ci: add linux-x64-cuda-13.3 build (dev + release)#40
Merged
Conversation
Linux until now shipped only the Vulkan backend; NVIDIA users got the dequant path and no access to native CUDA kernels (incl. FP4 on Blackwell). Add a CUDA 13.3 build mirroring the Windows cuda-13.3 variant: GGML_BACKEND_DL keeps it a dlopen'd libggml-cuda.so next to the CPU variants, cudart/cublas are bundled like the Windows DLLs, archs cover A100/RTX30/RTX40/H100/RTX50 (80;86;89;90;120).
…p dotted display name
This was referenced Jul 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Linux currently ships only
linux-x64-vulkan; NVIDIA users never get native CUDA kernels — including native FP4 on Blackwell, which the Vulkan path cannot use (fp4: 0in device caps, verified on a rented RTX 5090).Adds a
linux-x64-cuda-13.3job todev-build.ymlandrelease-turboquant.yml, mirroring the Windowscuda-13.3variant:GGML_BACKEND_DL=ON: CUDA backend is a dlopen'dlibggml-cuda.sonext to the CPU variants — one archive, backend picked at runtimeCMAKE_CUDA_ARCHITECTURES=80;86;89;90;120(A100 / RTX 30 / RTX 40 / H100+H200 / RTX 50)libcudart/libcublas/libcublasLtlike the Windows job bundles the DLLsllama-turboquant-linux-x64-cuda-13.3.{tar.gz,zip}, artifactarchive-linux-x64-cuda-13.3, added to both publish jobs'needsNote: first CI run builds cold (no ccache) — expect the new job to take a while; subsequent runs drop to minutes. The job is intentionally NOT in the required-checks list yet — add it after it proves stable.
After merge the plan is: manually smoke the new artifact on a rented 5090 (native FP4 path), then wire both linux artifacts into the upcoming
backend-smoke.yml.