Package Details: ollama-cuda13-bin 0.35.1-1

Git Clone URL: https://aur.archlinux.org/ollama-bin.git (read-only, click to copy)
Package Base: ollama-bin
Description: Create, run and share large language models (LLMs) with CUDA 13
Upstream URL: https://github.com/ollama/ollama
Keywords: ai llm local
Licenses: MIT
Conflicts: ollama-cuda, ollama-cuda12, ollama-vulkan
Provides: ollama-cuda13
Submitter: Dominiquini
Maintainer: Dominiquini (csjewell)
Last Packager: Dominiquini
Votes: 8
Popularity: 0.42
First Submitted: 2025-09-26 06:29 (UTC)
Last Updated: 2026-10-02 22:40 (UTC)

Latest Comments

1 2 3 Next › Last »

Dominiquini commented on 2026-10-05 23:31 (UTC)

@tiziodcaio: I created this package primarily to provide the 'vulkan' version of ollama, as it didn't exist in the 'extra' section for a while. Now that it's been published there, I don't really see much point in keeping this package. I only continue maintaining it because it costs me almost nothing...

tiziodcaio commented on 2026-10-05 17:20 (UTC)

Has it any difference from the package in extra repo? https://archlinux.org/packages/extra/x86_64/ollama/

ticpu commented on 2026-08-14 19:54 (UTC)

I can help in maintaining the latest version if you wish.

csjewell commented on 2026-08-13 14:09 (UTC) (edited on 2026-08-13 14:10 (UTC) by csjewell)

I want to say "Please give Dominiquini (and I, although he's fast enough I don't do much) 48 or 72 hours before marking it out-of-date, please." There's only so much time in a day, after all, and I have even less, plus ollama pushes new builds up FREQUENTLY.

omnigenous commented on 2026-07-29 09:12 (UTC) 

==> Validating source_x86_64 files with sha256sums...
    ollama-x86_64-0.32.4.tzst ... FAILED

kodx commented on 2026-05-23 17:32 (UTC) (edited on 2026-05-23 17:32 (UTC) by kodx)

For CUDA 12, I had to patch the libggml-cuda.so library, otherwise the program couldn't find other libraries in the directory. This should be done after installing the program. Here are the commands:

Installing patchelf

sudo pacman -S patchelf

Patching the library to add $ORIGIN to rpath

sudo patchelf --set-rpath '$ORIGIN' /usr/lib/ollama/libggml-cuda.so

Another option is to add /usr/lib/ollama to ldconfig, but this should be considered harmfull and lead to errors with other sofrware.

This error is in upstream ollama binary libs.

xenoxaos commented on 2026-05-08 00:08 (UTC)

I had to roll back to f450a5df747d to rebuild 0.22.1 in order to get it working on my 1070 ti.

csjewell commented on 2026-05-07 05:11 (UTC)

Looks good, and thank you!

Dominiquini commented on 2026-05-07 01:38 (UTC)

@csjewell: I basically implemented your suggestion, with a few minor modifications. Could you take a look? I'll add you as a co-maintainer, in case you're interested in contributing anything else! Thanks for your help!

csjewell commented on 2026-05-06 15:14 (UTC) (edited on 2026-05-06 15:52 (UTC) by csjewell)

'Drive-by-patching' to stop building mlx-cuda-13 version, and to split what is needed for cuda and vulkan out, is at https://paste.c-net.org/b0eb0a98-a412-2ba5-f0fc-6209365fa761 - @Dominiquini, you may wish to take it, or to do something different.

I use Vulkan with ollama myself is how I know that the OLLAMA_VULKAN variable is needed.

1 2 3 Next › Last »