| 1 |
On Sun, 14 Jun 2009 06:07:16 +0000 (UTC) |
| 2 |
Jason Lynch <jason@×××××××××.com> wrote: |
| 3 |
|
| 4 |
> Thus, I have five processes, four at nice 19, one at nice 0, a load |
| 5 |
> average of just over 5, but only 3 out of the 4 cores are actually doing |
| 6 |
> anything. |
| 7 |
|
| 8 |
That's an interesting observation with quite a trivial scenario. |
| 9 |
|
| 10 |
So I thought to check it out and ran 8 niced copies "while True: pass" |
| 11 |
script on 8-core machine, atop showed 799-800% load, 100% for each core. |
| 12 |
Ninth, non-niced process indeed drops the load to 700-710%, with one |
| 13 |
core absolutely free. |
| 14 |
|
| 15 |
Then, I've tried to remove nice form the equation and load held at 800% |
| 16 |
with 8, 9, 10 and more processes. Nice-only processes behave similary, |
| 17 |
loading all eight cores. |
| 18 |
|
| 19 |
So I guess the problem (or feature?) is related to nice / non-nice |
| 20 |
processes' scheduling and exists at least in 2.6.29 kernel. |
| 21 |
Gotta google it a bit later, bet someone on lkml should've noticed it. |
| 22 |
|
| 23 |
-- |
| 24 |
Mike Kazantsev // fraggod.net |
Archives