github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1024] Packaging ollama + cuda for Arch Linux #47011

New Issue
Closed
opened 2026-04-28 02:40:23 -05:00 by GiteaMirror · 11 comments
GiteaMirror commented 2026-04-28 02:40:23 -05:00
Owner
Copy Link

Originally created by @xyproto on GitHub (Nov 7, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1024

Hi! Arch Linux package maintainer for the ollama and ollama-cuda packages here.

Thanks for creating Ollama, it makes LLMs more fun to deal with!

When compiling v0.1.8 for Arch Linux, using this PKGBUILD:

pkgname=ollama-cuda
pkgdesc='Create, run and share large language models (LLMs) with CUDA'
pkgver=0.1.8
pkgrel=1
arch=(x86_64)
url='https://github.com/jmorganca/ollama'
license=(MIT)
provides=(ollama)
conflicts=(ollama)
makedepends=(cmake cuda git go setconf)
_ollamacommit=e21579a0f1548e2d1f77411af3df2037c1f144fe # tag: v0.1.8
# The git submodule commit hashes are here:
# https://github.com/jmorganca/ollama/tree/v0.1.8/llm/llama.cpp
_ggmlcommit=9e232f0234073358e7031c1b8d7aa45020469a3b
_ggufcommit=9e70cc03229df19ca2d28ce23cc817198f897278
source=(git+$url#commit=$_ollamacommit
        ggml::git+https://github.com/ggerganov/llama.cpp#commit=$_ggmlcommit
        gguf::git+https://github.com/ggerganov/llama.cpp#commit=$_ggufcommit)
b2sums=('SKIP'
        'SKIP'
        'SKIP')

prepare() {
  cd ${pkgname/-cuda}

  rm -frv llm/llama.cpp/gg{ml,uf}

  # Copy git submodule files instead of symlinking because the build process is sensitive to symlinks.
  cp -r "$srcdir/ggml" llm/llama.cpp/ggml
  cp -r "$srcdir/gguf" llm/llama.cpp/gguf

  # Do not git clone when "go generate" is being run.
  sed -i 's,git submodule,true,g' llm/llama.cpp/generate_linux.go

  # Set the version number
  setconf version/version.go 'var Version string' "\"$pkgver\""
}

build() {
  export CGO_CFLAGS="$CFLAGS" CGO_CPPFLAGS="$CPPFLAGS" CGO_CXXFLAGS="$CXXFLAGS" CGO_LDFLAGS="$LDFLAGS"
  cd ${pkgname/-cuda}
  go generate ./...
  go build -buildmode=pie -trimpath -mod=readonly -modcacherw -ldflags=-linkmode=external -ldflags=-buildid=''
}

check() {
  cd ${pkgname/-cuda}
  go test ./...
}

package() {
  cd ${pkgname/-cuda}
  install -Dm755 ${pkgname/-cuda} "$pkgdir/usr/bin/${pkgname/-cuda}"
  install -Dm644 LICENSE "$pkgdir/usr/share/licenses/$pkgname/LICENSE"
}

I get:

- Found CUDAToolkit: /opt/cuda/include (found version "12.3.52")
-- cuBLAS found
CMake Error at /usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:753 (message):
Compiling the CUDA compiler identification source file
"CMakeCUDACompilerId.cu" failed.

Compiler: /opt/cuda/bin/nvcc

Build flags:

Id flags: --keep;--keep-dir;tmp -v



The output was:

2

#$ _NVVM_BRANCH_=nvvm

#$ _SPACE_=

#$ _CUDART_=cudart

#$ _HERE_=/opt/cuda/bin

#$ _THERE_=/opt/cuda/bin

#$ _TARGET_SIZE_=

#$ _TARGET_DIR_=

#$ _TARGET_DIR_=targets/x86_64-linux

#$ TOP=/opt/cuda/bin/..

