Matlab OpenCL toolbox max work group size

3 views (last 30 days)
mcg on 10 Sep 2018
Question for anyone that has used the opencl-toolbox from the file exchange. The demo code, which worked out of the box for me, adds two vectors of floats together. The issue I'm running into is that if you send vectors into the kernel that are larger than the max work group size of the device (in my case, 256 for an AMD 8790M), the function does not work (it doesn't crash, but the output is unmodified). How can I send very large arrays to the GPU using this toolbox as a baseline? The data I want to send are 3xn, where n can get very large (1e6+). I am not proficient in C++, so having to start from there would be a much larger effort for me.

Answers (0)




Community Treasure Hunt

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

Start Hunting!