SolitaryThinker
9b8838834c
Wire the capture/replay lifecycle into the driven-loop step boundary — the other half of Path A (hand-fused kernels behind the registry + piecewise cudagraphs, no compiler). Models and tests the correctness-critical control logic; replay re-runs the current step thunk (CPU models the lifecycle, not the GPU speedup). - GraphCapturer on the instance (cross-request cache), wired into RuntimeLoopContext.execute and gated by LoopSpec.graph_capture == "breakable_cudagraph". Capture key = (device, arch, loop, shape_sig, resident-weight-versions, graph_key). Eager-break for non-capturable (SDE) and interceptor-overridden steps. Never executes a stored thunk, so interleaved requests can't smear state. - WorkPlan.capturable / graph_key. wan21 sets capturable=not sde and folds compute dtype (shape_sig.dtype) + CFG branch set + expert + scheduler-precision into the key — closing a cross-precision key-collision corruption path a non-fp32 build would otherwise hit on a real GPU (audit finding). - Version-in-key auto-invalidation + real eviction: set_weights_version evicts the synced component's graphs (duck-typed, so card/ imports no runtime), preventing a GPU graph leak across FlowGRPO's per-iteration syncs. - register_kernel gains the workspace_bytes capture-safety contract (declared in the matrix; cuda cells declare real scratch, numpy reference is 0). - 11 tests: capture-once/replay-many, eager-break (SDE + override), capture ≡ pure-eager bit-identical, recapture on shape + weight-version change with eviction, eager-loop gating, accel-backend capture, and capturer unit tests (key discrimination, eager-break, workspace-collision safety net, eviction). Honestly deferred (bite a real GPU, not the CPU tests): the static-buffer refactor of the step body, admission budgeting of capture cost (GRAPH_CAPTURE), and per-card opt-in beyond wan21 — all documented in cudagraph.py + README. 195 tests pass.
152 lines
8.4 KiB
Python
152 lines
8.4 KiB
Python
"""WanDenoiseLoop — the canonical driven denoise loop (design_v3 §5, §15a).
|
|
|
|
Bidirectional video diffusion: N flow-match steps over the full latent. Composed from policies
|
|
(CFG / flow-shift / precision / expert routing) — the default decomposition. ``next`` is
|
|
kernel-free (it builds the forward+combine+solver thunk); ``advance`` folds the result and,
|
|
under the ROLLOUT profile, captures a behavior slice — so the *same* loop serves and rolls out
|
|
for RL (design_v3 §10). All per-request state lives in ``LoopState`` (interleave-safe).
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
import numpy as np
|
|
|
|
from ..._enums import ExecutionProfile, WorkUnitKind
|
|
from ...loop.contracts import (
|
|
Done,
|
|
LoopResult,
|
|
LoopState,
|
|
ResourceRequest,
|
|
ShapeSignature,
|
|
StepContext,
|
|
StepResult,
|
|
WorkPlan,
|
|
)
|
|
from ...platform import FLOW_MATCH_STEP, FLOW_SDE_STEP
|
|
from ...request.streams import StreamChunk
|
|
from ..backend import LATENT_CHANNELS
|
|
|
|
|
|
def latent_shape(req) -> tuple[int, int, int, int]:
|
|
"""Map request geometry to a small (deterministic) latent for the CPU toy backend."""
|
|
d = req.diffusion
|
|
t = max(1, d.num_frames // 40)
|
|
h = max(2, d.height // 120)
|
|
w = max(2, d.width // 120)
|
|
return (LATENT_CHANNELS, t, h, w)
|
|
|
|
|
|
class WanDenoiseLoop:
|
|
def __init__(self, *, loop_id, cfg, flow_shift, precision, expert, cost):
|
|
self.loop_id = loop_id
|
|
self.cfg = cfg
|
|
self.flow_shift = flow_shift
|
|
self.precision = precision
|
|
self.expert = expert
|
|
self.cost = cost
|
|
|
|
def init(self, req, model, ctx) -> LoopState:
|
|
seed = req.diffusion.seed if req.diffusion.seed is not None else 0
|
|
rng = np.random.default_rng(seed)
|
|
sig = self.flow_shift.build_schedule(req.diffusion.num_steps, req.diffusion.height,
|
|
req.diffusion.width, sigmas=req.diffusion.sigmas or None)
|
|
shape = latent_shape(req)
|
|
x = (rng.standard_normal(shape) * float(sig[0])).astype("float32")
|
|
st = LoopState(loop_id=self.loop_id, instance_id=model.card.model_id,
|
|
request_id=req.request_id, profile=ctx.profile, rng=rng, seed=seed,
|
|
latents={"video": x}, sigmas=[float(s) for s in sig],
|
|
timesteps=[float(s) * 1000.0 for s in sig])
|
|
st.cond["prompt_embeds"] = ctx.slots.get("text_embeds")
|
|
st.cond["negative_prompt_embeds"] = ctx.slots.get("neg_text_embeds")
|
|
st.scratch["guidance_scale"] = float(req.diffusion.guidance_scale)
|
|
st.scratch["stream_video"] = bool(req.outputs.stream.get("video"))
