Inference with ArrayFire* and oneAPI

This session demonstrates a simple machine-learning inference pipeline using the OpenCL™ software interoperability of the oneAPI platform. The talk introduces the ArrayFire* library and the derivative Flashlight project used as motivating examples. Data flows from the oneAPI Video Processing Library to these libraries as an example of integrating the oneAPI tool with existing GPU codebases.

Stefan Yurkevitch is a software engineer at ArrayFire where he helps maintain the ArrayFire library and obsesses over all things high performance.

 

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.