Files
huchukato 9e6d4c195d Fix CUDA graph exec handle released with wrong cudart call
cuda_graph_instance is a cudaGraphExec_t from cudaGraphInstantiate()
and must be released with cudaGraphExecDestroy(), not cudaGraphDestroy().
The handle-type mismatch corrupted the CUDA context on CUDA 13, causing
a silent process death after the node finished (keep_model_loaded=False)
or cudaErrorInvalidValue on the next run (keep_model_loaded=True).
2026-10-01 22:08:17 +02:00

15 lines
652 B
TOML

[project]
name = "Auto-TensorRT-RIFE"
description = "Ultra fast video frame interpolation for ComfyUI using RIFE with TensorRT acceleration. Auto-detects CUDA version, installs matching TensorRT wheels, and downloads RIFE models automatically."
version = "1.2.4"
license = { file = "LICENSE" }
[project.urls]
Repository = "https://github.com/huchukato/ComfyUI-RIFE-TensorRT-Auto"
[tool.comfy]
PublisherId = "huchukato"
DisplayName = "Auto TensorRT RIFE"
Icon = "https://raw.githubusercontent.com/huchukato/ComfyUI-RIFE-TensorRT-Auto/master/icon.png"
Banner = "https://raw.githubusercontent.com/huchukato/ComfyUI-RIFE-TensorRT-Auto/master/banner.png"