# How to make CUDA.jl work in WSL2

**URL:** <https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474>\
**Category:** GPU\
**Tags:** wsl2\
**Created:** [July 3, 2020, 1:08pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474 "2020-07-03T13:08:44Z")\
**Posts on this page:** 14\
**Page:** 1

<div class="post-metadata">

**Author:** ![zwwi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zwwi/32/11932_2.png) [@zwwi](https://discourse.julialang.org/u/zwwi)\
**Post date:** [July 3, 2020, 1:08pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/1 "2020-07-03T13:08:45Z")

</div>

I am using the WSL2 (Ubuntu) with version `4.19.121-microsoft-standard`, and have installed the CUDA driver provided here: [NVIDIA Drivers for CUDA on WSL](https://developer.nvidia.com/cuda/wsl).  
I also have installed `nvidia-cuda-toolkit`.  
I’m curious to know how I can make CUDA.jl work under this circumstance.

When I `test CUDA`, I got the error message:

```julia
Error: Could not initialize CUDA
│ exception =
│ CUDA error (code 100, CUDA_ERROR_NO_DEVICE)

```

Thank you in advance.

---

<div class="post-metadata">

**Author:** ![maleadt](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/maleadt/32/10097_2.png) [@maleadt](https://discourse.julialang.org/u/maleadt)\
**Post date:** [July 3, 2020, 2:17pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/2 "2020-07-03T14:17:26Z")

</div>

I take it you have a GPU? Can you use `nvidia-smi` or such in WSL2? This is basically a CUDA set-up issue, so try and make sure CUDA itself works first (e.g. by compiling and executing a CUDA C binary).

---

<div class="post-metadata">

**Author:** ![EHBaozi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ehbaozi/32/14854_2.png) [@EHBaozi](https://discourse.julialang.org/u/EHBaozi)\
**Post date:** [July 3, 2020, 2:43pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/3 "2020-07-03T14:43:24Z")

</div>

Are you on Windows Insiders?

---

<div class="post-metadata">

**Author:** ![Keno](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/keno/32/285_2.png) [@Keno](https://discourse.julialang.org/u/Keno)\
**Post date:** [July 3, 2020, 9:18pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/4 "2020-07-03T21:18:50Z")

</div>

Also note that WSL needs a different libcuda despite being linux, since it uses the passthrough drivers. Not sure how that interacts with CUDA.jl.

---

<div class="post-metadata">

**Author:** ![zwwi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zwwi/32/11932_2.png) [@zwwi](https://discourse.julialang.org/u/zwwi)\
**Post date:** [July 4, 2020, 6:00am UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/5 "2020-07-04T06:00:27Z")

</div>

Thank you all for your replies.  
Yes, I’m in the fast ring of the windows insiders program. Running `nvidia-smi` in wsl2-ubuntu gives

```julia
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 455.41 Driver Version: 455.41 CUDA Version: 11.1 |
|-------------------------------+----------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1060 WDDM | 00000000:02:00.0 Off | N/A |
| N/A 32C P8 3W / N/A | 90MiB / 6144MiB | 0% Default |
+-------------------------------+----------------------+----------------------+

```

This is indeed a cuda set-up issue:  
I could compile and execute a simple cu code, but the CUDA kernel function somehow did nothing…

EDIT:  
I now could compile and execute cu codes successfully by, e.g.,  
`nvcc -gencode arch=compute_60,code=sm_60 add.cu -o add_cuda`  
The key is to have the `-gencode` option. This shows that CUDA works at my wsl2 ubuntu.

---

<div class="post-metadata">

**Author:** ![zwwi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zwwi/32/11932_2.png) [@zwwi](https://discourse.julialang.org/u/zwwi)\
**Post date:** [July 6, 2020, 10:26am UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/6 "2020-07-06T10:26:04Z")

</div>

The issue was solved by adding

```julia
export LD_LIBRARY_PATH="/usr/lib/wsl/lib/:$LD_LIBRARY_PATH"

```

to the `.zshrc` file.

---

<div class="post-metadata">

**Author:** ![ozmaden](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ozmaden/32/14438_2.png) [@ozmaden](https://discourse.julialang.org/u/ozmaden)\
**Post date:** [August 3, 2020, 2:15pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/7 "2020-08-03T14:15:15Z")

</div>

I too am having dificulties with running CUDA.jl under WSL 2.

