Expanding GP's Computing Capabilities
Posted: Aug 1st, '18, 02:38
Hello! a suggestion;
So GP seems rather stuck on the slow speeds of CPU computation, which although is good enough for small playing around with small images, small amounts of numbers, and small arrays however when you move on to bigger things such as large arrays or modifying large images everything seems to slow down, especially on things that happen sequentially.
What I am saying is to use a general GPU library that supports ARM, OpenCL, and CUDA such as https://arrayfire.com/ and by also using the fastmath.h library. This seems like it would speed things up considerably.
Is there anyone who has access to the interpreter that could implement this?
I would love to help if I can!
So GP seems rather stuck on the slow speeds of CPU computation, which although is good enough for small playing around with small images, small amounts of numbers, and small arrays however when you move on to bigger things such as large arrays or modifying large images everything seems to slow down, especially on things that happen sequentially.
What I am saying is to use a general GPU library that supports ARM, OpenCL, and CUDA such as https://arrayfire.com/ and by also using the fastmath.h library. This seems like it would speed things up considerably.
Is there anyone who has access to the interpreter that could implement this?
I would love to help if I can!