International Business Machines Corporation

Arash Ashari

Berthold Reinwald

Shirish Tatikonda

Matthias Boehm

Alexandre Valentinovich Evfimievski

John D Keenleyside

Keith W Campbell

A method for optimization of machine learning (ML) workloads on a graphics processor unit (GPU). The method includes identifying a computation having a generic pattern commonly observed in ML processes. Hierarchical aggregation spanning a memory hierarchy of the GPU for processing is performed for the identified computation including maintaining partial output vector results in shared memory of the GPU. Hierarchical aggregation for vectors is performed including performing intra-block aggregation for multiple thread blocks of a partial output vector results on GPU global memory.

Pipelined approach to fused kernels for optimization of machine learning workloads on graphical processing units

A method for optimization of machine learning (ML) workloads on a graphics processor unit (GPU). The method includes identifying a computation having a generic pattern commonly observed in ML processes. An optimized fused GPU kernel is employed to exploit temporal locality for inherent data-flow dependencies in the identified computation. Hierarchical aggregation spanning a memory hierarchy of the GPU for processing for the identified computation is performed. GPU kernel launch parameters are estimated following an analytical model that maximizes thread occupancy and minimizes atomic writes to GPU global memory.

Growing community of inventors

Kirkland, WA, United States of America

Arash Ashari