How to utilize power of GPU computing in Matlab

Question

Atinesh Singh 2017-5-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/341197-how-to-utilize-power-of-gpu-computing-in-matlab

回答： Joss Knight 2017-5-22

My research work is in the field of Evolutionary computation, Sometimes I need to run the algorithm on high dimensional data (dimension may go beyond 1000). For such high dimensional problems, sometimes it takes days for me to collect the data. I have heard of Matlab's "Parallel computing toolbox" which distributes computation over GPU and other multiple clusters and speeds up the computation process. I checked the documentation page it says it requires "NVIDIA graphics card with compute capability 1.3", But I have AMD Radeon graphics card. Is it possible for me use my GPU computing power in Matlab computation.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2017-5-21

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/341197-how-to-utilize-power-of-gpu-computing-in-matlab#answer_267803

Sorry, there is no support for AMD GPU or Intel GPU, only for NVidia GPU.

Apparently it turns out that different vendors and different models implement OpenCL differently enough that Mathworks would have to be able to generate a number of different variations depending on exactly what was being targeted. It is not clear that would be worth doing until OpenCL implementations become more standardized.

2 个评论
显示无隐藏无

Atinesh Singh 2017-5-21

编辑：Atinesh Singh 2017-5-21

Can I use Matlab's "Distributed Computing Server". On the documentation page, it says it requires "Parallel computing toolbox".

Walter Roberson 2017-5-21

DCS requires that you have access to a cluster.

If you have access to a machine with multiple cores, then you can use the Parallel Computing Toolbox to access multiple CPU cores, without using the DCS. The NVidia restriction is for GPU use, not CPU use.

See also https://www.mathworks.com/products/parallel-computing/parallel-computing-on-the-cloud.html

请先登录，再进行评论。