Matlab stop recognizing K80 after some time of operation.

Run gpuDeviceCount, initially the result is 3 (1 for NVS 315, 2 for K80) and the default gpu device is K80. But after some time, maybe 5 mins, the K80 disappear and cannot be used.
Our system:
Dell Workstation 7910
Xeon® CPU E5-2620 2.4GHz
Windows 7 64 bit operation system.
GPU card: NVS 315, K80
CUDA 7.5

1 件のコメント

Joss Knight
Joss Knight 2015 年 11 月 24 日
It would be helpful if you provide more detail as to what you are doing in those five minutes and exactly what causes the K80 to disappear. If you run nvidia-smi on the command line (C:\Program Files\NVIDIA Corporation\NVSMI\nvidia-smi.exe) is the device still there?

サインインしてコメントする。

回答 (2 件)

Madhu Govindarajan
Madhu Govindarajan 2015 年 11 月 12 日

0 投票

Have you contacted Technical Support from MathWorks with this? http://www.mathworks.com/support/contact_us/
Daniel Moss
Daniel Moss 2017 年 12 月 22 日
編集済み: Daniel Moss 2017 年 12 月 22 日

0 投票

Did you find a solution to this problem....?
I am experiencing the same kind of problem on the Tesla K80 I have two other versions of Matlab successfully running on two of the gpus of the K80 system (7/8 in total)....
But when I run gpuDeviceCount in a freshly opened version of Matlab ...I get an error... looking inside gpuDeviceCount at the error that parallel.gpu.GPUDevice.count gives,
i get:
parallel:gpu:device:UnknownCUDAError
An unexpected error occurred during CUDA execution. The CUDA error was:CUDA_ERROR_OUT_OF_MEMORY

カテゴリ

ヘルプ センター および File ExchangeGPU Computing についてさらに検索

タグ

質問済み:

2015 年 11 月 12 日

編集済み:

2017 年 12 月 22 日

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by