On Tue, Jul 11, 2017 at 4:32 AM, Matt Fleming <m...@codeblueprint.co.uk> wrote: > On Fri, 30 Jun, at 01:44:22PM, Matt Fleming wrote: >> On Thu, 29 Jun, at 08:53:12AM, Andy Lutomirski wrote: >> > *** Ingo, even if this misses 4.13, please apply the first patch before >> > *** the merge window. >> > >> > There are three performance benefits here: >> > >> > 1. TLB flushing is slow. (I.e. the flush itself takes a while.) >> > This avoids many of them when switching tasks by using PCID. In >> > a stupid little benchmark I did, it saves about 100ns on my laptop >> > per context switch. I'll try to improve that benchmark. >> > >> > 2. Mms that have been used recently on a given CPU might get to keep >> > their TLB entries alive across process switches with this patch >> > set. TLB fills are pretty fast on modern CPUs, but they're even >> > faster when they don't happen. >> > >> > 3. Lazy TLB is way better. We used to do two stupid things when we >> > ran kernel threads: we'd send IPIs to flush user contexts on their >> > CPUs and then we'd write to CR3 for no particular reason as an excuse >> > to stop further IPIs. With this patch, we do neither. >> >> Heads up, I'm gonna queue this for a run on SUSE's performance test >> grid. > > FWIW, I didn't see any change in performance with this series on a > PCID-capable machine. On the plus side, I didn't see any weird-looking > bugs either. > > Are your benchmarks available anywhere?
https://git.kernel.org/pub/scm/linux/kernel/git/luto/misc-tests.git/ I did: $ ./context_switch_latency_64 0 process same and $ ./madvise_bounce_64 10k [IIRC -- it might have been a different loop count] --Andy
