Adaptable Register File Organization for Vector Processors

TL;DR

This paper introduces AVA, an adaptable vector processor architecture that dynamically reconfigures vector length to optimize performance and resource utilization across applications with varying data parallelism levels.

Contribution

AVA is a novel vector processor design that combines efficiency for short vectors with the ability to reconfigure for longer vectors, improving performance and resource use.

Findings

AVA achieves 2X speedup over its default configuration.

AVA saves 50% area compared to long vector VP.

Reconfigurable MVL enhances performance for diverse applications.

Abstract

Modern scientific applications are getting more diverse, and the vector lengths in those applications vary widely. Contemporary Vector Processors (VPs) are designed either for short vector lengths, e.g., Fujitsu A64FX with 512-bit ARM SVE vector support, or long vectors, e.g., NEC Aurora Tsubasa with 16Kbits Maximum Vector Length (MVL). Unfortunately, both approaches have drawbacks. On the one hand, short vector length VP designs struggle to provide high efficiency for applications featuring long vectors with high Data Level Parallelism (DLP). On the other hand, long vector VP designs waste resources and underutilize the Vector Register File (VRF) when executing low DLP applications with short vector lengths. Therefore, those long vector VP implementations are limited to a specialized subset of applications, where relatively high DLP must be present to achieve excellent performance with high efficiency. To overcome these limitations, we propose an Adaptable Vector Architecture (AVA) that leads to having the best of both worlds. AVA is designed for short vectors (MVL=16 elements) and is thus area and energy-efficient. However, AVA has the functionality to reconfigure the MVL, thereby allowing to exploit the benefits of having a longer vector (up to 128 elements) microarchitecture when abundant DLP is present. We model AVA on the gem5 simulator and evaluate the performance with six applications taken from the RiVEC Benchmark Suite. To obtain area and power consumption metrics, we model AVA on McPAT for 22nm technology. Our results show that by reconfiguring our small VRF (8KB) plus our novel issue queue scheme, AVA yields a 2X speedup over the default configuration for short vectors. Additionally, AVA shows competitive performance when compared to a long vector VP, while saving 50% of area.