#$ NVVMIR_LIBRARY_DIR=/opt/cuda/bin/../nvvm/libdevice

#$ LD_LIBRARY_PATH=/opt/cuda/bin/../lib:

#$
PATH=/opt/cuda/bin/../nvvm/bin:/opt/cuda/bin:/usr/lib/go/bin:/usr/local/sbin:/usr/local/bin:/usr/bin:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_pn


#$ INCLUDES="-I/opt/cuda/bin/../targets/x86_64-linux/include"

#$ LIBRARIES= "-L/opt/cuda/bin/../targets/x86_64-linux/lib/stubs"
"-L/opt/cuda/bin/../targets/x86_64-linux/lib"

#$ CUDAFE_FLAGS=

#$ PTXAS_FLAGS=

#$ rm tmp/a_dlink.reg.c

#$ gcc -D__CUDA_ARCH_LIST__=520 -E -x c++ -D__CUDACC__ -D__NVCC__
"-I/opt/cuda/bin/../targets/x86_64-linux/include" -D__CUDACC_VER_MAJOR__=12
-D__CUDACC_VER_MINOR__=3 -D__CUDACC_VER_BUILD__=52
-D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=3
-D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64
"CMakeCUDACompilerId.cu" -o "tmp/CMakeCUDACompilerId.cpp4.ii"

#$ cudafe++ --c++17 --gnu_version=130201 --display_error_number
--orig_src_file_name "CMakeCUDACompilerId.cu" --orig_src_path_name
"/build/ollama-cuda/src/ollama/llm/llama.cpp/ggml/build/cuda/CMakeFiles/3.27.7/CompilerIdCUDA/CMakeCUDACompilerId.cu"
--allow_managed --m64 --parse_templates --gen_c_file_name
"tmp/CMakeCUDACompilerId.cudafe1.cpp" --stub_file_name
"CMakeCUDACompilerId.cudafe1.stub.c" --gen_module_id_file
--module_id_file_name "tmp/CMakeCUDACompilerId.module_id"
"tmp/CMakeCUDACompilerId.cpp4.ii"

/usr/include/bits/floatn.h(86): error: invalid combination of type
specifiers

typedef __float128 _Float128;
^



/usr/include/bits/floatn-common.h(214): error: invalid combination of type
specifiers

typedef float _Float32;
^



/usr/include/bits/floatn-common.h(251): error: invalid combination of type
specifiers

typedef double _Float64;
^



/usr/include/bits/floatn-common.h(268): error: invalid combination of type
specifiers

typedef double _Float32x;
^



/usr/include/bits/floatn-common.h(285): error: invalid combination of type
specifiers

typedef long double _Float64x;
^



5 errors detected in the compilation of "CMakeCUDACompilerId.cu".

# --error 0x2 --





Call Stack (most recent call first):
/usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:8 (CMAKE_DETERMINE_COMPILER_ID_BUILD)
/usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:53 (__determine_compiler_id_test)
/usr/share/cmake/Modules/CMakeDetermineCUDACompiler.cmake:307 (CMAKE_DETERMINE_COMPILER_ID)
CMakeLists.txt:250 (enable_language)


-- Configuring incomplete, errors occurred!
llm/llama.cpp/generate_linux.go:21: running "cmake": exit status 1

Building the regular ollama package, where cuda/nvcc are not installed before building, works fine.

Attaching logs.

ollama-cuda-0.1.8-1-x86_64-build.log
ollama-cuda-0.1.8-1-x86_64-prepare.log

Advice for how to package ollama + cuda for Arch Linux is welcome.

