()
+
+
+def recipe_fingerprint(recipe: Mapping[str, Any]) -> str:
+    # Runtime-resolved values are audit metadata, not identity: hashing the
+    # resolved HF snapshot revision would churn the cohort (and via the
+    # no-baseline init path, silently reset the rolling baseline) on ANY
+    # upstream repo commit, including model-card edits with unchanged
+    # weights. The fingerprint is pinned to declared inputs only; the
+    # resolved values stay in the stored recipe for auditability.
+    pruned = {
+        key: (dict(value) if isinstance(value, Mapping) else value)
+        for key, value in recipe.items()
+    }
+    model = pruned.get("model")
+    if isinstance(model, dict):
+        model.pop("resolved_revision", None)
+    attention = pruned.get("attention")
+    if isinstance(attention, dict):
+        # Same reasoning as resolved_revision: with requested_backend="auto",
+        # a silent runtime fallback (e.g. a broken flash-attn wheel) must NOT
+        # open a fresh ungated cohort and seed degraded numbers as the new
+        # baseline — it should fail the regression gate in the old cohort.
+        # Intentional backend changes are declared via requested_backend,
+        # which stays in the hash.
+        attention.pop("resolved_backend", None)
+    return sha256_hexdigest(canonical_json(pruned))
+
+
+def profile_id(prefix: str, profile: Mapping[str, Any]) -> str:
+    return f"{prefix}-{sha256_hexdigest(canonical_json(profile))[:PROFILE_ID_LENGTH]}"
+
+
+def hardware_profile_id(profile: Mapping[str, Any]) -> str:
+    return profile_id("hw", profile)
+
+
+def software_profile_id(profile: Mapping[str, Any]) -> str:
+    return profile_id("sw", profile)
+
+
+def environment_fingerprint(metadata: Mapping[str, Any]) -> str:
+    return profile_id("env", metadata)
+
+
+def benchmark_identity_from_config(cfg: Mapping[str, Any]) -> dict[str, Any]:
+    missing = [
+        key for key in REQUIRED_BENCHMARK_IDENTITY_KEYS
+        if _none_if_empty(cfg.get(key)) is None
+    ]
+    if missing:
+        raise ValueError("Benchmark config missing required identity fields: " + ", ".join(missing))
+    return {
+        key: cfg[key]
+        for key in REQUIRED_BENCHMARK_IDENTITY_KEYS
+    }
+
+
+def build_recipe_from_benchmark_config(
+    cfg: Mapping[str, Any],
+    *,
+    attention_backend: str | None = None,
+    resolved_attention_backend: Any | None = None,
+    resolved_model_revision: Any | None = None,
+    measured_prompts: Sequence[Any] | None = None,
+) -> dict[str, Any]:
+    """Build a deterministic recipe document from a benchmark config.
+
+    The recipe captures fields that describe what benchmark workload was run. It
+    intentionally excludes timings, timestamps, commit metadata, and output paths.
+    """
+
+    model = dict(cfg.get("model") or {})
+    init_kwargs = dict(cfg.get("init_kwargs") or {})
+    generation_kwargs = dict(cfg.get("generation_kwargs") or {})
+    benchmark_identity = benchmark_identity_from_config(cfg)
+    prompts = list(measured_prompts) if measured_prompts is not None else list(
+        cfg.get("test_prompts") or ["A cinematic video."])
+    if attention_backend is None:
+        attention_backend = os.environ.get("FASTVIDEO_ATTENTION_BACKEND")
+
+    negative_prompt = generation_kwargs.pop("negative_prompt", generation_kwargs.pop("neg_prompt", None))
+    generation_recipe = {
+        key: value
+        for key, value in generation_kwargs.items()
+        if key not in _PATH_EXCLUDED_GENERATION_KEYS and key not in _PROMPT_KEYS
+    }
+
+    return {
+        "recipe_schema_version": RECIPE_SCHEMA_VERSION,
+        "benchmark": benchmark_identity,
+        "model": {
+            "model_path": normalize_model_path(model.get("model_path")),
+            "revision": _none_if_empty(model.get("revision") or cfg.get("revision")),
+            "resolved_revision": _none_if_empty(resolved_model_revision),
+        },
+        "init_kwargs": init_kwargs,
+        "generation_kwargs": generation_recipe,
+        "inputs": {
+            "prompt_count": len(prompts),
+            "prompt_sha256": [_optional_digest(prompt) for prompt in prompts],
+            "negative_prompt_sha256": _optional_digest(negative_prompt),
+        },
+        "attention": {
+            "requested_backend": _none_if_empty(attention_backend) or "auto",
+            "resolved_backend": _none_if_empty(resolved_attention_backend),
+        },
+    }
+
+
+def normalize_model_path(value: Any) -> str | None:
+    if value is None:
+        return None
+    text = os.fspath(value) if isinstance(value, os.PathLike) else str(value)
+    text = text.strip()
+    if not text:
+        return None
+    if _looks_like_local_path(text):
+        return Path(text).expanduser().as_posix()
+    return re.sub(r"/+", "/", text)
+
+
+def resolved_revision_from_model_path(value: Any) -> str | None:
+    normalized = normalize_model_path(value)
+    if normalized is None:
+        return None
+    parts = Path(normalized).parts
+    for index, part in enumerate(parts[:-1]):
+        if part == "snapshots":
+            revision = parts[index + 1]
+            if re.fullmatch(r"[0-9a-f]{40}", revision):
+                return revision
+    return None
+
+
+def hardware_profile(
+    *,
+    num_gpus: int | None = None,
+    gpu_devices: Sequence[Mapping[str, Any]] | None = None,
+    device_type: str | None = None,
+    interconnect: str | None = None,
+) -> dict[str, Any]:
+    """Return normalized hardware cohort metadata.
+
+    ``gpu_devices`` is an injection point for tests and non-CUDA callers.
+    """
+
+    if gpu_devices is not None:
+        devices = [_normalize_gpu_device(device) for device in gpu_devices]
+        return {
+            "device_type": device_type or "cuda",
+            "gpu_count": num_gpus if num_gpus is not None else len(devices),
+            "gpus": devices,
+            "interconnect": interconnect or "unknown",
+        }
+
+    if not torch.cuda.is_available():
+        return {
+            "device_type": device_type or "cpu",
+            "gpu_count": 0,
+            "gpus": [],
+            "interconnect": interconnect or "none",
+        }
+
+    gpu_count = int(num_gpus if num_gpus is not None else torch.cuda.device_count())
+    devices = []
+    cuda_device_count = torch.cuda.device_count()
+    for device_id in range(gpu_count):
+        if device_id >= cuda_device_count:
+            devices.append(_normalize_gpu_device({}))
+            continue
+        props = torch.cuda.get_device_properties(device_id)
+        capability = torch.cuda.get_device_capability(device_id)
+        devices.append(
+            _normalize_gpu_device({
+                "name": props.name,
+                "memory_bytes": props.total_memory,
+                "compute_capability": f"{capability[0]}.{capability[1]}",
+            }))
+
+    return {
+        "device_type": device_type or "cuda",
+        "gpu_count": gpu_count,
+        "gpus": devices,
+        "interconnect": interconnect or _detect_cuda_interconnect(gpu_count),
+    }
+
+
+def software_profile(
+    *,
+    python_version: str | None = None,
+    torch_version: str | None = None,
+    cuda_version: str | None = None,
+    package_versions: Mapping[str, str | None] | None = None,
+) -> dict[str, Any]:
+    """Return normalized software cohort metadata.
+
+    Python, PyTorch, and CUDA use major/minor cohorts. Attention and kernel
+    packages keep exact versions because patch releases can change performance.
+    """
+
+    versions = dict(package_versions) if package_versions is not None else _installed_package_versions()
+    return {
+        "python": _major_minor(python_version or platform.python_version()),
+        "pytorch": _major_minor(torch_version or torch.__version__),
+        "cuda": _major_minor(cuda_version or torch.version.cuda),
+        "packages": {
+            name: str(version)
+            for name, version in sorted(versions.items())
+            if version
+        },
+    }
+
+
+def environment_metadata(
+    *,
+    env: Mapping[str, str] | None = None,
+    package_versions: Mapping[str, str | None] | None = None,
+    hardware: Mapping[str, Any] | None = None,
+    software: Mapping[str, Any] | None = None,
+) -> dict[str, Any]:
+    """Return audit metadata kept separate from comparable identity keys."""
+
+    source_env = env if env is not None else os.environ
+    full_package_versions = dict(package_versions) if package_versions is not None else _installed_package_versions()
+    return {
+        "python": {
+            "version": platform.python_version(),
+            "implementation": platform.python_implementation(),
+        },
+        "platform": {
+            "system": platform.system(),
+            "machine": platform.machine(),
+            "release": platform.release(),
+        },
+        "torch": {
+            "version": torch.__version__,
+            "cuda": torch.version.cuda,
+            "cudnn": torch.backends.cudnn.version(),
+        },
+        "env": {
+            key: source_env.get(key)
+            for key in _PROFILE_ENV_VARS
+            if source_env.get(key) is not None
+        },
+        "packages": {
+            key: value
+            for key, value in sorted(full_package_versions.items())
+            if value
+        },
+        "hardware_profile": dict(hardware) if hardware is not None else hardware_profile(),
+        "software_profile": dict(software) if software is not None else software_profile(
+            package_versions=full_package_versions),
+    }
+
+
+def _canonicalize(value: Any) -> Any:
+    if dataclasses.is_dataclass(value) and not isinstance(value, type):
+        return _canonicalize(dataclasses.asdict(value))
+    if isinstance(value, Mapping):
+        return {
+            str(key): _canonicalize(value[key])
+            for key in sorted(value, key=lambda item: str(item))
+        }
+    if isinstance(value, (set, frozenset)):
+        return [_canonicalize(item) for item in sorted(value, key=lambda item: str(item))]
+    if isinstance(value, tuple):
+        return [_canonicalize(item) for item in value]
+    if isinstance(value, list):
+        return [_canonicalize(item) for item in value]
+    if isinstance(value, enum.Enum):
+        return value.name
+    if isinstance(value, Path):
+        return value.expanduser().as_posix()
+    if isinstance(value, os.PathLike):
+        return os.fspath(value)
+    if isinstance(value, type):
+        return f"{value.__module__}.{value.__qualname__}"
+    if isinstance(value, torch.dtype):
+        return str(value).removeprefix("torch.")
+    return value
+
+
+def _none_if_empty(value: Any) -> Any | None:
+    if value is None:
+        return None
+    if isinstance(value, str) and not value.strip():
+        return None
+    return value
+
+
+def _optional_digest(value: Any) -> str | None:
+    if value is None:
+        return None
+    return sha256_hexdigest(str(value))
+
+
+def _looks_like_local_path(value: str) -> bool:
+    if value.startswith(("/", "./", "../", "~")):
+        return True
+    looks_like_hf_repo = re.match(r"^[A-Za-z0-9_.-]+/[A-Za-z0-9_.-]+$", value)
+    has_separator = "/" in value or "\\" in value or os.sep in value
+    return has_separator and looks_like_hf_repo is None
+
+
+def _normalize_gpu_device(device: Mapping[str, Any]) -> dict[str, Any]:
+    memory_gb = device.get("memory_gb")
+    if memory_gb is None and device.get("memory_bytes") is not None:
+        memory_gb = round(float(device["memory_bytes"]) / (1024**3))
+    return {
+        "name": _none_if_empty(device.get("name")) or "unknown",
+        "memory_gb": memory_gb,
+        "compute_capability": _none_if_empty(device.get("compute_capability")),
+    }
+
+
+def _detect_cuda_interconnect(gpu_count: int) -> str:
+    if gpu_count <= 1:
+        return "single_gpu"
+    try:
+        from fastvideo.platforms import current_platform
+
+        if current_platform.is_cuda() and current_platform.is_full_nvlink(list(range(gpu_count))):
+            return "full_nvlink"
+        return "none_or_partial"
+    except Exception:
+        return "unknown"
+
+
+def _installed_package_versions() -> dict[str, str | None]:
+    return {
+        name: _distribution_version(*distribution_names)
+        for name, distribution_names in _PACKAGE_DISTRIBUTIONS.items()
+    }
+
+
+def _distribution_version(*names: str) -> str | None:
+    for name in names:
+        try:
+            return importlib.metadata.version(name)
+        except importlib.metadata.PackageNotFoundError:
+            continue
+    return None
+
+
+def _major_minor(version: str | None) -> str | None:
+    if not version:
+        return None
+    match = re.search(r"(\d+)\.(\d+)", version)
+    if match is None:
+        return version
+    return f"{match.group(1)}.{match.group(2)}"
+
+
+__all__ = [
+    "RECIPE_SCHEMA_VERSION",
+    "REQUIRED_BENCHMARK_IDENTITY_KEYS",
+    "benchmark_identity_from_config",
+    "canonical_json",
+    "environment_fingerprint",
+    "environment_metadata",
+    "hardware_profile",
+    "hardware_profile_id",
+    "normalize_model_path",
+    "profile_id",
+    "recipe_fingerprint",
+    "resolved_revision_from_model_path",
+    "build_recipe_from_benchmark_config",
+    "sha256_hexdigest",
+    "software_profile",
+    "software_profile_id",
+]

