More QPU magic from Pete Warden

Back in June, we mentioned Pete Warden’s port of the Deep Belief image-recognition SDK to the Pi, which used the VideoCore IV QPUs to provide an accelerated GEMM matrix-multiply function. Since then, Pete’s been optimizing his code, and has reduced the time required to process an image to 3 seconds (versus 20 seconds for the baseline ARM implementation and 6 seconds for his original QPU version).

Classifying dogs and their balls

Classifying dogs and their balls

In the spirit of “leaving a trail of breadcrumbs through the forest”, Pete has written up an excellent summary of his experiences here. Head on over and check it out.

6 comments

3xBackups avatar

I would have included the image that was being processed.
https://raw.githubusercontent.com/jetpacapp/DeepBeliefSDK/gh-pages/source/data/dog.jpg

Asad Kazmi avatar

thanks :)

Craig Van Degrift avatar

Thanks for the insight into how tricky GPU programming can be! The summary given in the link was fascinating even to this mere mortal who has only done a littl assembly programming on simple processors.

It is marvelous that a computer aimed at education of children doubles as a scientific tool. I wonder how long it will be before a 12-year old creates an interesting GPU program on the Pi.

JPW avatar

Is that supposed to read ‘QPU’ or ‘GPU’ ? Wikipedia defines ‘QPU’ as a ‘Quantum processing unit’……I know the Pi is brilliant; but I didn’t realize it was doing Quantum Computing just yet !!

Liz Upton avatar

QPU stands for Quad Processor Unit.

jpw avatar

@Liz Thanks for the clarification !

Comments are closed