Originally created by @xyproto on GitHub (Nov 7, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1024 Hi! Arch Linux package maintainer for the `ollama` and `ollama-cuda` packages here. Thanks for creating Ollama, it makes LLMs more fun to deal with! When compiling `v0.1.8` for Arch Linux, using this PKGBUILD: ```bash pkgname=ollama-cuda pkgdesc='Create, run and share large language models (LLMs) with CUDA' pkgver=0.1.8 pkgrel=1 arch=(x86_64) url='https://github.com/jmorganca/ollama' license=(MIT) provides=(ollama) conflicts=(ollama) makedepends=(cmake cuda git go setconf) _ollamacommit=e21579a0f1548e2d1f77411af3df2037c1f144fe # tag: v0.1.8 # The git submodule commit hashes are here: # https://github.com/jmorganca/ollama/tree/v0.1.8/llm/llama.cpp _ggmlcommit=9e232f0234073358e7031c1b8d7aa45020469a3b _ggufcommit=9e70cc03229df19ca2d28ce23cc817198f897278 source=(git+$url#commit=$_ollamacommit ggml::git+https://github.com/ggerganov/llama.cpp#commit=$_ggmlcommit gguf::git+https://github.com/ggerganov/llama.cpp#commit=$_ggufcommit) b2sums=('SKIP' 'SKIP' 'SKIP') prepare() { cd ${pkgname/-cuda} rm -frv llm/llama.cpp/gg{ml,uf} # Copy git submodule files instead of symlinking because the build process is sensitive to symlinks. cp -r "$srcdir/ggml" llm/llama.cpp/ggml cp -r "$srcdir/gguf" llm/llama.cpp/gguf # Do not git clone when "go generate" is being run. sed -i 's,git submodule,true,g' llm/llama.cpp/generate_linux.go # Set the version number setconf version/version.go 'var Version string' "\"$pkgver\"" } build() { export CGO_CFLAGS="$CFLAGS" CGO_CPPFLAGS="$CPPFLAGS" CGO_CXXFLAGS="$CXXFLAGS" CGO_LDFLAGS="$LDFLAGS" cd ${pkgname/-cuda} go generate ./... go build -buildmode=pie -trimpath -mod=readonly -modcacherw -ldflags=-linkmode=external -ldflags=-buildid='' } check() { cd ${pkgname/-cuda} go test ./... } package() { cd ${pkgname/-cuda} install -Dm755 ${pkgname/-cuda} "$pkgdir/usr/bin/${pkgname/-cuda}" install -Dm644 LICENSE "$pkgdir/usr/share/licenses/$pkgname/LICENSE" } ``` I get: ``` - Found CUDAToolkit: /opt/cuda/include (found version "12.3.52") -- cuBLAS found CMake Error at /usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:753 (message): Compiling the CUDA compiler identification source file "CMakeCUDACompilerId.cu" failed. Compiler: /opt/cuda/bin/nvcc Build flags: Id flags: --keep;--keep-dir;tmp -v The output was: 2 #$ _NVVM_BRANCH_=nvvm #$ _SPACE_= #$ _CUDART_=cudart #$ _HERE_=/opt/cuda/bin #$ _THERE_=/opt/cuda/bin #$ _TARGET_SIZE_= #$ _TARGET_DIR_= #$ _TARGET_DIR_=targets/x86_64-linux #$ TOP=/opt/cuda/bin/.. #$ NVVMIR_LIBRARY_DIR=/opt/cuda/bin/../nvvm/libdevice #$ LD_LIBRARY_PATH=/opt/cuda/bin/../lib: #$ PATH=/opt/cuda/bin/../nvvm/bin:/opt/cuda/bin:/usr/lib/go/bin:/usr/local/sbin:/usr/local/bin:/usr/bin:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_pn #$ INCLUDES="-I/opt/cuda/bin/../targets/x86_64-linux/include" #$ LIBRARIES= "-L/opt/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/opt/cuda/bin/../targets/x86_64-linux/lib" #$ CUDAFE_FLAGS= #$ PTXAS_FLAGS= #$ rm tmp/a_dlink.reg.c #$ gcc -D__CUDA_ARCH_LIST__=520 -E -x c++ -D__CUDACC__ -D__NVCC__ "-I/opt/cuda/bin/../targets/x86_64-linux/include" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=3 -D__CUDACC_VER_BUILD__=52 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=3 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "CMakeCUDACompilerId.cu" -o "tmp/CMakeCUDACompilerId.cpp4.ii" #$ cudafe++ --c++17 --gnu_version=130201 --display_error_number --orig_src_file_name "CMakeCUDACompilerId.cu" --orig_src_path_name "/build/ollama-cuda/src/ollama/llm/llama.cpp/ggml/build/cuda/CMakeFiles/3.27.7/CompilerIdCUDA/CMakeCUDACompilerId.cu" --allow_managed --m64 --parse_templates --gen_c_file_name "tmp/CMakeCUDACompilerId.cudafe1.cpp" --stub_file_name "CMakeCUDACompilerId.cudafe1.stub.c" --gen_module_id_file --module_id_file_name "tmp/CMakeCUDACompilerId.module_id" "tmp/CMakeCUDACompilerId.cpp4.ii" /usr/include/bits/floatn.h(86): error: invalid combination of type specifiers typedef __float128 _Float128; ^ /usr/include/bits/floatn-common.h(214): error: invalid combination of type specifiers typedef float _Float32; ^ /usr/include/bits/floatn-common.h(251): error: invalid combination of type specifiers typedef double _Float64; ^ /usr/include/bits/floatn-common.h(268): error: invalid combination of type specifiers typedef double _Float32x; ^ /usr/include/bits/floatn-common.h(285): error: invalid combination of type specifiers typedef long double _Float64x; ^ 5 errors detected in the compilation of "CMakeCUDACompilerId.cu". # --error 0x2 -- Call Stack (most recent call first): /usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:8 (CMAKE_DETERMINE_COMPILER_ID_BUILD) /usr/share/cmake/Modules/CMakeDetermineCompilerId.cmake:53 (__determine_compiler_id_test) /usr/share/cmake/Modules/CMakeDetermineCUDACompiler.cmake:307 (CMAKE_DETERMINE_COMPILER_ID) CMakeLists.txt:250 (enable_language) -- Configuring incomplete, errors occurred! llm/llama.cpp/generate_linux.go:21: running "cmake": exit status 1 ``` Building the regular `ollama` package, where `cuda`/`nvcc` are not installed before building, works fine. Attaching logs. [ollama-cuda-0.1.8-1-x86_64-build.log](https://github.com/jmorganca/ollama/files/13274355/ollama-cuda-0.1.8-1-x86_64-build.log) [ollama-cuda-0.1.8-1-x86_64-prepare.log](https://github.com/jmorganca/ollama/files/13274356/ollama-cuda-0.1.8-1-x86_64-prepare.log) Advice for how to package ollama + cuda for Arch Linux is welcome.
GiteaMirror commented 2026-04-28 02:40:25 -05:00
Author
Owner
Copy Link