fastvideo/tests/performance/test_benchmark_config_validation.py

@@ -0,0 +1,116 @@
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+
+from fastvideo.tests.performance.test_inference_performance import (
+    _benchmark_display_id,
+    _config_identity_metadata,
+    _is_v2_config,
+    _validate_benchmark_config,
+)
+
+
+def _v2_config():
+    return {
+        "benchmark_id": "wan-t2v-1.3b-2gpu",
+        "config_schema_version": 2,
+        "workload_id": "wan-t2v-1.3b",
+        "variant_id": "canonical",
+        "benchmark_version": 1,
+    }
+
+
+def test_v1_benchmark_config_without_schema_version_validates():
+    cfg = {
+        "benchmark_id": "legacy-benchmark",
+    }
+
+    _validate_benchmark_config(cfg, "legacy.json")
+
+    assert _is_v2_config(cfg) is False
+    assert _config_identity_metadata(cfg) == {}
+    assert _benchmark_display_id(cfg) == "legacy-benchmark"
+
+
+def test_v2_benchmark_config_identity_validates_and_is_preserved():
+    cfg = _v2_config()
+    cfg["quality_metadata"] = {"some": "data"}
+
+    _validate_benchmark_config(cfg, "wan.json")
+
+    assert _is_v2_config(cfg) is True
+    assert _config_identity_metadata(cfg) == {
+        "config_schema_version": 2,
+        "workload_id": "wan-t2v-1.3b",
+        "variant_id": "canonical",
+        "benchmark_version": 1,
+        "quality_metadata": {"some": "data"},
+    }
+
+
+def test_v2_benchmark_config_missing_identity_fields_fails_clearly():
+    cfg = _v2_config()
+    del cfg["variant_id"]
+    del cfg["benchmark_version"]
+
+    expected = "wan.json: missing required v2 identity fields: variant_id, benchmark_version"
+    with pytest.raises(ValueError, match=expected):
+        _validate_benchmark_config(cfg, "wan.json")
+
+
+@pytest.mark.parametrize(
+    ("field", "value"),
+    [
+        ("workload_id", {}),
+        ("workload_id", ""),
+        ("workload_id", "   "),
+        ("variant_id", []),
+        ("variant_id", ""),
+        ("variant_id", "   "),
+    ],
+)
+def test_v2_benchmark_config_rejects_invalid_string_identity_values(field, value):
+    cfg = _v2_config()
+    cfg[field] = value
+
+    expected = f"wan.json: v2 identity field {field!r} must be a non-empty string"
+    with pytest.raises(ValueError, match=expected):
+        _validate_benchmark_config(cfg, "wan.json")
+
+
+@pytest.mark.parametrize("value", [None, "1", 1.5, True])
+def test_v2_benchmark_config_rejects_invalid_benchmark_version_values(value):
+    cfg = _v2_config()
+    cfg["benchmark_version"] = value
+
+    expected = "wan.json: v2 identity field 'benchmark_version' must be an integer"
+    with pytest.raises(ValueError, match=expected):
+        _validate_benchmark_config(cfg, "wan.json")
+
+
+def test_partial_v2_identity_requires_schema_version():
+    cfg = {
+        "benchmark_id": "wan-t2v-1.3b-2gpu",
+        "workload_id": "wan-t2v-1.3b",
+        "variant_id": "canonical",
+        "benchmark_version": 1,
+    }
+
+    expected = "wan.json: v2 benchmark identity fields require config_schema_version=2"
+    with pytest.raises(ValueError, match=expected):
+        _validate_benchmark_config(cfg, "wan.json")
+
+
+def test_optional_v2_metadata_fields_must_be_objects():
+    cfg = {
+        "benchmark_id": "wan-t2v-1.3b-2gpu",
+        "config_schema_version": 2,
+        "workload_id": "wan-t2v-1.3b",
+        "variant_id": "canonical",
+        "benchmark_version": 1,
+        "quality_metadata": ["not", "an", "object"],
+    }
+
+    expected = "wan.json: optional v2 metadata field 'quality_metadata' must be an object"
+    with pytest.raises(ValueError, match=expected):
+        _validate_benchmark_config(cfg, "wan.json")

fastvideo/tests/performance/test_compare_baseline_policy.py

@@ -1,6 +1,9 @@
 # SPDX-License-Identifier: Apache-2.0
 
+import pytest
+
 from fastvideo.tests.performance import compare_baseline
+from fastvideo.performance.metric_policy import resolve_metric_policies
 
 
 def _raw_result():
@@ -72,9 +75,261 @@ def test_normalized_record_includes_source_metadata(monkeypatch):
     assert record["job_id"] == "job-1"
 