> [@zwwi](#):
>
> `export LD_LIBRARY_PATH="/usr/lib/wsl/lib/:$LD_LIBRARY_PATH"`

This has solved the problem of not detecting the core libraries I guess. Since then I get a `true` response from `CUDA.functional()`.

However any basic operation like the one below results in:

```julia
julia> a = CuArray{Float32}(undef, 10)
10-element CuArray{Float32,1}:

julia> b = CuArray{Float32}(undef, 10)
10-element CuArray{Float32,1}:

julia> a + b
ERROR: CUDA error: PTX JIT compiler library not found (code 221, ERROR_JIT_COMPILER_NOT_FOUND)
Stacktrace:
 [1] throw_api_error(::CUDA.cudaError_enum) at /home/coz/.julia/packages/CUDA/7vLVC/lib/cudadrv/error.jl:103
 [2] CuModule(::String, ::Dict{CUDA.CUjit_option_enum,Any}) at /home/coz/.julia/packages/CUDA/7vLVC/lib/cudadrv/module.jl:42
 [3] _cufunction(::GPUCompiler.FunctionSpec{GPUArrays.var"#broadcast_kernel#20",Tuple{CUDA.CuKernelContext,CuDeviceArray{Float32,1,CUDA.AS.Global},Base.Broadcast.Broadcasted{Nothing,Tuple{Base.OneTo{Int64}},typeof(+),Tuple{Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}},Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}}}},Int64}}; kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at /home/coz/.julia/packages/CUDA/7vLVC/src/compiler/execution.jl:337
 [4] _cufunction at /home/coz/.julia/packages/CUDA/7vLVC/src/compiler/execution.jl:304 [inlined]
 [5] check_cache(::typeof(CUDA._cufunction), ::GPUCompiler.FunctionSpec{GPUArrays.var"#broadcast_kernel#20",Tuple{CUDA.CuKernelContext,CuDeviceArray{Float32,1,CUDA.AS.Global},Base.Broadcast.Broadcasted{Nothing,Tuple{Base.OneTo{Int64}},typeof(+),Tuple{Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}},Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}}}},Int64}}, ::UInt64; kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at /home/coz/.julia/packages/GPUCompiler/pCBTA/src/cache.jl:24
 [6] broadcast_kernel at /home/coz/.julia/packages/GPUArrays/PkHCM/src/host/broadcast.jl:60 [inlined]
 [7] cached_compilation(::typeof(CUDA._cufunction), ::GPUCompiler.FunctionSpec{GPUArrays.var"#broadcast_kernel#20",Tuple{CUDA.CuKernelContext,CuDeviceArray{Float32,1,CUDA.AS.Global},Base.Broadcast.Broadcasted{Nothing,Tuple{Base.OneTo{Int64}},typeof(+),Tuple{Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}},Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}}}},Int64}}, ::UInt64; kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at /home/coz/.julia/packages/GPUCompiler/pCBTA/src/cache.jl:0
 [8] cached_compilation at /home/coz/.julia/packages/GPUCompiler/pCBTA/src/cache.jl:44 [inlined]
 [9] cufunction(::GPUArrays.var"#broadcast_kernel#20", ::Type{Tuple{CUDA.CuKernelContext,CuDeviceArray{Float32,1,CUDA.AS.Global},Base.Broadcast.Broadcasted{Nothing,Tuple{Base.OneTo{Int64}},typeof(+),Tuple{Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}},Base.Broadcast.Extruded{CuDeviceArray{Float32,1,CUDA.AS.Global},Tuple{Bool},Tuple{Int64}}}},Int64}}; name::Nothing, kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at /home/coz/.julia/packages/CUDA/7vLVC/src/compiler/execution.jl:298
 [10] cufunction at /home/coz/.julia/packages/CUDA/7vLVC/src/compiler/execution.jl:293 [inlined]
 [11] #launch_heuristic#816 at /home/coz/.julia/packages/CUDA/7vLVC/src/gpuarrays.jl:19 [inlined]
 [12] launch_heuristic at /home/coz/.julia/packages/CUDA/7vLVC/src/gpuarrays.jl:17 [inlined]
 [13] copyto! at /home/coz/.julia/packages/GPUArrays/PkHCM/src/host/broadcast.jl:66 [inlined]
 [14] copyto!(::CuArray{Float32,1}, ::Base.Broadcast.Broadcasted{CUDA.CuArrayStyle{1},Tuple{Base.OneTo{Int64}},typeof(+),Tuple{CuArray{Float32,1},CuArray{Float32,1}}}) at ./broadcast.jl:864
 [15] copy(::Base.Broadcast.Broadcasted{CUDA.CuArrayStyle{1},Tuple{Base.OneTo{Int64}},typeof(+),Tuple{CuArray{Float32,1},CuArray{Float32,1}}}) at ./broadcast.jl:840
 [16] materialize at ./broadcast.jl:820 [inlined]
 [17] broadcast_preserving_zero_d at ./broadcast.jl:809 [inlined]
 [18] +(::CuArray{Float32,1}, ::CuArray{Float32,1}) at ./arraymath.jl:39
 [19] top-level scope at REPL[7]:1

