Comments (5)
Could you show the full logs, and let me know the nvcc, gcc, and cmake versions you are using? We don't have official support for building FBGEMM_GPU on WSL2, but if WSL2 works exactly the same as a regular Linux environment, then it may be just an issue with the nvcc and/or gcc versions.
from fbgemm.
Here is the full logs.
nvcc version: Cuda compilation tools, release 12.3, V12.3.52
gcc version: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0
cmake version: cmake version 3.25.3
OS: wsl2 Ubuntu20.4 on Windows10
wsl kernel version: 5.15.133.1-1
wsl version: 2.0.9.0
C:\Windows\system32\wsl.exe --distribution Ubuntu-20.04 --exec /bin/bash -c "export CMAKE_COLOR_DIAGNOSTICS=ON && export CLION_IDE=TRUE && export JETBRAINS_IDE=TRUE && cd /home/wtx/workspace/cpp_project/FBGEMM/fbgemm_gpu/cmake-build-fbgemm-gpu-debug && /usr/bin/cmake -DCMAKE_BUILD_TYPE=Debug -DCMAKE_C_COMPILER=/usr/bin/gcc '-DCMAKE_CXX_COMPILER=/usr/bin/g++' '-DCMAKE_PREFIX_PATH=D:\workspace\cpp_project\libtorch' -G 'CodeBlocks - Unix Makefiles' -S /home/wtx/workspace/cpp_project/FBGEMM/fbgemm_gpu -B /home/wtx/workspace/cpp_project/FBGEMM/fbgemm_gpu/cmake-build-fbgemm-gpu-debug"
================================================================================
Building the CUDA variant of FBGEMM-GPU
================================================================================
================================================================================
Default C++ compiler flags
(values may be overridden by CMAKE_CXX_STANDARD and CXX_STANDARD):
================================================================================
-- The CXX compiler identification is GNU 9.4.0
-- The C compiler identification is GNU 9.4.0
-- The CUDA compiler identification is NVIDIA 12.3.52
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/g++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/gcc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CUDA compiler ABI info
-- Detecting CUDA compiler ABI info - done
-- Check for working CUDA compiler: /usr/local/cuda/bin/nvcc - skipped
-- Detecting CUDA compile features
-- Detecting CUDA compile features - done
-- Performing Test C_HAS_AVX_1
-- Performing Test C_HAS_AVX_1 - Failed
-- Performing Test C_HAS_AVX_2
-- Performing Test C_HAS_AVX_2 - Success
-- Performing Test C_HAS_AVX2_1
-- Performing Test C_HAS_AVX2_1 - Failed
-- Performing Test C_HAS_AVX2_2
-- Performing Test C_HAS_AVX2_2 - Success
-- Performing Test C_HAS_AVX512_1
-- Performing Test C_HAS_AVX512_1 - Failed
-- Performing Test C_HAS_AVX512_2
-- Performing Test C_HAS_AVX512_2 - Failed
-- Performing Test C_HAS_AVX512_3
-- Performing Test C_HAS_AVX512_3 - Failed
-- Performing Test CXX_HAS_AVX_1
-- Performing Test CXX_HAS_AVX_1 - Failed
-- Performing Test CXX_HAS_AVX_2
-- Performing Test CXX_HAS_AVX_2 - Success
-- Performing Test CXX_HAS_AVX2_1
-- Performing Test CXX_HAS_AVX2_1 - Failed
-- Performing Test CXX_HAS_AVX2_2
-- Performing Test CXX_HAS_AVX2_2 - Success
-- Performing Test CXX_HAS_AVX512_1
-- Performing Test CXX_HAS_AVX512_1 - Failed
-- Performing Test CXX_HAS_AVX512_2
-- Performing Test CXX_HAS_AVX512_2 - Failed
-- Performing Test CXX_HAS_AVX512_3
-- Performing Test CXX_HAS_AVX512_3 - Failed
-- Skipping merge_pooled_embeddings sources
-- Configuring done
-- Generating done
-- Build files have been written to: /home/wtx/workspace/cpp_project/FBGEMM/fbgemm_gpu/cmake-build-fbgemm-gpu-debug
Cannot get compiler information:
Compiler exited with error code 1: \usr\local\cuda\bin\nvcc -Dfbgemm_gpu_py_EXPORTS -g --generate-code=arch=compute_75,code=[compute_75,sm_75] -Xcompiler=-fPIC -std=c++17 -mavx -mf16c -mfma -mavx2 -fopenmp --dryrun /mnt/c/Users/Lenovo/AppData/Local/Temp/compiler-file11626473163701549697.cu
nvcc fatal : Unknown option '-mavx'
from fbgemm.
Hmm, some of the tool versions you listed are not the recommended versions - in particular, CUDA should be at 12.1 and GCC should be at 10+.
I did quick Google search cases where users have run into a similar error signature before, and the solution was generally to upgrade the drivers and toolkit. That being said, we highly recommend using building and using FBGEMM in a full Linux environment instead, as WSL may contain intricacies that are beyond our scope for debugging issues like this.
from fbgemm.
@shadow150519 could you try latest main
, and run the usual FBGEMM_GPU build command, but with -DWSL_MODE=1
added to the args?
python setup.py ...<usual args>... -DWSL_MODE=1
(See here for build instructions in detail)
Let us know how this goes.
from fbgemm.
sorry for the late response, I've moved to a linux machine right now and it works correctly. I'll try this and give you the result as soon as I can. Thank for your help :)
@shadow150519 could you try latest
main
, and run the usual FBGEMM_GPU build command, but with-DWSL_MODE=1
added to the args?python setup.py ...<usual args>... -DWSL_MODE=1
(See here for build instructions in detail)
Let us know how this goes.
from fbgemm.
Related Issues (20)
- Support MacOS? HOT 1
- `.to("meta")` is leaked to the public main branch in the tests. HOT 2
- Can't compile FBGEMM with GCC 12.3.0 HOT 4
- compiling FBGEMM for ARM HOT 4
- How `partition_avx512` is auto-tuned? HOT 3
- Compiling on windows with mingw
- Error importing fbgemm_gpu HOT 15
- Build failure on MacOS HOT 6
- momentum for SGD/Adagrad HOT 2
- RuntimeError: No such operator fbgemm::jagged_2d_to_dense HOT 3
- Having issue installing FBGEMM-gpu on MacOS HOT 8
- Latest FBGEMM doesn't build with latest PyTorch HOT 2
- fbgemm_gpu doesn't build for CPU because impl_abstract_pystub is not found HOT 1
- [FBGEMM_GPU Question] When should I use FusedEmbeddingBagCollection over EmbeddingBagCollection?
- [Question] What does device / embedding_specs.compute_device parameter in ctor of TBE mean? HOT 4
- AttributeError: '_OpNamespace' 'fbgemm' object has no attribute 'jagged_2d_to_dense' HOT 5
- Building error after C++20 HOT 3
- quantize_embeddings + KeyedJaggedTensor+ vbe cannot work
- AttributeError: '_OpNamespace' 'fbgemm' object has no attribute 'jagged_2d_to_dense' HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from fbgemm.