Wrong NNlib bounds wiith CuArrays 1.0.2

[jamie-arcc]

Describe the bug
using CuArrays fails during precompilation with LoadError: UndefVarError: DenseConvDims not defined

To Reproduce

$ julia
               _
   _       _ _(_)_     |  Documentation: https://docs.julialang.org
  (_)     | (_) (_)    |
   _ _   _| |_  __ _   |  Type "?" for help, "]?" for Pkg help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 1.1.0
 _/ |\__'_|_|_|\__'_|  |
|__/                   |

julia> using CuArrays
[ Info: Precompiling CuArrays [3a865a2d-5b23-5a0f-bc46-62713ec82fae]
WARNING: could not import NNlib.spatial_dims into CUDNN
ERROR: LoadError: LoadError: LoadError: UndefVarError: DenseConvDims not defined
Stacktrace:
 [1] top-level scope at none:0
 [2] include at ./boot.jl:326 [inlined]
 [3] include_relative(::Module, ::String) at ./loading.jl:1038
 [4] include at ./sysimg.jl:29 [inlined]
 [5] include(::String) at /home/jschnaitter/.julia/packages/CuArrays/PwSdF/src/dnn/CUDNN.jl:1
 [6] top-level scope at none:0
 [7] include at ./boot.jl:326 [inlined]
 [8] include_relative(::Module, ::String) at ./loading.jl:1038
 [9] include at ./sysimg.jl:29 [inlined]
 [10] include(::String) at /home/jschnaitter/.julia/packages/CuArrays/PwSdF/src/CuArrays.jl:3
 [11] top-level scope at none:0
 [12] include at ./boot.jl:326 [inlined]
 [13] include_relative(::Module, ::String) at ./loading.jl:1038
 [14] include(::Module, ::String) at ./sysimg.jl:29
 [15] top-level scope at none:2
 [16] eval at ./boot.jl:328 [inlined]
 [17] eval(::Expr) at ./client.jl:404
 [18] top-level scope at ./none:3
in expression starting at /home/jschnaitter/.julia/packages/CuArrays/PwSdF/src/dnn/libcudnn.jl:265
in expression starting at /home/jschnaitter/.julia/packages/CuArrays/PwSdF/src/dnn/CUDNN.jl:35
in expression starting at /home/jschnaitter/.julia/packages/CuArrays/PwSdF/src/CuArrays.jl:52
ERROR: Failed to precompile CuArrays [3a865a2d-5b23-5a0f-bc46-62713ec82fae] to /home/jschnaitter/.julia/compiled/v1.1/CuArrays/7YFE0.ji.
Stacktrace:
 [1] error(::String) at ./error.jl:33
 [2] compilecache(::Base.PkgId, ::String) at ./loading.jl:1197
 [3] _require(::Base.PkgId) at ./loading.jl:960
 [4] require(::Base.PkgId) at ./loading.jl:858
 [5] require(::Module, ::Symbol) at ./loading.jl:853

Expected behavior
Expected the package to load

Build log

(v1.1) pkg> build
  Building ZipFile ─────────→ `~/.julia/packages/ZipFile/YHTbb/deps/build.log`
  Building CodecZlib ───────→ `~/.julia/packages/CodecZlib/9jDi1/deps/build.log`
  Building MbedTLS ─────────→ `~/.julia/packages/MbedTLS/X4xar/deps/build.log`
  Building CUDAdrv ─────────→ `~/.julia/packages/CUDAdrv/3cR2F/deps/build.log`
  Building LLVM ────────────→ `~/.julia/packages/LLVM/tg8MX/deps/build.log`
  Building CUDAnative ──────→ `~/.julia/packages/CUDAnative/wU0tS/deps/build.log`
  Building Conda ───────────→ `~/.julia/packages/Conda/CpuvI/deps/build.log`
  Building FFTW ────────────→ `~/.julia/packages/FFTW/p7sLQ/deps/build.log`
  Building CuArrays ────────→ `~/.julia/packages/CuArrays/PwSdF/deps/build.log`
  Building ZMQ ─────────────→ `~/.julia/packages/ZMQ/ABGOx/deps/build.log`
  Building SpecialFunctions → `~/.julia/packages/SpecialFunctions/fvheQ/deps/build.log`
  Building IJulia ──────────→ `~/.julia/packages/IJulia/gI2uA/deps/build.log`

Environment details (please complete this section)
Details on Julia:

Julia 1.1.0 running on CentOS 7.4 x86_64

julia> versioninfo()
Julia Version 1.1.0
Platform Info:
  OS: Linux (x86_64-redhat-linux)
  CPU: Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, skylake)
Environment:
  LMOD_FAMILY_JULIA_VERSION = julia-1.1.0
  LMOD_FAMILY_JULIA = julia

CUDA: CUDA 10.0 Driver Version: 410.79

Additional context
contents of ~/.julia/packages/CuArrays/deps/ext.jl:

# autogenerated file, do not edit
const libcufft = "/apps/cuda/cuda-10.0/lib64/libcufft.so"
const libcublas = "/apps/cuda/cuda-10.0/lib64/libcublas.so"
const configured = true
const libcusparse = "/apps/cuda/cuda-10.0/lib64/libcusparse.so"
const libcusolver = "/apps/cuda/cuda-10.0/lib64/libcusolver.so"
const libcurand = "/apps/cuda/cuda-10.0/lib64/libcurand.so"
const libcudnn = "/apps/cuda/cuda-10.0/lib64/libcudnn.so"

[ali-ramadhan]

I got around this by running ] add NNlib#master at the package manager. I think someone at NNlib.jl has to tag a new release?

[jamie-arcc]

I got around this by running ] add NNlib#master at the package manager. I think someone at NNlib.jl has to tag a new release?

Sadly, that seems to fail for me too:

(v1.1) pkg> add NNlib#master
  Updating git-repo `https://github.com/FluxML/NNlib.jl.git`
 Resolving package versions...
ERROR: Unsatisfiable requirements detected for package NNlib [872c559c]:
 NNlib [872c559c] log:
 ├─possible versions are: 0.6.0 or uninstalled
 ├─NNlib [872c559c] is fixed to version 0.6.0
 └─found to have no compatible versions left with Flux [587475ba]
   └─Flux [587475ba] log:
     ├─possible versions are: [0.1.0-0.1.1, 0.2.0-0.2.2, 0.3.0-0.3.4, 0.4.0-0.4.1, 0.5.0-0.5.4, 0.6.0-0.6.10, 0.7.0-0.7.3, 0.8.0-0.8.2] or uninstalled
     └─restricted to versions 0.8.2 by an explicit requirement, leaving only versions 0.8.2

[Menee]

I met the same error on Google Colab.
So I guess anyone who want to help solve it could repeat the issue conveniently on Google Colab.

[ali-ramadhan]

@jamie-arcc Ah sorry I'm not using Flux so updating NNlib was enough for me I guess. I'll let someone more experienced than me reply.

[jamie-arcc]

It looks like an issue with NNlib versions between CuArrays and Flux.jl, as mentioned in FluxML/Flux.jl#753

[jamie-arcc]

I was able to get it working by installing [email protected] instead of the current 1.0.2 release

[maleadt]

Should be fixed with JuliaRegistries/General#327 (at least, it should be impossible to install incompatible versions of CuArrays/NNlib now). Flux should be tagged with NNlib 0.6 compatibility.