Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CUDNN test failure on CUDA 11.x #2498

Open
maleadt opened this issue Sep 18, 2024 · 0 comments
Open

CUDNN test failure on CUDA 11.x #2498

maleadt opened this issue Sep 18, 2024 · 0 comments
Labels
bug Something isn't working

Comments

@maleadt
Copy link
Member

maleadt commented Sep 18, 2024

https://buildkite.com/julialang/cuda-dot-jl/builds/5412#019205a9-8db7-4962-b141-2413e84ed18b

multiheadattn: Error During Test at /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/runtests.jl:23
Got exception outside of a @test
LoadError: CUDNNError: CUDNN_STATUS_INTERNAL_ERROR (code 4000)
Stacktrace:
  [1] throw_api_error(res::cuDNN.cudnnStatus_t)
    @ cuDNN /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/libcudnn.jl:14
  [2] check
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/libcudnn.jl:27 [inlined]
  [3] cudnnMultiHeadAttnForward
    @ ~/.cache/julia-buildkite-plugin/depots/3cc01fab-3357-4a7a-9294-cde2d3115a97/packages/CUDA/z3j2H/lib/utils/call.jl:34 [inlined]
  [4] cudnnMultiHeadAttnForwardAD(weights::CuArray{Float32, 1, CUDA.DeviceMemory}, queries::CuArray{Float32, 3, CUDA.DeviceMemory}, keys::CuArray{Float32, 3, CUDA.DeviceMemory}, values::CuArray{Float32, 3, CUDA.DeviceMemory}, residuals::Nothing; dready::Base.RefValue{Bool}, dweights::Base.RefValue{Any}, dqueries::Base.RefValue{Any}, dkeys::Base.RefValue{Any}, dvalues::Base.RefValue{Any}, attnDesc::cudnnAttnDescriptor, currIdx::Int64, loWinIdx::Vector{Int32}, hiWinIdx::Vector{Int32}, devSeqLengthsQO::CuArray{Int32, 1, CUDA.DeviceMemory}, devSeqLengthsKV::CuArray{Int32, 1, CUDA.DeviceMemory}, qDesc::cudnnSeqDataDescriptor, kDesc::cudnnSeqDataDescriptor, vDesc::cudnnSeqDataDescriptor, oDesc::cudnnSeqDataDescriptor, out::CuArray{Float32, 3, CUDA.DeviceMemory}, workspace::Nothing, reserveSpace::Nothing)
    @ cuDNN /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/multiheadattn.jl:145
  [5] cudnnMultiHeadAttnForwardAD
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/multiheadattn.jl:129 [inlined]
  [6] cudnnMultiHeadAttnForward!(out::CuArray{Float32, 3, CUDA.DeviceMemory}, weights::CuArray{Float32, 1, CUDA.DeviceMemory}, queries::CuArray{Float32, 3, CUDA.DeviceMemory}, keys::CuArray{Float32, 3, CUDA.DeviceMemory}, values::CuArray{Float32, 3, CUDA.DeviceMemory}, attnDesc::cudnnAttnDescriptor; axes::Vector{cudnnSeqDataAxis_t}, seqLengthsQO::Vector{Int32}, seqLengthsKV::Vector{Int32}, devSeqLengthsQO::CuArray{Int32, 1, CUDA.DeviceMemory}, devSeqLengthsKV::CuArray{Int32, 1, CUDA.DeviceMemory}, qDesc::cudnnSeqDataDescriptor, kDesc::cudnnSeqDataDescriptor, vDesc::cudnnSeqDataDescriptor, oDesc::cudnnSeqDataDescriptor, residuals::Nothing, currIdx::Int64, loWinIdx::Vector{Int32}, hiWinIdx::Vector{Int32}, workspace::Nothing, reserveSpace::Nothing, dweights::Base.RefValue{Any}, dqueries::Base.RefValue{Any}, dkeys::Base.RefValue{Any}, dvalues::Base.RefValue{Any}, o::@Kwargs{attnMode::UInt32, nHeads::Int32, smScaler::Float64, mathType::cudnnMathType_t, qProjSize::Int64, kProjSize::Int64, vProjSize::Int64, oProjSize::Int64, qoMaxSeqLength::Int64, kvMaxSeqLength::Int64, maxBatchSize::Int64, maxBeamSize::Int64})