 
+def test_normalized_record_includes_effective_regression_thresholds(monkeypatch):
+    monkeypatch.setenv("PERF_RUN_SOURCE", "pr")
+    raw = _raw_result()
+    raw["regression_thresholds"] = {
+        "latency": {
+            "threshold_percent": 0.09,
+            "threshold_absolute": 0.75,
+            "gated": True,
+        },
+        "throughput": {
+            "gated": False,
+        },
+    }
+
+    record = compare_baseline.normalize_performance_result(raw)
+
+    assert record["regression_thresholds"]["latency"] == {
+        "threshold_percent": 0.09,
+        "threshold_absolute": 0.75,
+        "gated": True,
+    }
+    assert record["regression_thresholds"]["throughput"]["gated"] is False
+
+
+def test_invalid_regression_threshold_container_uses_defaults():
+    policies = resolve_metric_policies(["not", "a", "mapping"])
+
+    latency = next(policy for policy in policies if policy.key == "latency")
+    assert latency.threshold_percent == 0.08
+    assert latency.threshold_absolute == 0.5
+    assert latency.gated is True
+
+
+def test_boolean_regression_threshold_values_are_ignored():
+    policies = resolve_metric_policies({
+        "latency": {
+            "threshold_percent": True,
+            "threshold_absolute": False,
+            "gated": "false",
+        }
+    })
+
+    latency = next(policy for policy in policies if policy.key == "latency")
+    assert latency.threshold_percent == 0.08
+    assert latency.threshold_absolute == 0.5
+    assert latency.gated is False
+def test_normalized_record_preserves_identity_metadata(monkeypatch):
+    monkeypatch.setenv("PERF_RUN_SOURCE", "pr")
+    raw = _raw_result()
+    raw.update({
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+        "recipe": {
+            "recipe_schema_version": 1,
+        },
+        "recipe_fingerprint": "recipe-1",
+        "hardware_profile": {
+            "gpu_count": 1,
+        },
+        "hardware_profile_id": "hw-1",
+        "software_profile": {
+            "python": "3.12",
+        },
+        "software_profile_id": "sw-1",
+        "environment_metadata": {
+            "env": {
+                "IMAGE_VERSION": "latest",
+            },
+        },
+        "environment_fingerprint": "env-1",
+    })
+
+    record = compare_baseline.normalize_performance_result(raw)
+
+    assert record["workload_id"] == "wan-t2v"
+    assert record["variant_id"] == "1.3b-sp2"
+    assert record["benchmark_version"] == 2
+    assert record["recipe"] == {"recipe_schema_version": 1}
+    assert record["recipe_fingerprint"] == "recipe-1"
+    assert record["hardware_profile"] == {"gpu_count": 1}
+    assert record["hardware_profile_id"] == "hw-1"
+    assert record["software_profile"] == {"python": "3.12"}
+    assert record["software_profile_id"] == "sw-1"
+    assert record["environment_metadata"] == {"env": {"IMAGE_VERSION": "latest"}}
+    assert record["environment_fingerprint"] == "env-1"
+
+
+def test_normalized_record_reads_identity_labels_from_recipe(monkeypatch):
+    monkeypatch.setenv("PERF_RUN_SOURCE", "pr")
+    raw = _raw_result()
+    raw["recipe"] = {
+        "benchmark": {
+            "workload_id": "wan-t2v",
+            "variant_id": "1.3b-sp2",
+            "benchmark_version": 2,
+        },
+    }
+
+    record = compare_baseline.normalize_performance_result(raw)
+
+    assert record["workload_id"] == "wan-t2v"
+    assert record["variant_id"] == "1.3b-sp2"
+    assert record["benchmark_version"] == 2
+
+
+def test_comparison_identity_filters_use_full_issue_key():
+    record = {
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+        "recipe_fingerprint": "recipe-1",
+        "hardware_profile_id": "hw-1",
+        "software_profile_id": "sw-1",
+    }
+
+    assert compare_baseline._comparison_identity_filters(record) == {
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": "2",
+        "recipe_fingerprint": "recipe-1",
+        "hardware_profile_id": "hw-1",
+        "software_profile_id": "sw-1",
+    }
+
+
+def test_comparison_identity_filters_require_full_issue_key():
+    record = {
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 0,
+        "recipe_fingerprint": "recipe-1",
+        "hardware_profile_id": "hw-1",
+    }
+
+    with pytest.raises(ValueError, match="software_profile_id"):
+        compare_baseline._comparison_identity_filters(record)
+
+
+def test_comparison_identity_filters_keep_zero_version():
+    record = {
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 0,
+        "recipe_fingerprint": "recipe-1",
+        "hardware_profile_id": "hw-1",
+        "software_profile_id": "sw-1",
+    }
+
+    assert compare_baseline._comparison_identity_filters(record)["benchmark_version"] == "0"
+
+
 def test_baseline_eligibility_only_for_successful_scheduled_main():
     assert compare_baseline._is_baseline_eligible("scheduled_main", True) is True
     assert compare_baseline._is_baseline_eligible("scheduled_main", False) is False
     assert compare_baseline._is_baseline_eligible("pr", True) is False
     assert compare_baseline._is_baseline_eligible("local", True) is False
 
+
+def test_latency_regression_requires_percent_and_absolute_floors():
+    baseline = [{"latency": 10.0}]
+    current = {"model_id": "wan", "latency": 10.6}
+
+    percent_only = resolve_metric_policies({
+        "latency": {
+            "threshold_percent": 0.05,
+            "threshold_absolute": 0.75,
+        }
+    })
+    absolute_only = resolve_metric_policies({
+        "latency": {
+            "threshold_percent": 0.10,
+            "threshold_absolute": 0.5,
+        }
+    })
+    both = resolve_metric_policies({
+        "latency": {
+            "threshold_percent": 0.05,
+            "threshold_absolute": 0.5,
+        }
+    })
+
+    assert compare_baseline._check_regressions(current, baseline, percent_only) == []
+    assert compare_baseline._check_regressions(current, baseline, absolute_only) == []
+
+    failures = compare_baseline._check_regressions(current, baseline, both)
+    assert len(failures) == 1
+    assert "latency regressed by 6.0% and 0.600" in failures[0]
+
+
+def test_throughput_regression_uses_higher_is_better_direction():
+    baseline = [{"throughput": 10.0}]
+    current = {"model_id": "wan", "throughput": 9.0}
+    policies = resolve_metric_policies({
+        "throughput": {
+            "threshold_percent": 0.05,
+            "threshold_absolute": 0.5,
+        }
+    })
+
+    failures = compare_baseline._check_regressions(current, baseline, policies)
+
+    assert len(failures) == 1
+    assert "throughput regressed by 10.0% and 1.000" in failures[0]
+
+
+def test_memory_regression_uses_metric_specific_absolute_floor():
+    baseline = [{"memory": 10000.0}]
+    current = {"model_id": "wan", "memory": 10600.0}
+    policies = resolve_metric_policies({
+        "memory": {
+            "threshold_percent": 0.05,
+            "threshold_absolute": 256.0,
+        }
+    })
+
+    failures = compare_baseline._check_regressions(current, baseline, policies)
+
+    assert len(failures) == 1
+    assert "memory regressed by 6.0% and 600.0" in failures[0]
+
+
+def test_component_metric_can_gate_independently():
+    baseline = [{"dit_time_s": 8.0}]
+    current = {"model_id": "wan", "dit_time_s": 8.6}
+    policies = resolve_metric_policies({
+        "dit_time_s": {
+            "threshold_percent": 0.05,
+            "threshold_absolute": 0.25,
+        }
+    })
+
+    failures = compare_baseline._check_regressions(current, baseline, policies)
+
+    assert len(failures) == 1
+    assert "dit_time_s regressed by 7.5% and 0.600" in failures[0]
+
+
+def test_informational_metric_remains_visible_without_failing():
+    baseline = [{"throughput": 10.0}]
+    current = {"model_id": "wan", "gpu_type": "NVIDIA L40S", "throughput": 8.0}
+    policies = resolve_metric_policies({
+        "throughput": {
+            "threshold_percent": 0.01,
+            "threshold_absolute": 0.01,
+            "gated": False,
+        }
+    })
+
+    row = compare_baseline._build_summary_row(current, baseline, policies, False)
+
+    assert compare_baseline._check_regressions(current, baseline, policies) == []
+    assert row["metrics"]["throughput"]["regression_pct"] == 20.0
+    assert row["metrics"]["throughput"]["gated"] is False
+    assert row["metrics"]["throughput"]["threshold_exceeded"] is True
+    assert row["metrics"]["throughput"]["regressed"] is False
+    assert row["threshold_exceeded_metrics"] == ["throughput"]
+    assert row["failing_metrics"] == []

fastvideo/tests/performance/test_dashboard_api.py

@@ -56,8 +56,34 @@ def _record(model_id, gpu_type, ts, commit, latency, throughput, success=True, *
 
 def test_summary_endpoint_returns_latest_group_status():
     app = create_app(FakeStore([
-        _record("wan", "NVIDIA L40S", "2026-01-01T00:00:00+00:00", "a" * 40, 10.0, 10.0),
-        _record("wan", "NVIDIA L40S", "2026-01-02T00:00:00+00:00", "b" * 40, 11.0, 9.0),
+        _record(
+            "wan",
+            "NVIDIA L40S",
+            "2026-01-01T00:00:00+00:00",
+            "a" * 40,
+            10.0,
+            10.0,
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            recipe_fingerprint="recipe-a",
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+        ),
+        _record(
+            "wan",
+            "NVIDIA L40S",
+            "2026-01-02T00:00:00+00:00",
+            "b" * 40,
+            11.0,
+            9.0,
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            recipe_fingerprint="recipe-a",
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+        ),
     ]))
     client = TestClient(app)
 