```

`] test CUDA` also fails with a gazillion of outputs that overflow my terminal.

```julia
┌ Info: System information:
│ CUDA toolkit 11.0.194, artifact installation
│ CUDA driver 11.1.0
│
│ Libraries:
│ - CUBLAS: 11.1.0
│ - CURAND: 10.2.1
│ - CUFFT: 10.2.0
│ - CUSOLVER: 10.5.0
│ - CUSPARSE: 11.1.0
│ - CUPTI: 13.0.0
│ - NVML: missing
│ - CUDNN: 8.0.1 (for CUDA 11.0.0)
│ - CUTENSOR: 1.2.0 (for CUDA 11.0.0)
│
│ Toolchain:
│ - Julia: 1.4.2
│ - LLVM: 8.0.1
│ - PTX ISA support: 3.2, 4.0, 4.1, 4.2, 4.3, 5.0, 6.0, 6.1, 6.3
│ - 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(s):
└ - GeForce GTX 1050 Ti (sm_61, 3.297 GiB / 4.000 GiB available)
[ Info: Testing using 1 device(s): 1. GeForce GTX 1050 Ti (UUID a2d155cc-31af-3125-b9b2-bb5b43712f8d)
[ Info: Skipping the following tests: nvml, cutensor, device/wmma
                                     | | ---------------- GPU ---------------- | ---------------- CPU ---------------- |
Test (Worker) | Time (s) | GC (s) | GC % | Alloc (MB) | RSS (MB) | GC (s) | GC % | Alloc (MB) | RSS (MB) |
initialization (2) | 3.27 | 0.00 | 0.0 | 0.00 | N/A | 0.06 | 1.9 | 219.96 | 834.47 |
apiutils (2) | 0.27 | 0.00 | 0.0 | 0.00 | N/A | 0.00 | 0.0 | 5.74 | 834.88 |
array (2) | failed at 2020-08-03T16:10:57.567
.
.
.

```

I have tried Ubuntu 18.04 and 20.04, with julia v1.4.2 and v1.5.0. The Nvidia BlackScholes sample from [CUDA on WSL](https://docs.nvidia.com/cuda/wsl-user-guide/index.html#getting-started) docs compiles and runs fine.

Any idea or glipmse of hope?

---

<div class="post-metadata">

**Author:** ![zwwi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zwwi/32/11932_2.png) [@zwwi](https://discourse.julialang.org/u/zwwi)\
**Post date:** [August 3, 2020, 3:02pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/8 "2020-08-03T15:02:40Z")

</div>

> [@ozmaden](#):
>
> `ERROR: CUDA error: PTX JIT compiler library not found`

The [known-limitations](https://docs.nvidia.com/cuda/wsl-user-guide/index.html#known-limitations) page says

> PTX JIT is not supported (so PTX code will not be loaded from CUDA binaries for runtime compilation)

---

<div class="post-metadata">

**Author:** ![ozmaden](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ozmaden/32/14438_2.png) [@ozmaden](https://discourse.julialang.org/u/ozmaden)\
**Post date:** [August 3, 2020, 4:20pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/9 "2020-08-03T16:20:48Z")

</div>

So no CUDA.jl fun on WSL for the foreseeable future? Or did you manage to overcome this somehow?

