Re: -mm seems significanty slower than mainline on kernbench

["Martin J. Bligh" <mbligh@google.com>]

Peter Williams wrote:

> Peter Williams wrote:
>
>> Con Kolivas wrote:
>>
>>> On Wed, 11 Jan 2006 01:38 pm, Peter Williams wrote:
>>>
>>>> Con Kolivas wrote:
>>>> > I guess we need to check whether reversing this patch helps.
>>>>
>>>> It would be interesting to see if it does.
>>>>
>>>> If it does we probably have to wear the cost (and try to reduce it) as
>>>> without this change smp nice support is fairly ineffective due to the
>>>> fact that it moves exactly the same tasks as would be moved without 
>>>> it.
>>>>  At the most it changes the frequency at which load balancing occurs.
>>>
>>>
>>>
>>>
>>> I disagree. I think the current implementation changes the balancing 
>>> according to nice much more effectively than previously where by 
>>> their very nature, low priority tasks were balanced more frequently 
>>> and ended up getting their own cpu.
>>
>>
>>
>> I can't follow the logic here and I certainly don't see much 
>> difference in practice.
>
>
> I think I've figured out why I'm not seeing much difference in 
> practice.  I'm only testing on 2 CPU systems and it seems to me that 
> the main difference that the SMP nice patch will have is in selecting 
> which CPU to steal tasks from (grabbing the one with the highest 
> priority tasks) and this is a non issue on a 2 CPU system.  :-(
>
> So I should revise my statement to say that it doesn't make much 
> difference if there's only 2 CPUs.
>

If nothing's niced, why would it be affecting scheduling decisions at all?
That seems broken to me ?

M.