Techniques for enlarging instruction streams

Abstract

This work presents several techniques for enlarging instruction streams. We call stream to a sequence of instructions from the target of a taken branch to the next taken branch, potentially containing multiple basic blocks. The long size of instruction streams makes it possible for a fetch engine based on streams to provide high fetch bandwidth, which leads to obtaining performance results comparable to a trace cache. The long size of streams also enables the next stream predictor to tolerate the prediction table access latency. Therefore, enlarging instruction streams will improve the behavior of a fetch engine based on streams. We provide a comprehensive analysis of dynamic instruction streams, showing that focusing on particular kinds of stream is not a good strategy due to Amdahl's law. Consequently, we propose the multiple stream predictor, a novel mechanism that deals with all kinds of streams by combining single streams into long virtual streams. We show that our multiple stream predictor is able to tolerate the prediction access latency without requiring the complexity caused by additional hardware mechanisms like prediction overriding.

Postprint (published version)