andreas/ComfyUI
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Mixed Precision Quantization System (#10498)

Browse Source 
* Implement mixed precision operations with a registry design and metadate for quant spec in checkpoint.

* Updated design using Tensor Subclasses

* Fix FP8 MM

* An actually functional POC

* Remove CK reference and ensure correct compute dtype

* Update unit tests

* ruff lint

* Implement mixed precision operations with a registry design and metadate for quant spec in checkpoint.

* Updated design using Tensor Subclasses

* Fix FP8 MM

* An actually functional POC

* Remove CK reference and ensure correct compute dtype

* Update unit tests

* ruff lint

* Fix missing keys

* Rename quant dtype parameter

* Rename quant dtype parameter

* Fix unittests for CPU build
This commit is contained in:
contentis
2025-10-28 21:20:53 +01:00
committed by GitHub
parent 22e40d2ace
commit 8817f8fc14
8 changed files with 1030 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
comfy/supported_models_base.py
View File

@@ -50,6 +50,7 @@ class BASE:
manual_cast_dtype = None
custom_operations = None
scaled_fp8 = None
layer_quant_config = None # Per-layer quantization configuration for mixed precision
optimizations = {"fp8": False}
@classmethod