---

<div class="post-metadata">

**Author:** ![zwwi](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zwwi/32/11932_2.png) [@zwwi](https://discourse.julialang.org/u/zwwi)\
**Post date:** [August 3, 2020, 4:44pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/10 "2020-08-03T16:44:50Z")

</div>

no…

---

<div class="post-metadata">

**Author:** ![airheadr100](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@airheadr100](https://discourse.julialang.org/u/airheadr100)\
**Post date:** [February 17, 2021, 3:13pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/11 "2021-02-17T15:13:55Z")

</div>

Hi guys - there may be an avenue that might be worth investigating. I just tested running the official Julia Docker image with gpu:  
`sudo docker run --gpus all -it --rm julia`  
having followed (approximately) the guide from here:  
[CUDA on WSL docs](https://docs.nvidia.com/cuda/wsl-user-guide/index.html#installing-nvidia-drivers)  
The installation of the nvidia-docker2 being the key part. I did it in Arch Linux on WSL2 (AUR package nvidia-docker is the equivalent)  
first run - did little test:

```julia
julia> x_d = CUDA.fill(1.0f0, 2^10);
Downloading artifact: CUDA111
Downloading artifact: CUDNN_CUDA111
Downloading artifact: CUTENSOR_CUDA111

julia> y_d = CUDA.fill(2.0f0, 2^10);
julia> y_d .+= x_d;
julia> @test all(Array(y_d) .== 3.0f0)
Test Passed

```

the CUDA111 artifact download took forever for some reason, but it worked. I’m a total newb to Docker (and julia too!) but I just thought I’d throw it out here. And to be clear I’m running the docker instance all within linux on WSL.

snippet of `test CUDA` output

```julia
gpuarrays/mapreduce derivatives: Error During Test at none:1
  Test threw exception
  Expression: gpuarrays/mapreduce derivatives
  On worker 58:
  NVMLError: Unknown Error (code 999)
  throw_api_error at /root/.julia/packages/CUDA/wTQsK/lib/nvml/error.jl:22
  compute_processes at /root/.julia/packages/CUDA/wTQsK/lib/nvml/device.jl:124
  runtests at /root/.julia/packages/CUDA/wTQsK/test/setup.jl:70
  #106 at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.5/Distributed/src/process_messages.jl:294
  run_work_thunk at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.5/Distributed/src/process_messages.jl:79
  macro expansion at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.5/Distributed/src/process_messages.jl:294 [inlined]
  #105 at ./task.jl:356

Test Summary: | Pass Error Total
  Overall | 70 58 128
    initialization | 1 1
    apiutils | 1 1
    array | 1 1
    broadcast | 1 1
    codegen | 8 2 10
    cublas | 1 1
    cudnn | 1 1
    cufft | 1 1
    curand | 1 1
    cusolver | 1 1
    cusparse | 1 1
    examples | 1 1
    exceptions | 1 1
    execution | 56 1 57
    forwarddiff | 1 1
    iterator | 1 1
    nnlib | 1 1
    nvml | 6 1 7
    nvtx | 1 1

```

---

<div class="post-metadata">

**Author:** ![maleadt](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/maleadt/32/10097_2.png) [@maleadt](https://discourse.julialang.org/u/maleadt)\
**Post date:** [February 17, 2021, 3:21pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/12 "2021-02-17T15:21:49Z")

</div>

CUDA JIT is supported under WSL2 nowadays so you shouldn’t require a container.

---

<div class="post-metadata">

**Author:** ![airheadr100](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@airheadr100](https://discourse.julialang.org/u/airheadr100)\
**Post date:** [February 17, 2021, 3:55pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/13 "2021-02-17T15:55:02Z")

</div>

Thanks - good to know - though I had errors just trying within WSL itself (preferred path), I guess I’ll look again…

---

<div class="post-metadata">

**Author:** ![airheadr100](https://avatars.discourse-cdn.com/v4/letter/a/ed8c4c/32.png) [@airheadr100](https://discourse.julialang.org/u/airheadr100)\
**Post date:** [February 21, 2021, 2:30pm UTC](https://discourse.julialang.org/t/how-to-make-cuda-jl-work-in-wsl2/42474/14 "2021-02-21T14:30:34Z")

</div>

I tried again but failed to use CUDA on WSL/Arch (which is not really a supported WSL distro by Windows or Arch even)

So I installed Ubuntu and everything works so far!