| Summary: | sys-kernel/gentoo-sources-2.6.37-r4: powertop reports: High [kernel scheduler] Load balancing tick | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Plero H <plero_hero> |
| Component: | [OLD] Core system | Assignee: | Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel> |
| Status: | RESOLVED NEEDINFO | ||
| Severity: | normal | CC: | jer, mobile+disabled |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | AMD64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
powertop's output when idle (im off g15daemon off)
kernel config dmesg output lsmod output powertop's output when idle (im on, g15daemon on) powertop's output with load powertop's output when almost idle 2.6.38-r2 2.6.38 kernel config powertop's output under load 2.6.38-r2 |
||
|
Description
Plero H
2011-04-15 18:48:58 UTC
Created attachment 270129 [details]
powertop's output when idle (im off g15daemon off)
powertop -d -t 10
output
Created attachment 270131 [details]
kernel config
my kernel config
Created attachment 270133 [details]
dmesg output
dmesg output
Created attachment 270135 [details]
lsmod output
Created attachment 270137 [details]
powertop's output when idle (im on, g15daemon on)
booting with nohz=off, highres=off no problem arise.
Created attachment 270139 [details]
powertop's output with load
mprime on two cores
Can you try 2.6.38 and see if you get the same results? Created attachment 270599 [details]
powertop's output when almost idle 2.6.38-r2
using gentoo-sources-2.6.38-r2
firefox & transmission in the background. I don't think it changed.
Created attachment 270601 [details]
2.6.38 kernel config
Created attachment 270605 [details]
powertop's output under load 2.6.38-r2
mprime on 2 cores.
Did you notice different behavior with older kernels (ie 2.6.36)? I noticed that you have CONFIG_HZ=1000 in your config. Did you have the same setting in older kernels (in case you had fewer scheduler interrupts with older kernels). Thanks. |