Speed up a for loop by utilizing a gpu

18 次查看（过去 30 天）
tiwwexx
tiwwexx 2022-6-30
回答： tiwwexx 2022-6-30
Hello, I'm trying to speed up the following for loop
for n=1:size(Q_test,3)
Q_test(:,:,n) = sigma(n) / 2 / mass(n) * Q_in(:,:,n);
end
I tried implementing this with all the above arrays being gpuArrays and running
Q_gpu = arrayfun(@scale_Q,sigmaE_gpu,mass_gpu,Q_gpu,size(Q_gpu,3));
function output = scale_Q(sigma,mass,Q_in,itterations)
output = zeros(size(Q_in),'single');
for n=1:itterations
output(:,:,n) = sigma(n) / 2 / mass(n) * Q_in(:,:,n);
end
end
but I get the error
"Error using gpuArray/arrayfun
Arrays have incompatible sizes for this operation."
% Error using gpuArray/arrayfun
% Arrays have incompatible sizes for this operation.
Any help would be appriciated!

请先登录，再回答此问题。

采纳的回答

tiwwexx
tiwwexx 2022-6-30
This solution was given by Jan here.
Q_gpu = reshape(0.5*sigmaE_gpu./mass_gpu,1,1,size(sigmaE_gpu,1)).*Q_gpu;

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 GPU Computing 的更多信息

标签

产品


版本

R2021b

Community Treasure Hunt

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

Start Hunting!

Translated by