@geekodour commented on GitHub (Nov 7, 2023):

This error is more llama.cpp specific as far as i understand. I was trying to install the same in nixos yesterday and faced similar issues, https://github.com/NixOS/nixpkgs/pull/257760#issuecomment-1794967371

<!-- gh-comment-id:1797992841 --> @geekodour commented on GitHub (Nov 7, 2023): This error is more llama.cpp specific as far as i understand. I was trying to install the same in nixos yesterday and faced similar issues, https://github.com/NixOS/nixpkgs/pull/257760#issuecomment-1794967371
GiteaMirror commented 2026-04-28 02:40:26 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Nov 7, 2023):

I haven't tested this yet, but there is a new llama.cpp commit from 2 days ago that might fix this issue:

2833a6f63c

<!-- gh-comment-id:1798153020 --> @xyproto commented on GitHub (Nov 7, 2023): I haven't tested this yet, but there is a new llama.cpp commit from 2 days ago that might fix this issue: https://github.com/ggerganov/llama.cpp/commit/2833a6f63c1b87c7f4ac574bcf7a15a2f3bf3ede
GiteaMirror commented 2026-04-28 02:40:28 -05:00
Author
Owner
Copy Link

@orkutmuratyilmaz commented on GitHub (Nov 9, 2023):

@xyproto thanks for maintaining this package:)

<!-- gh-comment-id:1803283585 --> @orkutmuratyilmaz commented on GitHub (Nov 9, 2023): @xyproto thanks for maintaining this package:)
GiteaMirror commented 2026-04-28 02:40:29 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Nov 13, 2023):

#1028 did not work out.

Could the errors be related to the cmake defines -DLLAMA_FMA=off or -DLLAMA_F16C=off somehow?

