On Sun, 16 Jul 2017 10:23:08 +0200 Jesper Dangaard Brouer <bro...@redhat.com> wrote:
> On Tue, 11 Jul 2017 12:37:10 -0700 John Fastabend <john.fastab...@gmail.com> > wrote: > > [...] > > > > hmm maybe Jesse or Alex have some clues. Adding them to the CC list. > > This seems related to Hyper-Threading. I tried to disable > hyperthreading in the BIOS, and the problem goes away. That is, the > benchmarks are no-longer affected by the CPU tuned-adm profile. Wow, nice job figuring that out. I went and took a look at tuned-adm documentation but didn't see anything that stood out that would cause the behavior you were seeing. I suspect your toplev results showing this is a frontend problem mesh nicely with the hyper-threading configuration inducing the bad behavior, since there is still only one execution unit, and the fetchers would have to stall. I think you've rediscovered why the forwarding / routing crowd generally turns off hyperthreading.