More VM benchmarks
OK I'm not too unhappy with kbuild now. I've flattened the curve a bit more since you last saw it. Would be nice if we could get j8 and j10 faster but you can't win them all.
I'm not sure what happens further on - Roger indicates that perhaps 2.4 overtakes 2.6 again at j24 although the patchset he used (http://www.kerneltrap.org/~npiggin/vm/3/) performs far worse than this one at j16. This is really not a big deal IMO, but I might run it and see what happens.
The systime benchmarks are just a bit of fun. They don't mean too much because I didn't measure how much work kswapd is doing...
Oh, the base kernel is 2.6.2-rc3-mm1 for -np3. I'll release the patches shortly.
Best regards,
Nick
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Quoted 6 lines · Nick Piggin
http://www.kerneltrap.org/~npiggin/vm/5/
OK I'm not too unhappy with kbuild now. I've flattened the curve a bit more since you last saw it. Would be nice if we could get j8 and j10 faster but you can't win them all.
Quoted 6 lines
I'm not sure what happens further on - Roger indicates that perhaps 2.4 overtakes 2.6 again at j24 although the patchset he used (http://www.kerneltrap.org/~npiggin/vm/3/) performs far worse than this one at j16. This is really not a big deal IMO, but I might run it and see what happens.
They're about even here at an hour apiece. Thats good, not that I would ever try to optimise for this case...
2.6.2-rc3-mm1-np3
kbuild: make -j24 bzImage
144.45user 31.58system 1:00:08elapsed 4%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (380524major+919600minor)pagefaults 0swaps
npiggin@ropeable:~/vm$ cat bkb-24.out
2.4.23
kbuild: make -j24 bzImage
148.41user 43.11system 1:02:41elapsed 5%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (612476major+633988minor)pagefaults 0swaps
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Quoted 6 lines · Nick Piggin
Quoted 2 lines · Nick Piggin
Sorry to keep replying to myself. I've done some runs with my IO scheduler regression tests and these patches don't make a significant difference one way or the other although the results with the patches are generally a bit better.
Tested OraSim pgbench nickbench tiobench and the read/ls kernel tree during a streaming read/write. 256MB RAM this time.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>