diodiogod
efe8d295d4
- Clarify that global patch solves TWO issues: TorchCodec DLL + int16 normalization - Explain timing criticality: patch must apply before torchaudio import - Remove redundant pytorch_patches call from apply_all_compatibility_patches() - Update save_audio_safe docstring to explain defense-in-depth approach - Add detailed docstring to pytorch_patches explaining both issues and fixes
49 lines
1.7 KiB
Python
49 lines
1.7 KiB
Python
"""
|
|
ComfyUI Compatibility Patches for TTS Audio Suite
|
|
|
|
This module applies necessary patches to resolve compatibility issues
|
|
between ComfyUI versions and the TTS Audio Suite.
|
|
"""
|
|
|
|
import sys
|
|
import logging
|
|
|
|
_cudnn_fix_applied = False
|
|
|
|
def ensure_python312_cudnn_fix():
|
|
"""
|
|
Ensure Python 3.12 CUDNN benchmark fix is applied before TTS operations.
|
|
|
|
This is called from TTS nodes before generation to prevent VRAM spikes.
|
|
"""
|
|
global _cudnn_fix_applied
|
|
|
|
if _cudnn_fix_applied or sys.version_info[:2] != (3, 12):
|
|
return
|
|
|
|
try:
|
|
import torch
|
|
from comfy.cli_args import args, PerformanceFeature
|
|
|
|
# Check if CUDNN benchmark is enabled (the problematic setting)
|
|
if (torch.cuda.is_available() and
|
|
torch.backends.cudnn.is_available() and
|
|
hasattr(args, 'fast') and
|
|
args.fast and
|
|
PerformanceFeature.AutoTune in args.fast):
|
|
|
|
# Disable the problematic CUDNN benchmarking
|
|
torch.backends.cudnn.benchmark = False
|
|
print("🩹 TTS AUDIO SUITE CUDNN FIX APPLIED")
|
|
print(" Disabled CUDNN benchmark on Python 3.12 to prevent VRAM spikes")
|
|
print(" This fixes ComfyUI v0.3.57+ regression - VRAM spikes eliminated!")
|
|
_cudnn_fix_applied = True
|
|
|
|
except Exception as e:
|
|
print(f"🩹 Warning: Could not apply CUDNN fix: {e}")
|
|
|
|
def apply_all_compatibility_patches():
|
|
"""Apply all necessary ComfyUI compatibility patches."""
|
|
# Python 3.12 CUDNN fix
|
|
if sys.version_info[:2] == (3, 12):
|
|
print("🩹 TTS Audio Suite: Python 3.12 CUDNN fix ready (will apply before TTS generation)") |