why GPUarray is slower than CPU?

Hi. GPU vs CPU speed check. why GPUarray is slower than CPU? GPU : GTX 1080, CPU i7-8700K

 % ----------------CPU ---------------
 for m=1:100
   for n=1:100
            aa = rand(m,n);
            bb= rand(m,n);
            cc = corr2(aa,bb);
    end
end
fprintf('\n CPU Run time [%2f]', toc);
% ----------------GPU ---------------
 for m=1:100 
    for n=1:100
             aaa= rand(m,n,'gpuArray');
             bbb= rand(m,n,'gpuArray');
             ccc = corr2(aaa,bbb);
    end
end
fprintf('\n GPU Run time [%2f]', toc);

Result : CPU Run time [0.320389], GPU Run time [9.299645]

Answer 1

[Priyank Sharma]

There are multiple factors which determine a GPU's performance. The headline number of cores a GPU has is not enough to accurately gauge performance for double precision computing. You can follow the steps below to test your GPU performance:

1. Run a standard benchmark test.

2. If the benchmark shows different behavior between the precision types: calculate the expected ratio between Single and Double Precision for their GPU card.

3. If the card is a laptop (mobile) card, adjust expectation for performance.

4. If there is a kernel timeout, the paralleldemo_gpu_benchmark benchmark is likely to trigger it.

5. If performance in the benchmarks is good and clear then it is likely a problem originating in code.

Standard GPU Benchmark:

There is a benchmarking test written by MathWorks Parallel Computing Team and available on the File Exchange: http://www.mathworks.com/matlabcentral/fileexchange/34080-gpubench

This test will do a variety of tests involving both memory and compute intensive tasks in both single and double precision.

One thing to understand is most GPUs are optimized for single-precision maths (as is used by OpenGL etc.). GeForce cards, mobile or otherwise, are quite good for single-precision performance but usually about 8x worse for double.

MATLAB defaults to using double-precision everywhere. Of the NVIDIA cards, only the Tesla and top-end Quadro series do well at double-precision. Add to that the fact that a mobile GPU typically has far fewer cores than a desktop one.

Try the benchmark and check the result for your GPU.