can not install by setup.py because /ld: cannot find -lcublas_static

[meicale]

Hi,
I think this repo is great and tried it using pre-baked docker image for nvidi, and after following the instructions in Readme.md, I got these errors.
/root/anaconda3/envs/int/bin/../lib/gcc/x86_64-conda-linux-gnu/9.3.0/../../../../x86_64-conda-linux-gnu/bin/ld: cannot find -lcublas_static
/root/anaconda3/envs/int/bin/../lib/gcc/x86_64-conda-linux-gnu/9.3.0/../../../../x86_64-conda-linux-gnu/bin/ld: cannot find -lcublasLt_static

I guess maybe the cublass location is different, but I can not figure out how to make it right. Would you mind helping me out? Thx!

I also can not find cublasLt_static or cublas_static by using find. The results were:
(int) root@bafd706ba0bf:/workspace/torch-int# find / -name "cublasLt" -print
/root/anaconda3/lib/python3.10/site-packages/torch/lib/libcublasLt.so.11
/root/anaconda3/envs/int/lib/python3.8/site-packages/torch/lib/libcublasLt.so.11
/usr/local/cuda-11.3/targets/x86_64-linux/lib/stubs/libcublasLt.so
/usr/local/cuda-11.3/targets/x86_64-linux/lib/libcublasLt.so
/usr/local/cuda-11.3/targets/x86_64-linux/lib/libcublasLt.so.11
/usr/local/cuda-11.3/targets/x86_64-linux/lib/libcublasLt.so.11.5.1.109
/usr/local/cuda-11.3/targets/x86_64-linux/include/cublasLt.h

BTW, if I link cublas without static it can finish the compilation. But is will fail the test on readme. python tests/test_linear_modules.py

[yunzhongOvO]

I met the same problem as you, does it fixed?

[FurryMushroom]

The same problem. Have you dear friends fixed, it's already October now

[FurryMushroom]

Friends I've successfully finished the compilation and test. I just modified the lib names in setup.py, removing _static suffix. You can have a look at my page for more details.

[Hukongtao]

Friends I've successfully finished the compilation and test. I just modified the lib names in setup.py, removing _static suffix. You can have a look at my page for more details.

I use your torch-int，and still got an error:

[FurryMushroom]

Friends I've successfully finished the compilation and test. I just modified the lib names in setup.py, removing _static suffix. You can have a look at my page for more details.

I use your torch-int，and still got an error:

Please refer to Quantization work of DesignOrder, experiments on smoothquant