@@ -68,7 +94,13 @@ def test_summary_endpoint_returns_latest_group_status():
     assert body["count"] == 1
     assert body["status_counts"] == {"pass": 1, "fail": 0}
     assert body["rows"][0]["metrics"]["latency"]["baseline"] == 10.0
+    assert body["rows"][0]["metrics"]["latency"]["threshold_exceeded"] is True
+    assert body["rows"][0]["threshold_exceeded_metrics"] == ["latency", "throughput"]
     assert body["rows"][0]["computed_regression_status"] == "fail"
+    assert body["rows"][0]["workload_id"] == "wan-t2v"
+    assert body["rows"][0]["variant_id"] == "1.3b-sp2"
+    assert body["rows"][0]["benchmark_version"] == 2
+    assert body["rows"][0]["recipe_fingerprint"] == "recipe-a"
 
 
 def test_summary_status_is_independent_of_days_window():

fastvideo/tests/performance/test_dashboard_plotly.py

@@ -0,0 +1,98 @@
+# SPDX-License-Identifier: Apache-2.0
+
+import pandas as pd
+
+from fastvideo.tests.performance import dashboard
+
+
+def _record(recipe_fingerprint="recipe-a", software_profile_id="sw-a", **overrides):
+    record = {
+        "model_id": "wan-t2v-1.3b-2gpu",
+        "gpu_type": "NVIDIA L40S",
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+        "recipe_fingerprint": recipe_fingerprint,
+        "hardware_profile_id": "hw-l40s",
+        "software_profile_id": software_profile_id,
+        "timestamp": "2026-01-01T00:00:00+00:00",
+        "commit_sha": "a" * 40,
+        "config_id": "aaaaaaa",
+        "latency": 10.0,
+        "throughput": 4.5,
+        "memory": 10000.0,
+        "text_encoder_time_s": 2.0,
+        "dit_time_s": 8.0,
+        "vae_decode_time_s": 3.0,
+    }
+    record.update(overrides)
+    return record
+
+
+def test_group_data_uses_full_comparison_cohort():
+    df = pd.DataFrame([
+        _record(recipe_fingerprint="recipe-a", software_profile_id="sw-a"),
+        _record(recipe_fingerprint="recipe-b", software_profile_id="sw-b"),
+    ])
+
+    groups = list(dashboard.group_data(df))
+
+    assert len(groups) == 2
+    assert {key[5] for key, _group in groups} == {"recipe-a", "recipe-b"}
+    assert {key[7] for key, _group in groups} == {"sw-a", "sw-b"}
+
+
+def test_group_data_fills_legacy_cohort_columns():
+    df = pd.DataFrame([{
+        "model_id": "legacy-wan",
+        "gpu_type": "NVIDIA L40S",
+        "timestamp": "2026-01-01T00:00:00+00:00",
+        "commit_sha": "a" * 40,
+        "config_id": "aaaaaaa",
+        "latency": 10.0,
+    }])
+
+    groups = list(dashboard.group_data(df))
+
+    assert len(groups) == 1
+    assert groups[0][0][2:] == ("", "", "", "", "", "")
+
+
+def test_build_plots_labels_distinct_cohorts():
+    df = pd.DataFrame([
+        _record(recipe_fingerprint="recipe-a", software_profile_id="sw-a"),
+        _record(recipe_fingerprint="recipe-b", software_profile_id="sw-b"),
+    ])
+
+    figs, skipped_metrics = dashboard.build_plots(df)
+    titles = [fig.layout.title.text for fig in figs]
+
+    assert len(figs) == len(dashboard.METRICS) * 2
+    assert skipped_metrics == []
+    assert any("recipe recipe-a | hw-l40s | sw-a" in title for title in titles)
+    assert any("recipe recipe-b | hw-l40s | sw-b" in title for title in titles)
+
+
+def test_skipped_metric_table_includes_cohort_identity():
+    df = pd.DataFrame([{
+        "model_id": "wan-t2v-1.3b-2gpu",
+        "gpu_type": "NVIDIA L40S",
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+        "recipe_fingerprint": "recipe-a",
+        "hardware_profile_id": "hw-l40s",
+        "software_profile_id": "sw-a",
+        "timestamp": "2026-01-01T00:00:00+00:00",
+        "commit_sha": "a" * 40,
+        "config_id": "aaaaaaa",
+        "latency": 10.0,
+    }])
+
+    _figs, skipped_metrics = dashboard.build_plots(df)
+    html = dashboard.render_skipped_metrics(skipped_metrics[:1])
+
+    assert skipped_metrics[0]["cohort"] == "wan-t2v / 1.3b-sp2 / v2"
+    assert skipped_metrics[0]["cohort_detail"] == "recipe recipe-a | hw-l40s | sw-a"
+    assert "wan-t2v / 1.3b-sp2 / v2" in html
+    assert "recipe recipe-a | hw-l40s | sw-a" in html

fastvideo/tests/performance/test_dashboard_service.py

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: Apache-2.0
+from fastvideo.performance import hf_store
 from fastvideo.performance_dashboard.service import build_latest_summary, build_trends, filter_records
-from fastvideo.tests.performance import hf_store
 
 
 def _record(ts, commit, latency, throughput, success=True, **metadata):
@@ -28,16 +28,23 @@ def test_build_latest_summary_uses_previous_successful_records_for_baseline():
         _record("2026-01-03T00:00:00+00:00", "c" * 40, 11.0, 9.0),
     ]
 
-    rows = build_latest_summary(records, max_regression=0.05)
+    rows = build_latest_summary(records)
 
     assert len(rows) == 1
     row = rows[0]
     assert row["baseline_n"] == 1
     assert row["metrics"]["latency"]["baseline"] == 10.0
     assert row["metrics"]["latency"]["regression_pct"] == 10.0
+    assert row["metrics"]["latency"]["absolute_delta"] == 1.0
+    assert row["metrics"]["latency"]["threshold_percent"] == 8.0
+    assert row["metrics"]["latency"]["threshold_absolute"] == 0.5
+    assert row["metrics"]["latency"]["threshold_exceeded"] is True
+    assert row["metrics"]["latency"]["regressed"] is True
     assert row["metrics"]["throughput"]["regression_pct"] == 10.0
     assert row["status"] == "pass"
     assert row["computed_regression_status"] == "fail"
+    assert row["threshold_exceeded_metrics"] == ["latency", "throughput"]
+    assert row["failing_metrics"] == ["latency", "throughput"]
 
 
 def test_build_latest_summary_status_uses_latest_record_success_field():
@@ -73,7 +80,7 @@ def test_build_latest_summary_run_source_filter_keeps_canonical_baseline():
         ),
     ]
 
-    rows = build_latest_summary(records, max_regression=0.05, run_source="pr")
+    rows = build_latest_summary(records, run_source="pr")
 
     assert len(rows) == 1
     assert rows[0]["run_source"] == "pr"
@@ -83,6 +90,165 @@ def test_build_latest_summary_run_source_filter_keeps_canonical_baseline():
     assert rows[0]["computed_regression_status"] == "fail"
 
