sageattention-2.2.0+cu130torch2.9.1-cp313-cp313-win_amd64.whl

[windFFYL]

Can you help me compile this file? Thank you

[polymath-wtf]

Some things cannot be found on Civitai
I'm already learn to searching way on Huggingface and Github lol
https://huggingface.co/Kijai/PrecompiledWheels/tree/main

[windFFYL]

Some things cannot be found on Civitai I'm already learn to searching way on Huggingface and Github lol https://huggingface.co/Kijai/PrecompiledWheels/tree/main

cp313-cp313，Python 3.13 doesn't have this version, haha

[polymath-wtf]

cp313-cp313，Python 3.13 doesn't have this version, haha

lol didn't notice about cp313
Then try official whls https://github.com/woct0rdho/SageAttention/releases
or random 2.2.1 cp313 from HF https://huggingface.co/eddy1111111/sageattention-2.2.0-cp312-270-cu128-cp312-win_amd64/tree/main

im good at sageattention-2.2.0+cu130torch2.9.0andhigher.post4-cp39-abi3-win_amd64.whl

[loscrossos]

windows buiild is currently not working.

you can build applying my PR

#352

a better way is to use whoctors files :)

[boricuapab]

Was finally able to build from source using cuda toolkit 13.1, pytorch version: 2.10.0+cu130 with Python version: 3.12.7 using the modifications suggested by loscrossos in the setup.py for sage 2.2

    CXX_FLAGS = ["-g", "-O3", "-fopenmp", "-lgomp", "-std=c++17", "-DENABLE_BF16", "-DUSE_CUDA"]
    NVCC_FLAGS = [
        "-O3",
        "-std=c++17",
        "-U__CUDA_NO_HALF_OPERATORS__",
        "-U__CUDA_NO_HALF_CONVERSIONS__",
        "--use_fast_math",
        "--threads=8",
        "-Xptxas=-v",
        "-diag-suppress=174",
        "-DUSE_CUDA",
    ]