[Solved] cudaGetDevice returned error code 3,

Accelerated Computing CUDA CUDA Setup and Installation
1

I am getting started with CUDA. I have slackware 14.1 64-bit running on a Dell Precision m4400 machine.
I have installed CUDA toolkit 6.0
I have the NVIDIA driver installed : lspci | grep -i nvidia shows
“01:00.0 VGA compatible controller: NVIDIA Corporation G96GLM [Quadro FX 770M] (rev a1)”
This is the 331.20 driver, I am assuming that CUDA 6 works with this driver version.

I am trying to compile and run the example matrixMul from CUDA-Samples/0_Simple. It compiles correctly, but when I run it using ./matrixMul -device=0 -wA=10 -hA=10 -wB=10 -hB=10, I get an error
“cudaGetDevice returned error code 3, line(396)”

Does anyone have an idea on what is going wrong? Any help is most appreciated.

Update:

deviceQuery fails with the following error

"./deviceQuery Starting…

CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 3
→ initialization error
Result = FAIL"

1 Like
2

CUDA 6 on Linux requires a minimum driver version of 331.62

I don’t see slackware listed as a supported distro.

You haven’t indicated how you installed CUDA. I’d suggest starting over with an officially supported distro and using the runfile install method:

[url]CUDA Toolkit Documentation

3

Thank you! That fixed it.
I had 331.20, and now after installing 331.67 it works fine.

Though I could not find where the minimum driver version requirement is mentioned.

4

I infer the minimum driver version from the driver version that is included with the CUDA 6 runfile installer for linux. That particular driver is 331.62. I don’t know for certain that prior driver versions will not work, but I use that as a rule of thumb.