Concurrent TLB Flushing For Linux 5.13 Provide A Small Performance Benefit
Linux 5.13's x86 memory management work is bringing a minor performance optimization that is particularly beneficial in light of the CPU security mitigations in recent years that have an impact on the TLB.
VMware engineers for the past two years were looking at concurrent TLB flushes for Linux to flush local and remote translation lookaside buffers concurrently.
Benchmarks with prior rounds of the concurrent TLB flushing showed a 1~4% improvement for Sysbench.
The x86/mm pull request notes, "implement concurrent TLB flushes, which overlaps the local TLB flush with the remote TLB flush. In testing this improved sysbench performance measurably by a couple of percentage points, especially if TLB-heavy security mitigations are active."
Of course, once the merge window is through I'll be around with plenty of Linux 5.12 vs. 5.13 kernel benchmarking.