<!-- gh-comment-id:1808940600 --> @xyproto commented on GitHub (Nov 13, 2023): #1028 did not work out. Could the errors be related to the cmake defines `-DLLAMA_FMA=off` or `-DLLAMA_F16C=off` somehow?
GiteaMirror commented 2026-04-28 02:40:30 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Nov 19, 2023):

With the latest version of the Arch Linux packages + the latest version of Ollama, this now works again. 👍

<!-- gh-comment-id:1817858750 --> @xyproto commented on GitHub (Nov 19, 2023): With the latest version of the Arch Linux packages + the latest version of Ollama, this now works again. :+1:
GiteaMirror commented 2026-04-28 02:40:31 -05:00
Author
Owner
Copy Link

@orkutmuratyilmaz commented on GitHub (Nov 19, 2023):

@xyproto thanks for this update. I'd like to use the latest package on my local environment. Can you also update the stable branch too?

Many thanks for Sunday contribution:)

<!-- gh-comment-id:1817912721 --> @orkutmuratyilmaz commented on GitHub (Nov 19, 2023): @xyproto thanks for this update. I'd like to use the latest package on my local environment. Can you also update the [stable branch](https://packages.manjaro.org/?query=ollama) too? Many thanks for Sunday contribution:)
GiteaMirror commented 2026-04-28 02:40:32 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Nov 19, 2023):

Manjaro is a different Linux distribution, so I don't have access to those packages. They will probably update the stable version soon, though.

<!-- gh-comment-id:1817997306 --> @xyproto commented on GitHub (Nov 19, 2023): Manjaro is a different Linux distribution, so I don't have access to those packages. They will probably update the stable version soon, though.
GiteaMirror commented 2026-04-28 02:40:32 -05:00
Author
Owner
Copy Link

@orkutmuratyilmaz commented on GitHub (Nov 25, 2023):

@xyproto thanks for the explanation:)

<!-- gh-comment-id:1826397454 --> @orkutmuratyilmaz commented on GitHub (Nov 25, 2023): @xyproto thanks for the explanation:)
GiteaMirror commented 2026-04-28 02:40:34 -05:00
Author
Owner
Copy Link

@BaconIsAVeg commented on GitHub (Dec 25, 2023):

Hey sorry @xyproto to ping you here, but Arch registrations are down at the moment and there's no issue tracker on your fork. It looks like when you built the ollama-cuda package, the version number was left at 0.0.0. I'm assuming there should be something in the build process that updates that.

This causes the ollama-webui client to refuse connection, as it requires at least 1.1.16. Running the ollama bin manually (downloaded from ollama.ai) works.

<!-- gh-comment-id:1869013563 --> @BaconIsAVeg commented on GitHub (Dec 25, 2023): Hey sorry @xyproto to ping you here, but Arch registrations are down at the moment and there's no issue tracker on your fork. It looks like when you built the ollama-cuda package, the version number was left at 0.0.0. I'm assuming there should be something in the build process that updates that. This causes the ollama-webui client to refuse connection, as it requires at least 1.1.16. Running the ollama bin manually (downloaded from ollama.ai) works.
GiteaMirror commented 2026-04-28 02:40:35 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Dec 25, 2023):

@BaconIsAVeg Ah, yes. The latest tests added to ollama specifically checked if the version number was 0.0.0 before passing. Following the Arch philosophy of patching as little as possible, I removed the line in the PKGBUILD that patched in the correct version number, while figuring out if this really is on purpose or just a fluke.

<!-- gh-comment-id:1869124762 --> @xyproto commented on GitHub (Dec 25, 2023): @BaconIsAVeg Ah, yes. The latest tests added to ollama specifically checked if the version number was 0.0.0 before passing. Following the Arch philosophy of patching as little as possible, I removed the line in the PKGBUILD that patched in the correct version number, while figuring out if this really is on purpose or just a fluke.
GiteaMirror commented 2026-04-28 02:40:36 -05:00
Author
Owner
Copy Link

@xyproto commented on GitHub (Dec 25, 2023):

See also https://github.com/jmorganca/ollama/issues/1712

<!-- gh-comment-id:1869127151 --> @xyproto commented on GitHub (Dec 25, 2023): See also https://github.com/jmorganca/ollama/issues/1712
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#47011
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?