Pkg test CUDA on Windows 10, Julia 1.5.2

Specific Domains GPU
1

I just added CUDA.jl and ran Pkg test, did not pass. Wonder what the issue is.

Machine info:
Windows 10, GeForce MX150

Julia version info:
Julia Version 1.5.2
Commit 539f3ce943 (2020-09-23 23:17 UTC)
Platform Info:
OS: Windows (x86_64-w64-mingw32)
CPU: Intel(R) Core™ i5-8250U CPU @ 1.60GHz
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-9.0.1 (ORCJIT, skylake)

CUDA versioninfo:
CUDA toolkit 11.0.3, artifact installation
CUDA driver 11.1.0
NVIDIA driver 456.71.0

Libraries:

  • CUBLAS: 11.2.0
  • CURAND: 10.2.1
  • CUFFT: 10.2.1
  • CUSOLVER: 10.6.0
  • CUSPARSE: 11.1.1
  • CUPTI: 13.0.0
  • NVML: 11.0.0+456.71
  • CUDNN: 8.0.3 (for CUDA 11.0.0)
  • CUTENSOR: 1.2.0 (for CUDA 11.0.0)

Toolchain:

  • Julia: 1.5.2
  • LLVM: 9.0.1
  • PTX ISA support: 3.2, 4.0, 4.1, 4.2, 4.3, 5.0, 6.0, 6.1, 6.3, 6.4
  • Device support: sm_35, sm_37, sm_50, sm_52, sm_53, sm_60, sm_61, sm_62, sm_70, sm_72, sm_75

1 device:
0: GeForce MX150 (sm_61, 1.841 GiB / 2.000 GiB available)

Pkg test output in the link below:
Pkg test output

It seems to say that it failed at cublas, execution and exceptions.

Appreciate any input.

2

The cooperative launch test is broken, and fixed in https://github.com/JuliaGPU/CUDA.jl/pull/517

The CUBLAS issues are unknown to me, and seem pretty problematic. It would be interesting to reduce them to isolated failures, because it’s hard to tell exactly which inputs cause failures (and whether the failures are reproducible).

1 Like
3

I tried it on WIndows 10 with just-installed Cuda 11.1 and get test failures in cublas and exceptions:

CUDA toolkit 11.1.1, artifact installation
│ CUDA driver 11.1.0
│ NVIDIA driver 456.81.0
│
│ Libraries:
│ - CUBLAS: 11.3.0
│ - CURAND: 10.2.2
│ - CUFFT: 10.3.0
│ - CUSOLVER: 11.0.1
│ - CUSPARSE: 11.3.0
│ - CUPTI: 14.0.0
│ - NVML: 11.0.0+456.81
│ - CUDNN: 8.0.4 (for CUDA 11.1.0)
│ - CUTENSOR: 1.2.1 (for CUDA 11.1.0)
│
│ Toolchain:
│ - Julia: 1.5.2
│ - LLVM: 9.0.1
│ - PTX ISA support: 3.2, 4.0, 4.1, 4.2, 4.3, 5.0, 6.0, 6.1, 6.3, 6.4
│ - Device support: sm_35, sm_37, sm_50, sm_52, sm_53, sm_60, sm_61, sm_62, sm_70, sm_72, sm_75
│
│ 1 device:
└ 0: GeForce 940MX (sm_50, 1.875 GiB / 2.000 GiB available)

Maybe this has to do with an absent feature in my GeForce940M card?

cublas: Error During Test at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\cublas.jl:1250
Got exception outside of a @test
CUBLASError: an absent device architectural feature is required (code 8, CUBLAS_STATUS_ARCH_MISMATCH)
Stacktrace:
[1] throw_api_error(::CUDA.CUBLAS.cublasStatus_t) at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\error.jl:47
[2] macro expansion at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\error.jl:58 [inlined]
[3] cublasGemmEx(::Ptr{Nothing}, ::Char, ::Char, ::Int64, ::Int64, ::Int64, ::Base.RefValue{Float16}, ::CuArray{Float16,2}, ::Type{T} where T, ::Int64, ::CuArray{Float16,2}, ::Type{T} where T, ::Int64, ::Base.RefValue{Float16}, ::CuArray{Float16,2}, ::Type{T} where T, ::Int64, ::CUDA.CUBLAS.cublasComputeType_t, ::CUDA.CUBLAS.cublasGemmAlgo_t) at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\utils\call.jl:93
[4] gemmEx!(::Char, ::Char, ::Number, ::Union{CuArray{T,2}, CuArray{T,1}} where T, ::Union{CuArray{T,2}, CuArray{T,1}} where T, ::Number, ::Union{CuArray{T,2}, CuArray{T,1}} where T; algo::CUDA.CUBLAS.cublasGemmAlgo_t) at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\wrappers.jl:836
[5] gemmEx! at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\wrappers.jl:818 [inlined]
[6] gemm_dispatch!(::CuArray{Float16,2}, ::CuArray{Float16,2}, ::CuArray{Float16,2}, ::Bool, ::Bool) at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\linalg.jl:216
[7] mul! at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\lib\cublas\linalg.jl:227 [inlined]
[8] mul!(::CuArray{Float16,2}, ::CuArray{Float16,2}, ::CuArray{Float16,2}) at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\LinearAlgebra\src\matmul.jl:208
[9] top-level scope at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\cublas.jl:1275
[10] top-level scope at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Test\src\Test.jl:1115
[11] top-level scope at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\cublas.jl:1251
[12] top-level scope at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Test\src\Test.jl:1115
[13] top-level scope at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\cublas.jl:438
[14] include(::String) at .\client.jl:457
[15] #9 at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\runtests.jl:78 [inlined]
[16] macro expansion at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\setup.jl:47 [inlined]
[17] macro expansion at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Test\src\Test.jl:1115 [inlined]
[18] macro expansion at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\setup.jl:47 [inlined]
[19] macro expansion at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\src\utilities.jl:35 [inlined]
[20] macro expansion at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\src\pool.jl:564 [inlined]
[21] top-level scope at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\setup.jl:46
[22] eval at .\boot.jl:331 [inlined]
[23] runtests(::Function, ::String, ::Symbol, ::Nothing) at C:\Users\pi96doc.julia\packages\CUDA\YeS8q\test\setup.jl:58
[24] (::Distributed.var"#106#108"{Distributed.CallMsg{:call_fetch}})() at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Distributed\src\process_messages.jl:294
[25] run_work_thunk(::Distributed.var"#106#108"{Distributed.CallMsg{:call_fetch}}, ::Bool) at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Distributed\src\process_messages.jl:79
[26] macro expansion at D:\buildbot\worker\package_win64\build\usr\share\julia\stdlib\v1.5\Distributed\src\process_messages.jl:294 [inlined]
[27] (::Distributed.var"#105#107"{Distributed.CallMsg{:call_fetch},Distributed.MsgHeader,Sockets.TCPSocket})() at .\task.jl:356

1 Like