|
|
# FlowGRPO RL rollout: switch the sampler to SDE-with-logprob (else deterministic ODE serve).
|
|
st.scratch["sde"] = bool(getattr(req.diffusion, "sde_rollout", False))
|
|
st.scratch["sde_noise_scale"] = float(getattr(req.diffusion, "sde_noise_scale", 0.7))
|
|
st.plugin_state["cfg"] = {}
|
|
return st
|
|
|
|
def next(self, st: LoopState):
|
|
i = st.step_idx
|
|
if i >= len(st.sigmas) - 1:
|
|
return Done()
|
|
sigma_t, sigma_next = st.sigmas[i], st.sigmas[i + 1]
|
|
expert_id = self.expert.expert_for(StepContext(i, st.timesteps[i], sigma_t))
|
|
sctx = StepContext(step_idx=i, timestep=st.timesteps[i], sigma=sigma_t, active_expert_id=expert_id)
|
|
cfg_state = st.plugin_state["cfg"]
|
|
branches = self.cfg.branches_this_step(sctx, cfg_state)
|
|
x = st.latents["video"]
|
|
pe, ne = st.cond["prompt_embeds"], st.cond["negative_prompt_embeds"]
|
|
scale = st.scratch["guidance_scale"]
|
|
cfg, precision = self.cfg, self.precision
|
|
sde, noise_scale, rng = st.scratch.get("sde", False), st.scratch.get("sde_noise_scale", 0.7), st.rng
|
|
|
|
def run(model, override=None):
|
|
# Solver primitives are dispatched through the platform's kernel table: on CPU these
|
|
# resolve to the numpy reference; on a GPU/accel backend they resolve to that device's
|
|
# kernel — same loop, same StepResult shape (design_v3 §17).
|
|
kernels = model.platform.kernels
|
|
if override is not None and "noise_pred" in override:
|
|
velocity = np.asarray(override["noise_pred"], dtype="float32")
|
|
else:
|
|
dit = model.component(expert_id)
|
|
preds = {b: dit(x, pe if b == "cond" else ne, sigma_t) for b in branches}
|
|
velocity = cfg.combine(preds, scale, sctx, cfg_state)
|
|
velocity = precision.cast(velocity)
|
|
if sde: # FlowGRPO rollout: stochastic + log-prob
|
|
noise = rng.standard_normal(x.shape)
|
|
x_next, logp, _m, _s = kernels.get(FLOW_SDE_STEP)(
|
|
precision.cast(x), velocity, sigma_t, sigma_next, noise=noise, noise_scale=noise_scale)
|
|
return StepResult(output={"noise_pred": np.asarray(velocity, dtype="float32"),
|
|
"latents": x_next.astype("float32"),
|
|
"sde_logprob": logp, "prev": np.asarray(x, dtype="float32")})
|
|
x_next = kernels.get(FLOW_MATCH_STEP)(precision.cast(x), velocity, sigma_t, sigma_next)
|
|
return StepResult(output={"noise_pred": np.asarray(velocity, dtype="float32"),
|
|
"latents": x_next.astype("float32")})
|
|
|
|
cond_bytes = sum(int(np.asarray(e).nbytes) for e in (pe, ne) if e is not None)
|
|
res = ResourceRequest(
|
|
compute_seconds=self.cost.predict(int(np.prod(x.shape)), float(len(branches))),
|
|
resident_bytes=int(x.nbytes) + cond_bytes, # latents + conditioning held for the loop
|
|
peak_activation_bytes=int(x.nbytes)) # one step's transient working buffer
|
|
emits = []
|
|
if st.scratch.get("stream_video"):
|
|
emits.append(StreamChunk(stream_id=st.request_id, modality="video", seq=i,
|
|
data=x, preview=True)) # carry the latent as a preview payload
|
|
return WorkPlan(
|
|
loop_id=self.loop_id, instance_id=st.instance_id, kind=WorkUnitKind.DIFFUSION_STEP,
|
|
shape_sig=ShapeSignature(WorkUnitKind.DIFFUSION_STEP, dims=tuple(x.shape),
|
|
dtype=precision.compute_dtype, # compute dtype is part of the key
|
|
extra=(("cfg", type(cfg).__name__),)),
|
|
resources=res, payload={"branch": "combined", "step": i}, run=run,
|
|
label=f"wan.denoise.{i}", emits=emits,
|
|
# CUDA-graph capture (Path A): the deterministic ODE step is capturable; the stochastic
|
|
# SDE rollout step has host RNG / a data-dependent branch, so it must eager-break. The
|
|
# op-structure key carries the CFG branch set, active expert, and the scheduler-precision
|
|
# flag so a step with a different branch set / expert / solver precision never replays an
|
|
# incompatible captured graph. (Compute dtype rides in shape_sig.dtype above.) NOTE: this
|
|
# is sound for branch-set/expert-determined policies (ClassicCFG); a policy whose op
|
|
# structure forks on other state (PerModalityCFG modality) would need a graph_key hook.
|
|
capturable=not sde,
|
|
graph_key=(tuple(sorted(branches)), expert_id, precision.scheduler_step_in_fp32))
|
|
|
|
def advance(self, st: LoopState, result: StepResult) -> LoopState:
|
|
st.latents["video"] = result.output["latents"]
|
|
if st.profile == ExecutionProfile.ROLLOUT:
|
|
i = st.step_idx
|
|
rec = {"step": i, "sigma": st.sigmas[i],
|
|
"velocity": np.asarray(result.output["noise_pred"]).copy(),
|
|
"latents": np.asarray(st.latents["video"]).copy()}
|
|
if "sde_logprob" in result.output: # FlowGRPO: capture the PPO log-prob slice
|
|
rec.update(sde_logprob=result.output["sde_logprob"],
|
|
prev=np.asarray(result.output["prev"]).copy(),
|
|
sample=np.asarray(result.output["latents"]).copy(),
|
|
sigma_t=st.sigmas[i], sigma_next=st.sigmas[i + 1])
|
|
st.trajectory.append(rec)
|
|
st.step_idx += 1
|
|
return st
|
|
|
|
def finalize(self, st: LoopState) -> LoopResult:
|
|
return LoopResult(outputs={"latents": st.latents["video"]},
|
|
metrics={"denoise_steps": float(st.step_idx)},
|
|
behavior=st.trajectory or None)
|