A Versatile Software Systolic Execution Model for GPU Memory-Bound Kernels

by   Peng Chen, et al.

This paper proposes a versatile high-performance execution model, inspired by systolic arrays, for memory-bound regular kernels running on CUDA-enabled GPUs. We formulate a systolic model that shifts partial sums by CUDA warp primitives for the computation. We also employ register files as a cache resource in order to operate the entire model efficiently. We demonstrate the effectiveness and versatility of the proposed model for a wide variety of stencil kernels that appear commonly in HPC, and also convolution kernels (increasingly important in deep learning workloads). Our algorithm outperforms the top reported state-of-the-art stencil implementations, including implementations with sophisticated temporal and spatial blocking techniques, on the two latest Nvidia architectures: Tesla V100 and P100. For 2D convolution of general filter sizes and shapes, our algorithm is on average 2.5x faster than Nvidia's NPP on V100 and P100 GPUs.


Fast convolution kernels on pascal GPU with high memory efficiency

The convolution computation is widely used in many fields, especially in...

Revisiting Temporal Blocking Stencil Optimizations

Iterative stencils are used widely across the spectrum of High Performan...

AN5D: Automated Stencil Framework for High-Degree Temporal Blocking on GPUs

Stencil computation is one of the most widely-used compute patterns in h...

GraphCage: Cache Aware Graph Processing on GPUs

Efficient Graph processing is challenging because of the irregularity of...

Persistent Kernels for Iterative Memory-bound GPU Applications

Iterative memory-bound solvers commonly occur in HPC codes. Typical GPU ...

Performance Models for Data Transfers: A Case Study with Molecular Chemistry Kernels

With increasing complexity of hardwares, systems with different memory n...

Please sign up or login with your details

Forgot password? Click here to reset