lkml.org 
[lkml]   [2003]   [Jul]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [patch] scheduler fix for 1cpu/node case
>> I am going to ask a silly question, do we have any data showing this really
>> is a problem on AMD? I would think, even if we have an idle cpu, sometimes
>> a little delay on task migration (on NUMA) may not be a bad thing. If it
>> is too long, can we just make the rebalance ticks arch specific?
>
> The fact that global rebalances are done only in the timer interrupt
> is simply bad! It complicates rebalance_tick() and wastes the
> opportunity to get feedback from the failed local balance attempts.

The whole point of the NUMA scheduler is to rebalance globally less
often. Now I'd agree that in the idle case we want to be more agressive,
as your patch does, but we need to be careful not to end up in a cacheline
thrashing war.

Aiming for perfect balance is not always a good idea - the expense of both
the calculation and the migrations has to be taken into account. For some
arches, it's less important than others ... one thing we're missing is
a per arch "node-balance-agressiveness" factor.

Having said that, the NUMA scheduler is clearly broken on Hammer at the
moment, from Andi's observations.

> If you want data supporting my assumptions: Ted Ts'o's talk at OLS
> shows the necessity to rebalance ASAP (even in try_to_wake_up). There
> are plenty of arguments towards this, starting with the steal delay
> parameter scans from the early days of multi-queue schedulers (Davide
> Libenzi), over my experiments with NUMA schedulers and the observation
> of Andi Kleen that on Opteron you better run from the wrong CPU than
> wait (if the tasks returns to the right cpu, all's fine anyway).

That's a drastic oversimplification. It may be better in some
circumstances, on some benchmarks. For now, let's just get your patch
tested on Hammer, and see if it works better to have the NUMA scheduler
on than off after your patch ...

M.



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:47    [W:0.129 / U:6.604 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site