Later this week it looks like the latest round of Cache Aware Scheduling fixes will be merged for enhancing this kernel scheduler functionality for grouping related tasks together on CPU cores that share the same last level cache (LLC). Cache Aware Scheduling can lead to better cache locality and lower latency by avoiding the costs of potential cache bouncing -- that is when bugs don't get in the way. Intel engineer TIm Chen has been sorting out some CAS bugs since the introduction of this feature in the mainline Linux 7.2 kernel. These patches keep the nr_pref_llc_running in the runnable domain, honor the migrate_llc_task semantics in the active load balance, refresh the LLC capacity across CPU hotplugs, and skip the kernel thread for cache aware scheduling and other low-level adjustments. The patches today were merged into the sched/urgent branch of tip/tip.git. Now being in a TIP "urgent" branch, look for them to be submitted for the current Linux 7.3 kernel cycle later this week. All of these patches are also marked for back-porting to the Linux 7.2 kernel too. Earlier this cycle was also fixing Cache Aware Scheduling for hybrid CPUs so it will be interesting to revisit the Cache Aware Scheduling benchmarking with these latest fixes.
Cache Aware Scheduling Fixes Queued For Submission To Linux 7.3, Backporting To 7.2
Full Article
Original Source
Read the full article at Phoronix →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.