Re: ULE/yielding patch for testing.

From: Kris Kennaway <kris_at_FreeBSD.org>
Date: Sat, 06 Oct 2007 14:58:23 +0200
Benjamin Close wrote:
> Marc Fonvieille wrote:
>> On Sat, Oct 06, 2007 at 09:44:29AM +0200, Marc Fonvieille wrote:
>>  
>>> On Tue, Oct 02, 2007 at 04:53:33PM -0700, Jeff Roberson wrote:
>>>    
>>>> Enclosed is a patch that does two things:
>>>>
>>>> 1)  Reduces UP context switch time by over 10% making it faster than 
>>>> 4BSD on UP.  On SMP it's hard to compare since ULE can do as many as 
>>>> 30x as many switches per second on my 8way system.
>>>>
>>>> 2)  Restores old sched_yield() behavior from 6.x.  This was changed 
>>>> in -current unintentionally I think.
>>>>
>>>> I'd appreciate any extra testing.  The ULE context switch time 
>>>> improvements required some changes to the frequency that we 
>>>> recalculate priorities.  I'm mostly interested in hearing whether 
>>>> this causes any regression in normal workloads.
>>>>
>>>>       
>>> [tested with _ULE]
>>>
>>> This seems to help a bit the things for me (desktop use during
>>> compilations).  But 6.X is still faster, i.e, less lags and less jerky
>>> mouse movements during port builds.
>>> Thanks for spending time improving things, I really appreciate it.
>>>
>>>     
>>
>> After more testing, the lag problem is caused by the use of firefox.
>> Once I try to open a heavy webpage, both firefox and Xorg become slow,
>> the rest is fine.  By the "rest" I mean: audacious playing mp3s, wget and
>> ncftp downloading huge (100MB) files, compiling wine and aMule running.
>>   
> I find the lag occuring with the 4BSD scheduler as well. dailytech.com 
> is a particularly good site at lagging the system (though a great website).
> Perhaps this isn't scheduler related?

I assume you checked whether you are touching swap.

Kris
Received on Sat Oct 06 2007 - 10:58:18 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:18 UTC