# CUDA device support

**URL:** <https://discourse.julialang.org/t/cuda-device-support/41161>\
**Category:** GPU\
**Created:** [June 10, 2020, 8:52pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161 "2020-06-10T20:52:42Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![martincornejo](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/martincornejo/32/13436_2.png) [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Post date:** [June 10, 2020, 8:52pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/1 "2020-06-10T20:52:42Z")

</div>

I have an old [Nvidia 610M](https://www.geforce.com/hardware/notebook-gpus/geforce-610m) GPU on my laptop. It is not a powerful GPU but I wanted to play around and learn a bit of GPU computing. Although the GPU supports CUDA (I have version 9.1.112), I seem to encounter some problems.

When testing the CUDA package, the following error is thrown:

```julia
(@v1.4) pkg> test CUDA
    Testing CUDA
...
ERROR: LoadError: Could not find any suitable device for this configuration

```

And when following the tutorial for CUDA.jl:

```julia
julia> N = 2^20
1048576

julia> x_d = CUDA.fill(1.0f0, N) # a vector stored on the GPU filled with 1.0 (Float32)
1048576-element CuArray{Float32,1,Nothing}:
 1.0
 1.0
 1.0
 1.0
 1.0
 1.0
 ⋮
 1.0
 1.0
 1.0
 1.0
 1.0

julia> y_d = CUDA.fill(2.0f0, N)
1048576-element CuArray{Float32,1,Nothing}:
 2.0
 2.0
 2.0
 2.0
 2.0
 2.0
 ⋮
 2.0
 2.0
 2.0
 2.0
 2.0

julia> y_d .+= x_d
ERROR: Device capability v2.1.0 not supported by available toolchain

```

Is that because my GPU is not supported, or my CUDA version? Is there a list of supported GPUs?

---

<div class="post-metadata">

**Author:** ![jling](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/jling/32/212909_2.png) [@jling](https://discourse.julialang.org/u/jling)\
**Post date:** [June 10, 2020, 9:29pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/2 "2020-06-10T21:29:47Z")

</div>

> [@martincornejo](#):
>
> [Nvidia 610M](https://www.geforce.com/hardware/notebook-gpus/geforce-610m)

[https://developer.nvidia.com/cuda-gpus](https://developer.nvidia.com/cuda-gpus) CUDA 2.1 is wayyy too old

Edit: this is capability not compatibility, for 610m, it supports till CUDA 8

---

<div class="post-metadata">

**Author:** ![martincornejo](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/martincornejo/32/13436_2.png) [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Post date:** [June 10, 2020, 9:36pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/3 "2020-06-10T21:36:51Z")

</div>

![Capture](https://global.discourse-cdn.com/julialang/original/3X/a/4/a48d678c6a9ea1c192980ed5d73a7ed95afc6616.png)

---

<div class="post-metadata">

**Author:** ![jling](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/jling/32/212909_2.png) [@jling](https://discourse.julialang.org/u/jling)\
**Post date:** [June 10, 2020, 9:37pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/4 "2020-06-10T21:37:21Z")

</div>

you can install DLL all you want. Doesn’t mean your device support it

> **[Nvidia GeForce 610m mobile, could not install cuda toolkit](https://forums.developer.nvidia.com/t/nvidia-geforce-610m-mobile-could-not-install-cuda-toolkit/66076)**
>
> I was see, graphic card support cuda, but i can’t install, because installer say driver can’t support. But drivers is last, 391.35 https://www.nvidia.com/Download/driverResults.aspx/132841/en-us What can I do, how to work around?

---

<div class="post-metadata">

**Author:** ![martincornejo](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/martincornejo/32/13436_2.png) [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Post date:** [June 10, 2020, 9:40pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/5 "2020-06-10T21:40:41Z")

</div>

Interesting, I didn’t install v9 myself. Why would Nvidia choose then these drivers? Might installing v8 help? Or is it also unsupported?

---

<div class="post-metadata">

**Author:** ![martincornejo](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/martincornejo/32/13436_2.png) [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Post date:** [June 10, 2020, 10:30pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/6 "2020-06-10T22:30:50Z")

</div>

It seems CUDA.jl supports only CUDA v9 and above.

```julia
julia> x_d = CUDA.fill(1.0f0, N)
┌ Warning: CUDA.jl only supports NVIDIA drivers for CUDA 9.0 or higher (yours is for CUDA 8.0.0)

```

---

<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:** [June 11, 2020, 12:43pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/7 "2020-06-11T12:43:49Z")

</div>

If you upgrade your driver to have CUDA capability 9.0 or higher, CUDA.jl should automatically download a matching version of the toolkit. Although your screenshot above already mentions CUDA 9.1; could you try running with `JULIA_DEBUG=CUDA`? What does `CUDA.version()` show?

---

<div class="post-metadata">

**Author:** ![martincornejo](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/martincornejo/32/13436_2.png) [@martincornejo](https://discourse.julialang.org/u/martincornejo)\
**Post date:** [June 11, 2020, 1:42pm UTC](https://discourse.julialang.org/t/cuda-device-support/41161/8 "2020-06-11T13:42:15Z")

</div>

I installed CUDA 8.0 to test it, now I tried to install back a higher version but NVIDIA assigned and installed again CUDA 8.0. I don’t know why or how I had CUDA 9.1 installed since it shouldn’t support Fermi GPUs.

```julia
julia> CUDA.version()
┌ Warning: CUDA.jl only supports NVIDIA drivers for CUDA 9.0 or higher (yours is for CUDA 8.0.0)
└ @ CUDA C:\Users\Martin Cornejo\.julia\packages\CUDA\5t6R9\src\initialization.jl:109
ERROR: Could not find a suitable CUDA installation

```

I should have tried this before changing the drivers…