Forums > Gen

GEN and parallel computing


June 13, 2013 | 3:07 am

I was wondering as the GEN objects are implemented as some kind of stream processing technique, does it use the advantages of different CPU cores (i guess jit.gl.pix for sure on GPU)?

June 19, 2013 | 6:12 pm

Yes. Most jitter objects actually use parallel processing. Matrices are split up into chunks and dispatched to multiple threads for processing. How they are split up depends on the size of the matrix involved.

In jit.gen and jit.pix, the same applies. For jit.gl.pix, the GPU is used, which is inherently parallel.

June 19, 2013 | 7:24 pm

thanx Wesley!

January 28, 2015 | 7:30 am

Where is it possible to find more information about parallel processing with Jiter objects ?
(not using opengl/GPU processing)
Which objects, which treatments ??

February 5, 2015 | 4:30 pm

Where is it possible to find more information about parallel processing with Jiter objects ?
(not using opengl/GPU processing)
Which objects, which treatments ??

Check out the SDK… there is a little bit of documentation in there about how jitter breaks up matrices for parallel processing.

Each jitter object must explicitly use parallel specific routines – you don’t get parallel processing "for free". But I would imagine that any jitter object that has a matrix as an input and a matrix as an output, is most likely going to take advantage of this parallel processing. You can see a lot of examples in the jitter examples folder in the SDK.

Viewing 5 posts - 1 through 5 (of 5 total)