Need help building with CUDA or Vulkan

[ricardosantos79]

Getting errors while trying to build with CUDA or Vulkan, i can build fine without the flags on for CPU only, some guidance would be deeply appreciated.

System: AMD 5600g, NVIDIA RTX3050, Ubuntu 22.04.5 LTS, installed Cuda toolkit 11.5.1-1ubuntu1, installed Vulkan toolkit 1.3.204.1-2.

Cuda build error:

-- The C compiler identification is GNU 11.4.0
-- The CXX compiler identification is GNU 11.4.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- 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
-- Use CUDA as backend stable-diffusion
-- stable-diffusion.cpp version master-709-92a3b73-74-gea4e566
-- stable-diffusion.cpp commit ea4e566
-- Build static library
-- The ASM compiler identification is GNU
-- Found assembler: /usr/bin/cc
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE  
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: x86_64
-- GGML_SYSTEM_ARCH: x86
-- Found OpenMP_C: -fopenmp (found version "4.5") 
-- Found OpenMP_CXX: -fopenmp (found version "4.5") 
-- Found OpenMP: TRUE (found version "4.5")  
-- Including CPU backend
-- x86 detected
-- Adding CPU backend variant ggml-cpu: -march=native 
-- Found CUDAToolkit: /usr/include (found version "11.5.119") 
-- CUDA Toolkit found
-- 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
-- Using CMAKE_CUDA_ARCHITECTURES=50-virtual;61-virtual;70-virtual;75-virtual;80-virtual;86-real CMAKE_CUDA_ARCHITECTURES_NATIVE=86-real
-- Could NOT find NCCL (missing: NCCL_LIBRARY NCCL_INCLUDE_DIR) 
-- Warning: NCCL not found, performance for multiple CUDA GPUs will be suboptimal
-- CUDA host compiler is GNU 11.4.0
-- Including CUDA backend
-- ggml version: 0.15.3
-- ggml commit:  eced84c8
-- Performing Test HAVE_BUILTIN_BSWAP16
-- Performing Test HAVE_BUILTIN_BSWAP16 - Success
-- Performing Test HAVE_BUILTIN_BSWAP32
-- Performing Test HAVE_BUILTIN_BSWAP32 - Success
-- Performing Test HAVE_BUILTIN_BSWAP64
-- Performing Test HAVE_BUILTIN_BSWAP64 - Success
-- Found OpenGL: /usr/lib/x86_64-linux-gnu/libOpenGL.so   
-- Performing Test HAVE_MATH_LIBRARY
-- Performing Test HAVE_MATH_LIBRARY - Failed
-- Adding -lm flag.
-- Looking for include file GLUT/glut.h
-- Looking for include file GLUT/glut.h - not found
-- Looking for include file GL/glut.h
-- Looking for include file GL/glut.h - not found
-- Looking for include file OpenGL/glut.h
-- Looking for include file OpenGL/glut.h - not found
-- Looking for include file shlwapi.h
-- Looking for include file shlwapi.h - not found
-- Looking for include file unistd.h
-- Looking for include file unistd.h - found
-- Looking for include file wincodec.h
-- Looking for include file wincodec.h - not found
-- Looking for include file windows.h
-- Looking for include file windows.h - not found
-- Performing Test WEBP_HAVE_FLAG_AVX2
-- Performing Test WEBP_HAVE_FLAG_AVX2 - Failed
-- Performing Test WEBP_HAVE_FLAG_AVX2
-- Performing Test WEBP_HAVE_FLAG_AVX2 - Success
-- Performing Test WEBP_HAVE_FLAG_SSE41
-- Performing Test WEBP_HAVE_FLAG_SSE41 - Failed
-- Performing Test WEBP_HAVE_FLAG_SSE41
-- Performing Test WEBP_HAVE_FLAG_SSE41 - Success
-- Performing Test WEBP_HAVE_FLAG_SSE2
-- Performing Test WEBP_HAVE_FLAG_SSE2 - Success
-- Performing Test WEBP_HAVE_FLAG_MIPS32
-- Performing Test WEBP_HAVE_FLAG_MIPS32 - Failed
-- Performing Test WEBP_HAVE_FLAG_MIPS32
-- Performing Test WEBP_HAVE_FLAG_MIPS32 - Failed
-- Performing Test WEBP_HAVE_FLAG_MIPS_DSP_R2
-- Performing Test WEBP_HAVE_FLAG_MIPS_DSP_R2 - Failed
-- Performing Test WEBP_HAVE_FLAG_MIPS_DSP_R2
-- Performing Test WEBP_HAVE_FLAG_MIPS_DSP_R2 - Failed
-- Performing Test HAS_COMPILE_FLAG
-- Performing Test HAS_COMPILE_FLAG - Failed
-- Performing Test WEBP_HAVE_FLAG_NEON
-- Performing Test WEBP_HAVE_FLAG_NEON - Failed
-- Performing Test WEBP_HAVE_FLAG_NEON
-- Performing Test WEBP_HAVE_FLAG_NEON - Failed
-- Performing Test WEBP_HAVE_FLAG_MSA
-- Performing Test WEBP_HAVE_FLAG_MSA - Failed
-- Performing Test WEBP_HAVE_FLAG_MSA
-- Performing Test WEBP_HAVE_FLAG_MSA - Failed
-- Performing Test HAS_COMPILE_FLAG
-- Performing Test HAS_COMPILE_FLAG - Failed
Checking CXX compiler flag support for: -std=c++11
-- Performing Test LIBWEBM_HAVE_CXX_FLAG
-- Performing Test LIBWEBM_HAVE_CXX_FLAG - Success
Checking CXX compiler flag support for: -Wall
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wc++14-compat
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wc++17-compat
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wc++20-compat
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wextra
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wnarrowing
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wno-deprecated
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Success
Checking CXX compiler flag support for: -Wshorten-64-to-32
-- Performing Test CXX_FLAG_SUPPORTED
-- Performing Test CXX_FLAG_SUPPORTED - Failed
CMake Warning at examples/server/CMakeLists.txt:58 (message):
  pnpm not found; frontend build disabled.


