Skip to content

Allow CUDA.jl v6, bump to v6.1.2#1082

Merged
ChrisRackauckas merged 1 commit into
SciML:masterfrom
lamdor:la/cuda-compat-6
Jul 2, 2026
Merged

Allow CUDA.jl v6, bump to v6.1.2#1082
ChrisRackauckas merged 1 commit into
SciML:masterfrom
lamdor:la/cuda-compat-6

Conversation

@lamdor

@lamdor lamdor commented Jul 2, 2026

Copy link
Copy Markdown
Contributor

Summary

Widen the CUDA compat from "5.5.2" to "6" so the package works with CUDA.jl v6.x. The V100 GPU CI runners run nvidia driver 580, which requires CUDA.jl v6.2+ for correct CUDA runtime selection — versions below 6.2 select the runtime based on the driver version and pick an incompatible runtime (JuliaGPU/CUDA.jl#3134, issue #3079). Compat "6" lets the resolver pick the latest 6.x with the fix.

Changes

  • Bump CUDA compat from "5.5.2" to "6".
  • Bump the package version from 6.1.1 to 6.1.2 for the widened compat.

Note: Downgrade.yml will now test against the CUDA 6.0 floor.

The package pinned CUDA.jl to "5.5.2", but the V100 GPU CI runners now
use nvidia driver 580, which requires CUDA.jl v6.2+ for correct CUDA
runtime selection (CUDA.jl versions <6.2 select the runtime based on the
driver version and pick an incompatible runtime; see JuliaGPU/CUDA.jl#3134).
Widen the CUDA compat to "6" so the resolver picks the latest 6.x with
the fix, and bump the package version to 6.1.2 for the widened compat.
@lamdor
lamdor requested a review from ChrisRackauckas July 2, 2026 01:07
@lamdor
lamdor marked this pull request as ready for review July 2, 2026 01:08
@ChrisRackauckas
ChrisRackauckas merged commit c358df0 into SciML:master Jul 2, 2026
25 of 35 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants