Cuda backend fails to build

Zama TFHE-rs
1

Hello everyone, I’m trying to build cuda backend on Ubuntu 22.04.4 LTS Jammy, I have Cuda compilation tools, release 11.5, V11.5.119 installed and NVIDIA-SMI 535.171.04 Driver Version: 551.78 CUDA Version: 12.4. But I get the following error
Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/device.cu.o
/home/claoo/Projects/rust/tfhe-rs/backends/tfhe-cuda-backend/cuda/src/device.cu(91): error: identifier “cudaDevAttrClusterLaunch” is undefined

1 error detected in the compilation of “/home/claoo/Projects/rust/tfhe-rs/backends/tfhe-cuda-backend/cuda/src/device.cu”.
make[2]: *** [src/CMakeFiles/tfhe_cuda_backend.dir/build.make:107: src/CMakeFiles/tfhe_cuda_backend.dir/device.cu.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:116: src/CMakeFiles/tfhe_cuda_backend.dir/all] Error 2
make: *** [Makefile:146: all] Error 2
Does any of you knows what is the issue. I’m using Ubuntu over Windows 11 with WSL. can this be the issue?

2

Hello!

Are you on the latest main? What is your GPU model?

Cheers,
Agnès

3

Hi, yes I’m using the latest main and I see that the file with the issue was modified one week ago, and is also a comment on that specific line of code. My GPU is NVIDIA GeForce RTX 3060.

Thx,
Claoo

4

May I ask you to try and build on this branch: chore/gpu/tbcchecksupport, see if it fixes compilation for you? Thank you :pray:

5

this time I get a different error
– The CXX compiler identification is GNU 11.4.0
– Detecting CXX compiler ABI info
– Detecting CXX compiler ABI info - done
– Check for working CXX compiler: /usr/bin/c++ - skipped
– Detecting CXX compile features
– Detecting CXX compile features - done
– Looking for a CUDA compiler
– Looking for a CUDA compiler - /usr/bin/nvcc
– The CUDA compiler identification is NVIDIA 11.5.119
– Detecting CUDA compiler ABI info
– Detecting CUDA compiler ABI info - done
– Check for working CUDA compiler: /usr/bin/nvcc - skipped
– Detecting CUDA compile features
– Detecting CUDA compile features - done
– CUDA Architecture: -arch=sm_86
– CUDA Version:
– CUDA Path:
– CUDA Libraries:
– CUDA Performance Primitives:
– Found OpenMP_CXX: -fopenmp (found version “4.5”)
– Found OpenMP: TRUE (found version “4.5”)
– Configuring done (7.0s)
– Generating done (0.0s)
– Build files have been written to: /home/claoo/Projects/tfhe-rs/backends/tfhe-cuda-backend/cuda/build
claoo@DESKTOP-6SHLCUU:~/Projects/tfhe-rs/backends/tfhe-cuda-backend/cuda/build$ make
[ 3%] Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/crypto/ciphertext.cu.o
[ 6%] Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/crypto/keyswitch.cu.o
[ 10%] Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/device.cu.o
[ 13%] Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/fft/twiddles.cu.o
[ 17%] Building CUDA object src/CMakeFiles/tfhe_cuda_backend.dir/integer/bitwise_ops.cu.o
/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’
make[2]: *** [src/CMakeFiles/tfhe_cuda_backend.dir/build.make:137: src/CMakeFiles/tfhe_cuda_backend.dir/integer/bitwise_ops.cu.o] Error 1
claoo@DESKTOP-6SHLCUU:~/Projects/tfhe-rs/backends/tfhe-cuda-backend/cuda/build$
make: *** [Makefile:146: all] Error 2

6

Have you followed this guide for Cuda installation on WSL? CUDA on WSL

(Cuda download page: CUDA Toolkit 12.4 Update 1 Downloads | NVIDIA Developer)

1 Like
7

Thank you very much, problem solved. I reinstall everything from scratch and I manage to build cuda backend.
Happy coding!

1 Like