On 04/08/2010 02:44 PM, Avi Kivity wrote:
I'll try running this with a kernel build in parallel.
Results here are less than stellar. While khugepaged is pulling pages
together, something is breaking them apart. Even after memory
pressure is removed, this behaviour continues. Can it be that
compaction is tearing down huge pages?
ok, #19 is a different story. A 1.2GB sort vs 'make -j12' and a cat of
the source tree and some light swapping, all in 2GB RAM, didn't quite
reach 1.2GB but came fairly close. The sort was started while memory
was quite low so it had to fight its way up, but even then khugepaged
took less that 1.5 seconds total time after a _very_ long compile.
I observed huge pages being used for gcc as well, likely not bringing
much performance since kernel compiles don't use a lot of memory per
file. I'll look at the link stage, that will probably use a lot of
large pages.
--
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxxx For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>