Error registering a new package using CUDA

[RXGottlieb]

I recently attempted to register a new package here with CUDA.jl as the only dependency, but it fails during AutoMerge here with the error: "ERROR: LoadError: CUDA driver not functional".

I am able to import my package without error on one machine:

julia> versioninfo()
Julia Version 1.12.3
Commit 966d0af0fd (2025-12-15 11:20 UTC)
Build Info:
  Official https://julialang.org release
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 12 × Intel(R) Core(TM) i7-9850H CPU @ 2.60GHz
  WORD_SIZE: 64
  LLVM: libLLVM-18.1.7 (ORCJIT, skylake)
  GC: Built with stock GC
Threads: 4 default, 1 interactive, 4 GC (on 12 virtual cores)
Environment:
  JULIA_EDITOR = code
  JULIA_VSCODE_REPL = 1
  JULIA_NUM_THREADS = 4

julia> CUDA.versioninfo()
CUDA toolchain:
- runtime 13.0, artifact installation
- driver 580.92.0 for 13.0
- compiler 13.1

CUDA libraries:
- CUBLAS: 13.1.0
- CURAND: 10.4.0
- CUFFT: 12.0.0
- CUSOLVER: 12.0.4
- CUSPARSE: 12.6.3
- CUPTI: 2025.3.1 (API 13.0.1)
- NVML: 13.0.0+580.92

Julia packages:
- CUDA: 5.9.5
- CUDA_Driver_jll: 13.1.0+0
- CUDA_Compiler_jll: 0.3.0+1
- CUDA_Runtime_jll: 0.19.2+0

Toolchain:
- Julia: 1.12.3
- LLVM: 18.1.7

1 device:
  0: Quadro T2000 (sm_75, 2.837 GiB / 4.000 GiB available)

But the package does not import on this other machine:

julia> versioninfo()
Julia Version 1.10.8
Commit 4c16ff44be (2025-01-22 10:06 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 12 × AMD Ryzen 5 6600H with Radeon Graphics
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, znver3)
Threads: 1 default, 0 interactive, 1 GC (on 12 virtual cores)
Environment:
  JULIA_EDITOR = code
  JULIA_VSCODE_REPL = 1
  JULIA_NUM_THREADS = 0
 
julia> CUDA.versioninfo()
CUDA toolchain:
- runtime 13.1, artifact installation
- driver 591.59.0 for 13.1
- compiler 13.1
 
CUDA libraries:
- CUBLAS: 13.1.0
- CURAND: 10.4.0
- CUFFT: 12.0.0
- CUSOLVER: 12.0.4
- CUSPARSE: 12.6.3
- CUPTI: 2025.3.1 (API 13.0.1)
- NVML: 13.0.0+591.59
 
Julia packages:
- CUDA: 5.9.5
- CUDA_Driver_jll: 13.1.0+0
- CUDA_Compiler_jll: 0.3.0+1
- CUDA_Runtime_jll: 0.19.2+0
 
Toolchain:
- Julia: 1.10.8
- LLVM: 15.0.7
 
1 device:
  0: NVIDIA GeForce RTX 3050 Laptop GPU (sm_86, 2.688 GiB / 4.000 GiB available)

This second machine instead gives the error:

julia> import BatchPDLP
┌ Error: You are using CUDA 13.1.0, but CUDA.jl was precompiled for CUDA 13.0.0.
│ This is unexpected; please file an issue.
└ @ CUDA C:\Users\Robert\.julia\packages\CUDA\x8d2s\src\initialization.jl:148

I think this is because GitHub and my 2nd machine are using CUDA 13.1, as it worked on my machine with CUDA 13.0. Is there a way to force GitHub to use a specific CUDA Toolkit version, or is there another solution to this?