 
+def test_build_latest_summary_requires_absolute_floor_for_computed_regression():
+    records = [
+        _record("2026-01-01T00:00:00+00:00", "a" * 40, 10.0, 10.0),
+        _record(
+            "2026-01-02T00:00:00+00:00",
+            "b" * 40,
+            10.6,
+            10.0,
+            regression_thresholds={
+                "latency": {
+                    "threshold_percent": 0.05,
+                    "threshold_absolute": 0.75,
+                    "gated": True,
+                }
+            },
+        ),
+    ]
+
+    rows = build_latest_summary(records)
+
+    assert round(rows[0]["metrics"]["latency"]["regression_pct"], 1) == 6.0
+    assert round(rows[0]["metrics"]["latency"]["absolute_delta"], 3) == 0.6
+    assert rows[0]["metrics"]["latency"]["threshold_exceeded"] is False
+    assert rows[0]["metrics"]["latency"]["regressed"] is False
+    assert rows[0]["computed_regression_status"] == "pass"
+
+
+def test_build_latest_summary_separates_informational_threshold_crossing():
+    records = [
+        _record("2026-01-01T00:00:00+00:00", "a" * 40, 10.0, 10.0),
+        _record(
+            "2026-01-02T00:00:00+00:00",
+            "b" * 40,
+            10.6,
+            10.0,
+            regression_thresholds={
+                "latency": {
+                    "threshold_percent": 0.05,
+                    "threshold_absolute": 0.5,
+                    "gated": False,
+                }
+            },
+        ),
+    ]
+
+    rows = build_latest_summary(records)
+
+    assert rows[0]["metrics"]["latency"]["threshold_exceeded"] is True
+    assert rows[0]["metrics"]["latency"]["regressed"] is False
+    assert rows[0]["threshold_exceeded_metrics"] == ["latency"]
+    assert rows[0]["failing_metrics"] == []
+    assert rows[0]["computed_regression_status"] == "pass"
+def test_build_latest_summary_keeps_identity_cohorts_separate():
+    records = [
+        _record(
+            "2026-01-01T00:00:00+00:00",
+            "a" * 40,
+            10.0,
+            10.0,
+            recipe_fingerprint="recipe-a",
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+            run_source="scheduled_main",
+            baseline_eligible=True,
+        ),
+        _record(
+            "2026-01-02T00:00:00+00:00",
+            "b" * 40,
+            20.0,
+            5.0,
+            recipe_fingerprint="recipe-b",
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+            run_source="scheduled_main",
+            baseline_eligible=True,
+        ),
+        _record(
+            "2026-01-03T00:00:00+00:00",
+            "c" * 40,
+            22.0,
+            4.5,
+            recipe_fingerprint="recipe-b",
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+        ),
+    ]
+
+    rows = build_latest_summary(records, max_regression=0.05)
+    recipe_b_row = next(row for row in rows if row["recipe_fingerprint"] == "recipe-b")
+
+    assert len(rows) == 2
+    assert recipe_b_row["baseline_n"] == 1
+    assert recipe_b_row["metrics"]["latency"]["baseline"] == 20.0
+    assert recipe_b_row["workload_id"] == "wan-t2v"
+    assert recipe_b_row["variant_id"] == "1.3b-sp2"
+    assert recipe_b_row["benchmark_version"] == 2
+
+
+def test_build_latest_summary_keeps_variant_versions_separate():
+    records = [
+        _record(
+            "2026-01-01T00:00:00+00:00",
+            "a" * 40,
+            10.0,
+            10.0,
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=1,
+            recipe_fingerprint="recipe-a",
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+            run_source="scheduled_main",
+            baseline_eligible=True,
+        ),
+        _record(
+            "2026-01-02T00:00:00+00:00",
+            "b" * 40,
+            20.0,
+            5.0,
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            recipe_fingerprint="recipe-a",
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+            run_source="scheduled_main",
+            baseline_eligible=True,
+        ),
+        _record(
+            "2026-01-03T00:00:00+00:00",
+            "c" * 40,
+            22.0,
+            4.5,
+            workload_id="wan-t2v",
+            variant_id="1.3b-sp2",
+            benchmark_version=2,
+            recipe_fingerprint="recipe-a",
+            hardware_profile_id="hw-l40s",
+            software_profile_id="sw-cu130",
+        ),
+    ]
+
+    rows = build_latest_summary(records, max_regression=0.05)
+    version_2_row = next(row for row in rows if row["benchmark_version"] == 2)
+
+    assert len(rows) == 2
+    assert version_2_row["baseline_n"] == 1
+    assert version_2_row["metrics"]["latency"]["baseline"] == 20.0
+
+
 def test_filter_records_and_trends_preserve_metric_points():
     records = [
         _record("2026-01-01T00:00:00+00:00", "a" * 40, 10.0, 10.0),
@@ -157,3 +323,78 @@ def test_load_records_can_filter_baseline_eligible_records(tmp_path):
         "2026-01-02T00:00:00+00:00",
         "2026-01-03T00:00:00+00:00",
     }
+
+
+def test_load_records_for_model_filters_identity_cohort(tmp_path):
+    model_dir = tmp_path / "wan"
+    model_dir.mkdir()
+    (model_dir / "matching.json").write_text(
+        """
+        {
+          "model_id": "wan",
+          "gpu_type": "NVIDIA L40S",
+          "timestamp": "2026-01-01T00:00:00+00:00",
+          "success": true,
+          "baseline_eligible": true,
+          "workload_id": "wan-t2v",
+          "variant_id": "1.3b-sp2",
+          "benchmark_version": 2,
+          "recipe_fingerprint": "recipe-a",
+          "hardware_profile_id": "hw-l40s",
+          "software_profile_id": "sw-cu130"
+        }
+        """,
+        encoding="utf-8",
+    )
+    (model_dir / "other_recipe.json").write_text(
+        """
+        {
+          "model_id": "wan",
+          "gpu_type": "NVIDIA L40S",
+          "timestamp": "2026-01-02T00:00:00+00:00",
+          "success": true,
+          "baseline_eligible": true,
+          "workload_id": "wan-t2v",
+          "variant_id": "1.3b-sp2",
+          "benchmark_version": 2,
+          "recipe_fingerprint": "recipe-b",
+          "hardware_profile_id": "hw-l40s",
+          "software_profile_id": "sw-cu130"
+        }
+        """,
+        encoding="utf-8",
+    )
+    (model_dir / "other_version.json").write_text(
+        """
+        {
+          "model_id": "wan",
+          "gpu_type": "NVIDIA L40S",
+          "timestamp": "2026-01-03T00:00:00+00:00",
+          "success": true,
+          "baseline_eligible": true,
+          "workload_id": "wan-t2v",
+          "variant_id": "1.3b-sp2",
+          "benchmark_version": 3,
+          "recipe_fingerprint": "recipe-a",
+          "hardware_profile_id": "hw-l40s",
+          "software_profile_id": "sw-cu130"
+        }
+        """,
+        encoding="utf-8",
+    )
+
+    records = hf_store.load_records_for_model(
+        str(tmp_path),
+        "wan",
+        "NVIDIA L40S",
+        workload_id="wan-t2v",
+        variant_id="1.3b-sp2",
+        benchmark_version="2",
+        recipe_fingerprint="recipe-a",
+        hardware_profile_id="hw-l40s",
+        software_profile_id="sw-cu130",
+        baseline_eligible_only=True,
+    )
+
+    assert len(records) == 1
+    assert records[0]["recipe_fingerprint"] == "recipe-a"

fastvideo/tests/performance/test_identity.py

@@ -0,0 +1,422 @@
+# SPDX-License-Identifier: Apache-2.0
+
+from copy import deepcopy
+
+import pytest
+import torch
+
+from fastvideo.tests.performance import identity as identity_module
+from fastvideo.tests.performance.identity import (
+    build_recipe_from_benchmark_config,
+    canonical_json,
+    environment_fingerprint,
+    environment_metadata,
+    hardware_profile,
+    hardware_profile_id,
+    recipe_fingerprint,
+    resolved_revision_from_model_path,
+    software_profile,
+    software_profile_id,
+)
+
+
+def _benchmark_config():
+    return {
+        "benchmark_id": "wan-t2v-1.3b-2gpu",
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+        "model": {
+            "model_path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
+            "model_short_name": "Wan2.1-T2V-1.3B",
+        },
+        "init_kwargs": {
+            "num_gpus": 2,
+            "sp_size": 2,
+            "tp_size": 1,
+            "vae_sp": True,
+            "vae_tiling": True,
+            "text_encoder_precisions": ["fp32"],
+        },
+        "generation_kwargs": {
+            "height": 480,
+            "width": 832,
+            "num_frames": 45,
+            "num_inference_steps": 4,
+            "guidance_scale": 3,
+            "embedded_cfg_scale": 6,
+            "seed": 1024,
+            "fps": 24,
+            "neg_prompt": "low quality",
+        },
+        "test_prompts": ["A cinematic video."],
+    }
+
+
+def _fingerprint(cfg, attention_backend="FLASH_ATTN"):
+    recipe = build_recipe_from_benchmark_config(cfg, attention_backend=attention_backend)
+    return recipe_fingerprint(recipe)
+
+
+def test_canonical_json_is_stable_for_mapping_order():
+    left = {"b": [2, 1], "a": {"z": "last", "m": "middle"}}
+    right = {"a": {"m": "middle", "z": "last"}, "b": [2, 1]}
+
+    assert canonical_json(left) == canonical_json(right)
+
+
+def test_canonical_json_handles_sets_deterministically():
+    assert canonical_json({"values": {"b", "a"}}) == '{"values":["a","b"]}'
+    assert canonical_json({"values": frozenset(("b", "a"))}) == '{"values":["a","b"]}'
+
+
+def test_same_config_produces_same_recipe_fingerprint():
+    cfg = _benchmark_config()
+
+    assert _fingerprint(cfg) == _fingerprint(deepcopy(cfg))
+
+
+def test_recipe_includes_first_class_benchmark_identity():
+    recipe = build_recipe_from_benchmark_config(_benchmark_config())
+
+    assert recipe["benchmark"] == {
+        "benchmark_id": "wan-t2v-1.3b-2gpu",
+        "workload_id": "wan-t2v",
+        "variant_id": "1.3b-sp2",
+        "benchmark_version": 2,
+    }
+
+
+def test_recipe_requires_first_class_benchmark_identity():
+    cfg = _benchmark_config()
+    del cfg["workload_id"]
+
+    with pytest.raises(ValueError, match="workload_id"):
+        build_recipe_from_benchmark_config(cfg)
+
+
+def test_semantically_equivalent_sequence_forms_match():
+    cfg = _benchmark_config()
+    equivalent = deepcopy(cfg)
+    equivalent["init_kwargs"]["text_encoder_precisions"] = tuple(
+        equivalent["init_kwargs"]["text_encoder_precisions"])
+
+    assert _fingerprint(cfg) == _fingerprint(equivalent)
+
+
+def test_output_path_does_not_change_recipe_fingerprint():
+    cfg = _benchmark_config()
+    with_output_path = deepcopy(cfg)
+    with_output_path["generation_kwargs"]["output_path"] = "/tmp/generated"
+
+    assert _fingerprint(cfg) == _fingerprint(with_output_path)
+
+
+def test_runtime_resolved_values_do_not_change_recipe_fingerprint():
+    # Runtime-resolved values are audit metadata: an upstream repo commit
+    # (even a README-only edit) or a silent attention-backend fallback must
+    # not churn the cohort — that would open a fresh ungated cohort and let
+    # degraded numbers seed a new baseline. Declared inputs (including
+    # requested_backend) stay in the hash.
+    cfg = _benchmark_config()
+    recipe = build_recipe_from_benchmark_config(
+        cfg, attention_backend="FLASH_ATTN",
+        resolved_attention_backend="FLASH_ATTN",
+        resolved_model_revision="aaaa1111")
+    changed = build_recipe_from_benchmark_config(
+        cfg, attention_backend="FLASH_ATTN",
+        resolved_attention_backend="TORCH_SDPA",
+        resolved_model_revision="bbbb2222")
+
+    assert recipe["model"]["resolved_revision"] == "aaaa1111"  # kept for audit
+    assert recipe["attention"]["resolved_backend"] == "FLASH_ATTN"  # kept for audit
+    assert recipe_fingerprint(recipe) == recipe_fingerprint(changed)
+
+
+def test_measured_prompt_override_ignores_unused_configured_prompts():
+    cfg = _benchmark_config()
+    changed = deepcopy(cfg)
+    changed["test_prompts"] = [
+        "Unused prompt that should not affect this measured workload.",
+        cfg["test_prompts"][0],
+    ]
+
+    recipe = build_recipe_from_benchmark_config(cfg, measured_prompts=[cfg["test_prompts"][0]])
+    changed_recipe = build_recipe_from_benchmark_config(changed, measured_prompts=[cfg["test_prompts"][0]])
+
+    assert recipe_fingerprint(recipe) == recipe_fingerprint(changed_recipe)
+
+
+def test_num_inference_steps_changes_recipe_fingerprint():
+    cfg = _benchmark_config()
+    changed = deepcopy(cfg)
+    changed["generation_kwargs"]["num_inference_steps"] = 8
+
+    assert _fingerprint(cfg) != _fingerprint(changed)
+
+
+def test_benchmark_version_changes_recipe_fingerprint():
+    cfg = _benchmark_config()
+    changed = deepcopy(cfg)
+    changed["benchmark_version"] = 3
+
+    assert _fingerprint(cfg) != _fingerprint(changed)
+
+
+def test_attention_backend_changes_recipe_fingerprint():
+    cfg = _benchmark_config()
+
+    assert _fingerprint(cfg, attention_backend="FLASH_ATTN") != _fingerprint(
+        cfg, attention_backend="TORCH_SDPA")
+
+
+def test_precision_changes_recipe_fingerprint():
+    cfg = _benchmark_config()
+    changed = deepcopy(cfg)
+    changed["init_kwargs"]["text_encoder_precisions"] = ["bf16"]
+
+    assert _fingerprint(cfg) != _fingerprint(changed)
+
+
+def test_dimensions_and_seed_change_recipe_fingerprint():
+    cfg = _benchmark_config()
+    different_dimensions = deepcopy(cfg)
+    different_dimensions["generation_kwargs"]["width"] = 1024
+    different_seed = deepcopy(cfg)
+    different_seed["generation_kwargs"]["seed"] = 2048
+
+    assert _fingerprint(cfg) != _fingerprint(different_dimensions)
+    assert _fingerprint(cfg) != _fingerprint(different_seed)
+
+
+def test_distributed_layout_changes_recipe_fingerprint():
+    cfg = _benchmark_config()
+    changed = deepcopy(cfg)
+    changed["init_kwargs"]["sp_size"] = 1
+    changed["init_kwargs"]["tp_size"] = 2
+
+    assert _fingerprint(cfg) != _fingerprint(changed)
+
+
+def test_software_profile_uses_exact_attention_kernel_package_versions_for_id():
+    profile_a = software_profile(
+        python_version="3.12.4",
+        torch_version="2.12.0+cu130",
+        cuda_version="13.0",
+        package_versions={
+            "triton": "3.4.1",
+            "fastvideo_kernel": "0.3.2",
+            "flash_attn_4": "4.0.0.dev0",
+            "flashinfer": "0.2.11",
+            "nvidia_cutlass_dsl": "4.5.0",
+        },
+    )
+    profile_b = software_profile(
+        python_version="3.12.5",
+        torch_version="2.12.1+cu130",
+        cuda_version="13.0",
+        package_versions={
+            "triton": "3.4.9",
+            "fastvideo_kernel": "0.3.7",
+            "flash_attn_4": "4.0.0.dev1",
+            "flashinfer": "0.2.12",
+            "nvidia_cutlass_dsl": "4.5.1",
+        },
+    )
+
+    assert profile_a == {
+        "python": "3.12",
+        "pytorch": "2.12",
+        "cuda": "13.0",
+        "packages": {
+            "fastvideo_kernel": "0.3.2",
+            "flash_attn_4": "4.0.0.dev0",
+            "flashinfer": "0.2.11",
+            "nvidia_cutlass_dsl": "4.5.0",
+            "triton": "3.4.1",
+        },
+    }
+    assert profile_b["python"] == "3.12"
+    assert profile_b["pytorch"] == "2.12"
+    assert profile_b["cuda"] == "13.0"
+    assert software_profile_id(profile_a) != software_profile_id(profile_b)
+
+
+def test_installed_package_versions_checks_attention_kernel_distributions(monkeypatch):
+    versions = {
+        "fastvideo-kernel": "0.3.2",
+        "flash-attn": "2.8.1",
+        "flash-attn-4": "4.0.0.dev0",
+        "flash-attention-fp4": "0.1.0",
+        "flashinfer-python": "0.2.11",
+        "nvidia-cutlass-dsl": "4.5.0",
+    }
+
+    def fake_version(name):
+        if name not in versions:
+            raise identity_module.importlib.metadata.PackageNotFoundError(name)
+        return versions[name]
+
+    monkeypatch.setattr(identity_module.importlib.metadata, "version", fake_version)
+
+    installed = identity_module._installed_package_versions()
+
+    assert installed["fastvideo_kernel"] == "0.3.2"
+    assert installed["flash_attn"] == "2.8.1"
+    assert installed["flash_attn_4"] == "4.0.0.dev0"
+    assert installed["flash_attention_fp4"] == "0.1.0"
+    assert installed["flashinfer"] == "0.2.11"
+    assert installed["nvidia_cutlass_dsl"] == "4.5.0"
+
+
+def test_hardware_profile_id_uses_normalized_gpu_cohort():
+    profile = hardware_profile(
+        num_gpus=2,
+        gpu_devices=[
+            {
+                "name": "NVIDIA L40S",
+                "memory_bytes": 48 * 1024**3,
+                "compute_capability": "8.9",
+            },
+            {
+                "name": "NVIDIA L40S",
+                "memory_gb": 48,
+                "compute_capability": "8.9",
+            },
+        ],
+        interconnect="none_or_partial",
+    )
+
+    assert profile == {
+        "device_type": "cuda",
+        "gpu_count": 2,
+        "gpus": [
+            {
+                "name": "NVIDIA L40S",
+                "memory_gb": 48,
+                "compute_capability": "8.9",
+            },
+            {
+                "name": "NVIDIA L40S",
+                "memory_gb": 48,
+                "compute_capability": "8.9",
+            },
+        ],
+        "interconnect": "none_or_partial",
+    }
+    assert hardware_profile_id(profile).startswith("hw-")
+
+
+def test_hardware_profile_pads_missing_requested_cuda_devices(monkeypatch):
+    class Props:
+        name = "NVIDIA L40S"
+        total_memory = 48 * 1024**3
+
+    monkeypatch.setattr(torch.cuda, "is_available", lambda: True)
+    monkeypatch.setattr(torch.cuda, "device_count", lambda: 1)
+    monkeypatch.setattr(torch.cuda, "get_device_properties", lambda device_id: Props())
+    monkeypatch.setattr(torch.cuda, "get_device_capability", lambda device_id: (8, 9))
+
+    profile = hardware_profile(num_gpus=2, interconnect="unknown")
+
+    assert profile["gpu_count"] == 2
+    assert profile["gpus"] == [
+        {
+            "name": "NVIDIA L40S",
+            "memory_gb": 48,
+            "compute_capability": "8.9",
+        },
+        {
+            "name": "unknown",
+            "memory_gb": None,
+            "compute_capability": None,
+        },
+    ]
+
+
+def test_local_path_detection_checks_both_path_separators(monkeypatch):
+    monkeypatch.setattr(identity_module.os, "sep", "\\")
+
+    assert identity_module._looks_like_local_path("models/local/checkpoint") is True
+    assert identity_module._looks_like_local_path(r"C:\models\checkpoint") is True
+    assert identity_module._looks_like_local_path("Wan-AI/Wan2.1-T2V-1.3B-Diffusers") is False
+
+
+def test_environment_metadata_is_separate_audit_fingerprint():
+    cfg = _benchmark_config()
+    recipe_hash = _fingerprint(cfg)
+    audit_a = environment_metadata(
+        env={"IMAGE_VERSION": "py3.12-cuda13.0.0"},
+        package_versions={"triton": "3.4.1"},
+        hardware={"device_type": "cuda", "gpu_count": 2},
+        software={"python": "3.12", "pytorch": "2.12", "cuda": "13.0"},
+    )
+    audit_b = environment_metadata(
+        env={"IMAGE_VERSION": "py3.12-cuda13.0.1"},
+        package_versions={"triton": "3.4.1"},
+        hardware={"device_type": "cuda", "gpu_count": 2},
+        software={"python": "3.12", "pytorch": "2.12", "cuda": "13.0"},
+    )
+
+    assert recipe_hash == _fingerprint(cfg)
+    assert environment_fingerprint(audit_a) != environment_fingerprint(audit_b)
+
+
+def test_resolved_revision_from_hf_snapshot_path():
+    revision = "a" * 40
+
+    assert resolved_revision_from_model_path(
+        f"/root/.cache/huggingface/hub/models--org--repo/snapshots/{revision}") == revision
+    assert resolved_revision_from_model_path("/models/local-repo") is None
+
+
+def test_runtime_identity_from_generator_summarizes_worker_records():
+    from fastvideo.tests.performance.test_inference_performance import _runtime_identity_from_generator
+
+    revision = "b" * 40
+
+    class FakeExecutor:
+        def collective_rpc(self, _method):
+            return [
+                {
+                    "resolved_attention_backends": ["FLASH_ATTN"],
+                    "resolved_model_path": f"/root/.cache/huggingface/hub/models--org--repo/snapshots/{revision}",
+                },
+                {
+                    "resolved_attention_backends": ["FLASH_ATTN"],
+                    "resolved_model_path": f"/root/.cache/huggingface/hub/models--org--repo/snapshots/{revision}",
+                },
+            ]
+
+    class FakeGenerator:
+        executor = FakeExecutor()
+
+    assert _runtime_identity_from_generator(FakeGenerator()) == {
+        "resolved_attention_backend": "FLASH_ATTN",
+        "resolved_model_revision": revision,
+    }
+
+
+def test_collect_worker_identity_handles_torch_module_pipeline():
+    from fastvideo.tests.performance.test_inference_performance import _collect_worker_identity
+
+    class Backend:
+        name = "FLASH_ATTN"
+
+    class Leaf(torch.nn.Module):
+        backend = Backend()
+
+    class Pipeline(torch.nn.Module):
+        model_path = "/models/local"
+
+        def __init__(self):
+            super().__init__()
+            self.leaf = Leaf()
+
+    class Worker:
+        pipeline = Pipeline()
+
+    assert _collect_worker_identity(Worker()) == {
+        "resolved_attention_backends": ["FLASH_ATTN"],
+        "resolved_model_path": "/models/local",
+    }

