Speeding up maxpooling.
Speeding up Maxpooling.
Collecting timings per layer.
Updating OpenCL experiment.
Adding HasAVX2 and HasAVX512. Small polishing at supersimple.lpr example.
Minor fixes in uab and ubit.
Updating super simple examples with correlation calculation.
Fixing bug reported by Louis Morin (Convolution with small Y input fails).