Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: WAPBL patch for testing


I repeated the tests 4) and 5) from my PR#39564 followup, and got

1) a standard " -j 16 -x release":

   7046.891u 2494.916s 25:58.97 612.0%     0+0k 15+474695io 37619pf+0w

2) as 1) with a "sh -c 'while true; do sync; sleep 1; done'" loop running

   6901.178u 2622.343s 24:25.99 649.6%     0+0k 21+51892io 37587pf+0w
   0.8u 68.2s 25:15.17 4.5% 0+0k 0+56019io 0pf+0w

   for the shell loop.

The wallcklock difference is quite a lot smaller than previously,
but the number of I/Os is still many times larger in the first
case, even if you add the two.

I collected lockstat output for the two runs, they are at

respectively (35 corresponds to 1, 34 to 2 above).


- Håvard

Home | Main Index | Thread Index | Old Index