fastvideo/tests/performance/test_inference_performance.py

@@ -12,12 +12,25 @@ import os
 import time
 from collections.abc import Mapping
 from datetime import datetime, timezone
+from typing import Any
 
 import torch
 import pytest
 
 from fastvideo import VideoGenerator
 from fastvideo.logger import init_logger
+from fastvideo.tests.performance.identity import (
+    benchmark_identity_from_config,
+    build_recipe_from_benchmark_config,
+    environment_fingerprint,
+    environment_metadata,
+    hardware_profile,
+    hardware_profile_id,
+    recipe_fingerprint,
+    resolved_revision_from_model_path,
+    software_profile,
+    software_profile_id,
+)
 from fastvideo.worker.multiproc_executor import MultiprocExecutor
 
 logger = init_logger(__name__)
@@ -28,6 +41,20 @@ STAGE_METRIC_MAP: dict[str, str] = {
     "DmdDenoisingStage": "dit_time_s",
     "DecodingStage": "vae_decode_time_s",
 }
+V2_CONFIG_SCHEMA_VERSION = 2
+V2_REQUIRED_IDENTITY_FIELDS = (
+    "workload_id",
+    "variant_id",
+    "benchmark_version",
+)
+# "recipe" is no longer config-declarable: the fingerprint follow-up landed,
+# and the generated recipe document owns that key in emitted records. A config
+# declaring it now fails validation loudly instead of being silently
+# overwritten by the generated one.
+V2_OPTIONAL_METADATA_FIELDS = (
+    "metric_threshold_policy",
+    "quality_metadata",
+)
 
 # -- Config discovery -------------------------------------------------------
 
@@ -42,6 +69,71 @@ _BENCHMARKS_DIR = os.path.join(
 )
 
 
+def _has_v2_fields(cfg):
+    v2_fields = V2_REQUIRED_IDENTITY_FIELDS + V2_OPTIONAL_METADATA_FIELDS
+    return any(field in cfg for field in v2_fields)
+
+
+def _is_v2_config(cfg):
+    return cfg.get("config_schema_version") == V2_CONFIG_SCHEMA_VERSION
+
+
+def _validate_non_empty_string(value, field, path):
+    if not isinstance(value, str) or not value.strip():
+        raise ValueError(f"{path}: v2 identity field {field!r} must be a non-empty string")
+
+
+def _validate_integer(value, field, path):
+    if isinstance(value, bool) or not isinstance(value, int):
+        raise ValueError(f"{path}: v2 identity field {field!r} must be an integer")
+
+
+def _validate_benchmark_config(cfg, path="<memory>"):
+    missing_common = [field for field in ("benchmark_id",) if field not in cfg]
+    if missing_common:
+        raise ValueError(f"{path}: missing required benchmark config fields: {', '.join(missing_common)}")
+
+    schema_version = cfg.get("config_schema_version")
+    if schema_version is None:
+        if _has_v2_fields(cfg):
+            raise ValueError(f"{path}: v2 benchmark identity fields require config_schema_version=2")
+        return
+
+    if schema_version != V2_CONFIG_SCHEMA_VERSION:
+        raise ValueError(f"{path}: unsupported benchmark config_schema_version={schema_version!r}")
+
+    missing_v2 = [field for field in V2_REQUIRED_IDENTITY_FIELDS if field not in cfg]
+    if missing_v2:
+        raise ValueError(f"{path}: missing required v2 identity fields: {', '.join(missing_v2)}")
+
+    _validate_non_empty_string(cfg["workload_id"], "workload_id", path)
+    _validate_non_empty_string(cfg["variant_id"], "variant_id", path)
+    _validate_integer(cfg["benchmark_version"], "benchmark_version", path)
+
+    for field in V2_OPTIONAL_METADATA_FIELDS:
+        if field in cfg and not isinstance(cfg[field], Mapping):
+            raise ValueError(f"{path}: optional v2 metadata field {field!r} must be an object")
+
+
+def _config_identity_metadata(cfg):
+    if not _is_v2_config(cfg):
+        return {}
+    metadata = {
+        "config_schema_version": cfg["config_schema_version"],
+        "workload_id": cfg["workload_id"],
+        "variant_id": cfg["variant_id"],
+        "benchmark_version": cfg["benchmark_version"],
+    }
+    for field in V2_OPTIONAL_METADATA_FIELDS:
+        if field in cfg:
+            metadata[field] = cfg[field]
+    return metadata
+
+
+def _benchmark_display_id(cfg):
+    return cfg["benchmark_id"]
+
+
 def _discover_benchmarks():
     """Glob benchmark JSON configs and return list of (id, config) tuples."""
     pattern = os.path.join(_BENCHMARKS_DIR, "*.json")
@@ -49,6 +141,7 @@ def _discover_benchmarks():
     for path in sorted(glob.glob(pattern)):
         with open(path) as f:
             cfg = json.load(f)
+        _validate_benchmark_config(cfg, path)
         configs.append(cfg)
     return configs
 
@@ -150,6 +243,110 @@ def _write_results(results):
     logger.info("Performance results written to %s", filepath)
 
