poxular and Cursor
bd86190129
Match CUDA one-shot output to CPU MagicMatch (ONNX on GPU only).
...
Use CPU scene extract, develop, and apply; GPU develop apply caused brighter drift.
Optional MAGICMATCH_CUDA_GPU_APPLY=1 restores legacy GPU apply path.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 01:59:45 +02:00
poxular and Cursor
0361126045
Restore 1ccd29d CUDA one-shot path without breaking node imports.
...
Add pipeline_v1 (GPU build + GPU apply) as a standalone module so partial
git checkouts of inference_cuda no longer break GPU Full/Fast imports.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 01:54:41 +02:00
poxular and Cursor
aeae298c40
Restore CUDA one-shot to parity build with CPU apply.
...
Rolling back inference_cuda alone broke GPU Full/Fast imports and hid all nodes.
CUDA build uses pipeline_cuda again; one-shot apply matches golden CPU develop stack.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 01:39:20 +02:00
poxular and Cursor
c85abee31e
Speed up GPU path: fused one-shot, GPU net resize, vectorized luminance.
...
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 00:53:08 +02:00
poxular and Cursor
1ccd29df6f
Add Torch GPU pipeline for CUDA ComfyUI nodes.
...
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 00:43:01 +02:00
poxular and Cursor
d101418eb9
Add separate CUDA ONNX path for experimental GPU inference.
...
Keeps CPU parity nodes and providers unchanged; CUDA modules and ComfyUI
nodes run alongside with CPU fallback when CUDA EP is unavailable.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-06-02 00:22:25 +02:00