-- HAVE_INDEX_HTML disabled
-- Configuring done (5.1s)
-- Generating done (0.1s)
-- Build files have been written to:----------------------------/stable-diffusion.cpp/build
[  1%] Building C object thirdparty/CMakeFiles/zip.dir/zip.c.o
In file included from ----------------------------/stable-diffusion.cpp/thirdparty/zip.c:40:
----------------------------/stable-diffusion.cpp/thirdparty/miniz.h:4988:9: note: ‘#pragma message: Using fopen, ftello, fseeko, stat() etc. path for file I/O - this path may not support large files.’
 4988 | #pragma message(                                                               \
      |         ^~~~~~~
[  1%] Built target zip
[  1%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o
[  2%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml.cpp.o
[  2%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o
[  2%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o
[  2%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-backend-meta.cpp.o
[  3%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o
[  3%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o
[  3%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o
[  3%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o
[  4%] Linking CXX static library libggml-base.a
[  4%] Built target ggml-base
[  5%] Building CUDA object ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/acc.cu.o
nvcc warning : The 'compute_35', 'compute_37', 'compute_50', 'sm_35', 'sm_37' and 'sm_50' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
/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’
gmake[2]: *** [ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/build.make:77: ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/acc.cu.o] Error 1
gmake[1]: *** [CMakeFiles/Makefile2:391: ggml/src/ggml-cuda/CMakeFiles/ggml-cuda.dir/all] Error 2
gmake: *** [Makefile:136: all] Error 2

Vulkan build error:

-- The C compiler identification is GNU 11.4.0
-- The CXX compiler identification is GNU 11.4.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- 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
-- Use Vulkan as backend stable-diffusion
-- stable-diffusion.cpp version master-709-92a3b73-74-gea4e566
-- stable-diffusion.cpp commit ea4e566
-- Build static library
-- The ASM compiler identification is GNU
-- Found assembler: /usr/bin/cc
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE  
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: x86_64
-- GGML_SYSTEM_ARCH: x86
-- Found OpenMP_C: -fopenmp (found version "4.5") 
-- Found OpenMP_CXX: -fopenmp (found version "4.5") 
-- Found OpenMP: TRUE (found version "4.5")  
-- Including CPU backend
-- x86 detected
-- Adding CPU backend variant ggml-cpu: -march=native 
CMake Error at -------------------/.local/lib/python3.10/site-packages/cmake/data/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find Vulkan (missing: Vulkan_LIBRARY Vulkan_INCLUDE_DIR glslc)
  (found version "")
Call Stack (most recent call first):
 -------------------/.local/lib/python3.10/site-packages/cmake/data/share/cmake-3.26/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE)
 -------------------/.local/lib/python3.10/site-packages/cmake/data/share/cmake-3.26/Modules/FindVulkan.cmake:597 (find_package_handle_standard_args)
  ggml/src/ggml-vulkan/CMakeLists.txt:9 (find_package)


-- Configuring incomplete, errors occurred!
gmake: Makefile: No such file or directory
gmake: *** No rule to make target 'Makefile'.  Stop.

[darekwojciechowski]

The error isn't in sd.cpp:

/usr/include/c++/11/bits/std_function.h:435: error: parameter packs not expanded with '...'

That's CUDA 11.5's nvcc choking on GCC 11 headers. Ubuntu 22.04 ships both, so the apt combo is broken out of the box. Your CPU build works because nvcc never runs.

Best fix is CUDA 12.x from NVIDIA's repo — your 3050 is sm_86 and fully supported. Build in a fresh directory, the old one caches the bad paths:

rm -rf build
cmake -B build -DSD_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86
cmake --build build -j$(nproc)

If you want to keep 11.5, pin an older host compiler instead:

sudo apt install g++-10
cmake -B build -DSD_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86 -DCMAKE_CUDA_HOST_COMPILER=/usr/bin/g++-10

Setting the arch to 86 also cuts build time a lot. You'll know it worked when acc.cu.o gets past 5%, which is where it died.

For Vulkan I'd need the actual error, but the usual blocker is glslc missing — try sudo apt install glslc and reconfigure from a clean build dir.

[ricardosantos79]

thank you, i was able to build with the 2nd step, maintaining the current version.. \o/