 
+def _backend_name(value) -> str:
+    if hasattr(value, "name"):
+        return str(value.name)
+    return str(value)
+
+
+def _collect_worker_identity(worker) -> dict[str, Any]:
+    pipeline = getattr(worker, "pipeline", None)
+    model_path = getattr(pipeline, "model_path", None)
+    backends: set[str] = set()
+
+    modules = getattr(pipeline, "modules", None)
+    if isinstance(modules, Mapping):
+        module_iter = modules.values()
+    elif isinstance(pipeline, torch.nn.Module):
+        module_iter = (pipeline,)
+    else:
+        module_iter = ()
+
+    for module in module_iter:
+        if isinstance(module, torch.nn.Module):
+            for submodule in module.modules():
+                backend = getattr(submodule, "backend", None)
+                if backend is not None:
+                    backends.add(_backend_name(backend))
+        else:
+            backend = getattr(module, "backend", None)
+            if backend is not None:
+                backends.add(_backend_name(backend))
+
+    return {
+        "resolved_attention_backends": sorted(backends),
+        "resolved_model_path": model_path,
+    }
+
+
+def _single_or_list(values: set[str]) -> str | list[str] | None:
+    ordered = sorted(values)
+    if not ordered:
+        return None
+    if len(ordered) == 1:
+        return ordered[0]
+    return ordered
+
+
+def _runtime_identity_from_generator(generator) -> dict[str, Any]:
+    worker_records = generator.executor.collective_rpc(_collect_worker_identity)
+    resolved_backends: set[str] = set()
+    resolved_revisions: set[str] = set()
+
+    for record in worker_records:
+        resolved_backends.update(record.get("resolved_attention_backends") or [])
+        revision = resolved_revision_from_model_path(record.get("resolved_model_path"))
+        if revision is not None:
+            resolved_revisions.add(revision)
+
+    return {
+        "resolved_attention_backend": _single_or_list(resolved_backends),
+        "resolved_model_revision": _single_or_list(resolved_revisions),
+    }
+
+
+def _benchmark_identity_fields(cfg):
+    identity = benchmark_identity_from_config(cfg)
+    return {
+        key: identity[key]
+        for key in ("workload_id", "variant_id", "benchmark_version")
+    }
+
+
+def _build_identity_fields(cfg, init_kwargs, prompt, runtime_identity):
+    # Legacy v1 configs (no config_schema_version) stay on the legacy
+    # (model_id, gpu_type) cohort: they lack the required identity fields,
+    # and building a recipe for them would raise AFTER the GPU measurement
+    # completed. The validator already enforces that v2 configs carry the
+    # identity fields, so v2 records always get the full identity block.
+    if cfg.get("config_schema_version") is None:
+        return {}
+    recipe = build_recipe_from_benchmark_config(
+        cfg,
+        resolved_attention_backend=runtime_identity.get("resolved_attention_backend"),
+        resolved_model_revision=runtime_identity.get("resolved_model_revision"),
+        measured_prompts=[prompt],
+    )
+    num_gpus = init_kwargs.get("num_gpus", cfg.get("run_config", {}).get("required_gpus", 1))
+    hw_profile = hardware_profile(num_gpus=num_gpus)
+    sw_profile = software_profile()
+    env_metadata = environment_metadata(
+        hardware=hw_profile,
+        software=sw_profile,
+    )
+    return {
+        "recipe": recipe,
+        "recipe_fingerprint": recipe_fingerprint(recipe),
+        "hardware_profile": hw_profile,
+        "hardware_profile_id": hardware_profile_id(hw_profile),
+        "software_profile": sw_profile,
+        "software_profile_id": software_profile_id(sw_profile),
+        "environment_metadata": env_metadata,
+        "environment_fingerprint": environment_fingerprint(env_metadata),
+        **_benchmark_identity_fields(cfg),
+    }
+
+
 # -- Test -------------------------------------------------------------------
 
 def _run_benchmark(cfg):
@@ -187,6 +384,7 @@ def _run_benchmark(cfg):
             model_path=model_info["model_path"],
             **init_kwargs,
         )
+        runtime_identity = _runtime_identity_from_generator(generator)
 
         for i in range(num_warmup):
             logger.info("Warmup run %d/%d", i + 1, num_warmup)
@@ -219,6 +417,7 @@ def _run_benchmark(cfg):
 
     results = {
         "benchmark_id": cfg["benchmark_id"],
+        **_config_identity_metadata(cfg),
         "model_short_name": model_info.get("model_short_name", ""),
         "device": device_name,
         "num_gpus": init_kwargs.get("num_gpus", 1),
@@ -231,6 +430,7 @@ def _run_benchmark(cfg):
         "max_peak_memory_mb": round(max_peak_memory, 1),
         "individual_peak_memories_mb": [round(m, 1) for m in peak_memories],
         "thresholds": thresholds,
+        "regression_thresholds": cfg.get("regression_thresholds", {}),
         "commit": os.environ.get("BUILDKITE_COMMIT", ""),
         "pr_number": os.environ.get("BUILDKITE_PULL_REQUEST", ""),
         "timestamp": datetime.now(timezone.utc).isoformat(),
@@ -239,6 +439,7 @@ def _run_benchmark(cfg):
         "dit_time_s": _avg_component(all_component_times, "dit_time_s"),
         "vae_decode_time_s": _avg_component(all_component_times,
                                             "vae_decode_time_s"),
+        **_build_identity_fields(cfg, init_kwargs, prompt, runtime_identity),
     }
 
     logger.info(
@@ -275,7 +476,7 @@ def _run_benchmark(cfg):
 @pytest.mark.parametrize(
     "cfg",
     _BENCHMARK_CONFIGS,
-    ids=[c["benchmark_id"] for c in _BENCHMARK_CONFIGS],
+    ids=[_benchmark_display_id(c) for c in _BENCHMARK_CONFIGS],
 )
 def test_inference_performance(cfg):
     """Measure generation latency, peak GPU memory, and component-level timings

fastvideo/tests/stages/test_text_encoding.py

@@ -21,6 +21,26 @@ class FakeTokenizer:
             "attention_mask": torch.ones(B, seq_len, dtype=torch.long),
         })
 
+
+class FakeChatTokenizer:
+    def __init__(self):
+        self.last_messages = None
+        self.last_kwargs = None
+
+    def apply_chat_template(self, messages, **kwargs):
+        self.last_messages = messages
+        self.last_kwargs = kwargs
+        assert isinstance(messages[0], list)
+        assert messages[0][0]["role"] == "system"
+        assert messages[0][1]["role"] == "user"
+        B = len(messages)
+        seq_len = int(kwargs.get("max_length", 4))
+        return TensorDict({
+            "input_ids": torch.arange(B * seq_len).view(B, seq_len),
+            "attention_mask": torch.ones(B, seq_len, dtype=torch.long),
+        })
+
+
 class FakeTextEncoder(torch.nn.Module):
     def __init__(self, hidden_size=8):
         super().__init__()
@@ -38,6 +58,14 @@ class FakeTextEncoder(torch.nn.Module):
 def id_preprocess(x: str) -> str:
     return x
 
+
+def chat_list_preprocess(x: str):
+    return [
+        {"role": "system", "content": "Describe the video."},
+        {"role": "user", "content": x if x else " "},
+    ]
+
+
 def take_mean_postprocess(outputs: BaseEncoderOutput) -> torch.Tensor:
     # [B, T, H] -> [B, H]
     return outputs.last_hidden_state.mean(dim=1)
@@ -156,3 +184,32 @@ def test_encode_text_does_not_force_hidden_states_for_ltx2_prefix():
     stage.encode_text("a", fastvideo_args, encoder_index=[0])
 
     assert stage.text_encoders[0].last_output_hidden_states is False
+
+
+def test_chat_list_preprocess_output_is_not_stripped():
+    fastvideo_args, hidden = make_args(num_encoders=1, text_len=5, hidden_size=8)
+    encoder_config = fastvideo_args.pipeline_config.text_encoder_configs[0]
+    encoder_config.is_chat_model = True
+    encoder_config.treat_empty_as_dot = True
+    fastvideo_args.pipeline_config.preprocess_text_funcs = (chat_list_preprocess, )
+
+    tokenizer = FakeChatTokenizer()
+    stage = TextEncodingStage(
+        text_encoders=[FakeTextEncoder(hidden_size=hidden)],
+        tokenizers=[tokenizer],
+    )
+
+    embeds, masks = stage.encode_text(
+        "a robotic arm welding a metal structure",
+        fastvideo_args,
+        encoder_index=[0],
+        return_attention_mask=True,
+    )
+
+    assert embeds[0].shape == (1, hidden)
+    assert masks[0].shape == (1, 5)
+    assert tokenizer.last_messages == [[
+        {"role": "system", "content": "Describe the video."},
+        {"role": "user", "content": "a robotic arm welding a metal structure"},
+    ]]
+    assert tokenizer.last_kwargs["return_tensors"] == "pt"

fastvideo/tests/train/methods/test_wan_causal_dfsft.py

@@ -127,4 +127,12 @@ def test_wan_causal_dfsft_single_train_step(
 
     # 5a-ii: device-keyed grad-norm regression on top of the same harness.
     # Skips when the current GPU has no seeded reference.
-    check_grad_norm_regression("test_wan_causal_dfsft", model.transformer)
+    # rtol above the harness default: the causal model compiles flex_attention
+    # with max-autotune (required for Wan 1.3B's head config), and the
+    # timing-based kernel selection is bimodal across L40S containers —
+    # observed 3.2562 vs 3.5860 (10.13% apart) with identical code, straddling
+    # the default 10%. 12% covers both winners; real wiring breakage (dead
+    # grads, scale bugs) still lands far outside it.
+    check_grad_norm_regression("test_wan_causal_dfsft",
+                               model.transformer,
+                               rtol=0.12)