Get the latest tech news
Turning Off Zen 4's Op Cache for Curiosity and Giggles
CPUs start executing instructions by fetching those instruction bytes from memory and decoding them into internal operations (micro-ops).
From prior data, I know that SPEC CPU2017’s floating point workloads tend to emphasize backend core and memory performance. I suspect the only reason it doesn’t lose more is because 548.exchange2 was already heavily utilizing the core in 1T mode, and had lower than average (~11%) gains even with the op cache on. Workloads that stress those aspects, like high IPC code that fits within L1 caches, certainly take a hit from losing frontend bandwidth.
Or read this on Hacker News