    @ cuDNN /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/multiheadattn.jl:118
  [7] cudnnMultiHeadAttnForward
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/multiheadattn.jl:55 [inlined]
  [8] cudnnMultiHeadAttnForward(w::CuArray{Float32, 1, CUDA.DeviceMemory}, q::CuArray{Float32, 3, CUDA.DeviceMemory}, k::CuArray{Float32, 3, CUDA.DeviceMemory}, v::CuArray{Float32, 3, CUDA.DeviceMemory}; o::@Kwargs{axes::Vector{cudnnSeqDataAxis_t}, seqLengthsQO::Vector{Int32}, seqLengthsKV::Vector{Int32}, attnMode::UInt32, nHeads::Int32, smScaler::Float64, mathType::cudnnMathType_t, qProjSize::Int64, kProjSize::Int64, vProjSize::Int64, oProjSize::Int64, qoMaxSeqLength::Int64, kvMaxSeqLength::Int64, maxBatchSize::Int64, maxBeamSize::Int64, residuals::Nothing, currIdx::Int64, loWinIdx::Vector{Int32}, hiWinIdx::Vector{Int32}})
    @ cuDNN /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/src/multiheadattn.jl:54
  [9] mhatest(; axes::Vector{cudnnSeqDataAxis_t}, seqLengthsQO::Vector{Int32}, seqLengthsKV::Vector{Int32}, attnMode::UInt32, nHeads::Int32, smScaler::Float64, mathType::cudnnMathType_t, qProjSize::Int64, kProjSize::Int64, vProjSize::Int64, oProjSize::Int64, qoMaxSeqLength::Int64, kvMaxSeqLength::Int64, maxBatchSize::Int64, maxBeamSize::Int64, residuals::Nothing, currIdx::Int64, loWinIdx::Vector{Int32}, hiWinIdx::Vector{Int32})
    @ Main /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/multiheadattn.jl:113
 [10] mhatest()
    @ Main /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/multiheadattn.jl:55
 [11] top-level scope
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/multiheadattn.jl:135
 [12] include(fname::String)
    @ Base.MainInclude ./client.jl:489
 [13] macro expansion
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/runtests.jl:24 [inlined]
 [14] macro expansion
    @ ~/.cache/julia-buildkite-plugin/julia_installs/bin/linux/x64/1.10/julia-1.10-latest-linux-x86_64/share/julia/stdlib/v1.10/Test/src/Test.jl:1577 [inlined]
 [15] top-level scope
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/runtests.jl:24
 [16] eval
    @ ./boot.jl:385 [inlined]
 [17] macro expansion
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/runtests.jl:22 [inlined]
 [18] macro expansion
    @ ~/.cache/julia-buildkite-plugin/julia_installs/bin/linux/x64/1.10/julia-1.10-latest-linux-x86_64/share/julia/stdlib/v1.10/Test/src/Test.jl:1577 [inlined]
 [19] top-level scope
    @ /var/lib/buildkite-agent/builds/gpuci-16/julialang/cuda-dot-jl/lib/cudnn/test/runtests.jl:13
 [20] include(fname::String)
    @ Base.MainInclude ./client.jl:489
 [21] top-level scope
    @ none:6
 [22] eval
    @ ./boot.jl:385 [inlined]
 [23] exec_options(opts::Base.JLOptions)
    @ Base ./client.jl:291
 [24] _start()
    @ Base ./client.jl:552
@maleadt maleadt added the bug Something isn't working label Sep 18, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

1 participant