Improving the accuracy of snoop filtering using stream registers

Multi-core processors have become mainstream; they provide parallelism with relatively low complexity. As true on-chip SMPs evolve, coherence traffic between cores is becoming problematic, both in terms of performance and power. The negative effects of coherence (snoop) traffic can be significantly mitigated through snoop filtering. Shielding each cache with a device that can squash snoop requests for addresses known not to be in cache improves performance significantly for caches that cannot perform normal load and snoop lookups simultaneously. In addition, reducing snoop lookups yields power savings. This paper introduces Stream Register snoop filtering, which captures the spatial locality of multiple memory reference streams in a few registers. We propose a snoop filter that combines Stream Registers with "snoop caching", a mechanism that captures the temporal locality of frequently accessed addresses. Simulations of Splash- 2 benchmarks on a 4-core multiprocessor illustrate tradeoffs and strengths of these two techniques. Their combination is most effective, eliminating 94-99% of all snoop requests using very few stream registers and snoop cache lines.