Open jaschn12 opened 2 months ago
Hi @jaschn12!
Can you please let me know what the output of
nvcc --version
gcc --version
Are?
(base) MY_USERNAME@MY_SERVER_NAME:~$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Thu_Nov_18_09:45:30_PST_2021
Cuda compilation tools, release 11.5, V11.5.119
Build cuda_11.5.r11.5/compiler.30672275_0
(base) MY_USERNAME@MY_SERVER_NAME:~$ gcc --version
gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
@jaschn12 no idea unfortunately. @guoqingbao have you seen anything like this by chance?
One of the CUDA source files (src/cuda/nonzero_bitwise.cu) is unable to be compiled because it is not compatible with the system C++ library under current NVCC compiler. Some people said it was a bug in a specific CUDA version. I suggest @jaschn12 use a different CUDA package.
https://stackoverflow.com/questions/74350584/nvcc-compilation-error-using-thrust-in-cuda-11-5
Root cause:
/usr/include/c++/11/bits/std_function.h:435:145: error: parameter packs not expanded with ‘...’:
435 | function(_Functor&& __f)
| ^
/usr/include/c++/11/bits/std_function.h:435:145: note: ‘_ArgTypes’
/usr/include/c++/11/bits/std_function.h:530:146: error: parameter packs not expanded with ‘...’:
530 | operator=(_Functor&& __f)
| ^
/usr/include/c++/11/bits/std_function.h:530:146: note: ‘_ArgTypes’
# --error 0x1 --
thread 'main' panicked at /home/MY_USERNAME/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.5/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_86" "-c" "-o" "/tmp/pip-install-k3v_9rnc/mistralrs-cuda_d8edf27dd7b84807a46c329fbd384856/target/release/build/mistralrs-core-b94df057dc002129/out/nonzero_bitwise-b50867152df76f01.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "src/cuda/nonzero_bitwise.cu"
Perhaps he's using Windows Subsystem for Linux (WSL) for compiling Mistral.rs, do we support WSL at the moment?
Thanks for getting back @guoqingbao. @jaschn12 perhaps you could upgrade your CUDA package - upgrading to 11.8 worked in the linked thread.
Perhaps he's using Windows Subsystem for Linux (WSL) for compiling Mistral.rs, do we support WSL at the moment?
Yes, we support WSL.
@jaschn12 I think I found a fix. Given that you are on GCC 11, this is actually a compiler bug! Please install GCC 10 and run with
NVCC_CCBIN=/usr/lib/gcc/x86_64-linux-gnu/10
.
Describe the bug I tried to install the mistral.rs PyO3 Bindings: mistralrs but it failed. I followed the instructions under https://github.com/EricLBuehler/mistral.rs/blob/master/mistralrs-pyo3/README.md#installation-from-pypi
Here are my terminal outputs:
Latest commit Which commit you ran with.