On Mon, Feb 10, 2020 at 1:05 AM Floris Van Nee <florisvan...@optiver.com> wrote: > I ran all the tests on two different machines, several times for 1 hour each > time. I'm still having a hard time getting reliable results for the 30 > clients case though. I'm pretty certain the patches bring a performance > benefit, but how high exactly is difficult to say. As for applying only patch > 1+2 or all three patches - I found no significant difference between these > two cases. It looks like all the performance benefit is in the first two > patches.
Attached is v3, which no longer includes the third patch/optimization. It also inlines (in the second patch) by marking the _bt_compare definition as inline, while not changing anything in nbtree.h. I believe that this is portable C99 -- let's see what CF Tester thinks of it. I'm going to test this myself. It would be nice if you could repeat something like the previous experiments with v3, Floris. master vs v3 (both patches together). With a variable number of clients. Thanks -- Peter Geoghegan
v3-0002-Inline-_bt_compare.patch
Description: Binary data
v3-0001-Avoid-pipeline-stall-in-_bt_compare.patch
Description: Binary data