Hi guys,
I have some DBT-3 (decision support) results using Gavin's original
futex patch fix. It's on out 8-way Pentium III Xeon systems
in our STP environment. Definitely see some overall throughput
performance on the tests, about 15% increase, but not change with
respect to the number of context switches.
Perhaps it doesn't really address what's causing the incredible number
of context switches, but still helps. I think I'm seeing what Gavin
has, that it seems to solves some concurrency problems on x86 platforms.
Here's results without futexes:
http://khack.osdl.org/stp/298114/
Results with futexes:
http://khack.osdl.org/stp/298115/
Mark