I had misinterpreted the original results... it is interesting that the
runtime SE/FS selection outperforms compile time in some cases.  It's not
clear why that would be the case, unless it's some really subtle thing like
the code got moved around which happened to eliminate some icache misses,
or something like that.

This all looks good, with the one exception being that an SE compile now
takes 31% more user time.  It's true that that's negligible wall clock time
on a large-scale MP, but not everyone always has access to that.  I would
still really like us to figure out a way to make compilation more
configurable so we can leave out the components that no one needs in SE
mode (though ideally that's just one application of a more general
mechanism rather than a lingering hard-coded "SE vs FS" difference).

Steve

On Fri, Jan 27, 2012 at 1:37 AM, Gabriel Michael Black <
[email protected]> wrote:

> I was, although I'd have to check again what number I used. I set it to
> something with the SCONSOPTS (or similar) environment variable on each
> computer I use and then don't worry about it any more. You can see that the
> "real" time is substantially less than the other times added together,
> about 1/6th as much. Since I probably set -j to something more than 6 on
> that machine, I'd guess that's around where it saturates.
>
> Gabe
>
>
> Quoting Korey Sewell <[email protected]>:
>
>  Gabe,
>> do you think there will be any more of a difference when you build scons
>> with the "-j" flags?  (e.g scons -j2 or scons -j4)
>>
>>
>> On Sat, Jan 21, 2012 at 5:38 AM, Gabe Black <[email protected]>
>> wrote:
>>
>>  Here are some build time numbers.
>>>
>>> time scons build/X86/gem5.opt
>>> real    4m35.081s
>>> user    23m48.901s
>>> sys     1m34.119s
>>>
>>>
>>> time scons build/X86_SE/gem5.opt
>>> real    4m10.493s
>>> user    18m9.749s
>>> sys     1m6.644s
>>>
>>>
>>> time scons build/X86_FS/gem5.opt
>>> real    4m33.993s
>>> user    22m58.946s
>>> sys     1m31.277s
>>>
>>> Gabe
>>> ______________________________**_________________
>>> gem5-dev mailing list
>>> [email protected]
>>> http://m5sim.org/mailman/**listinfo/gem5-dev<http://m5sim.org/mailman/listinfo/gem5-dev>
>>>
>>>
>>
>>
>> --
>> - Korey
>> ______________________________**_________________
>> gem5-dev mailing list
>> [email protected]
>> http://m5sim.org/mailman/**listinfo/gem5-dev<http://m5sim.org/mailman/listinfo/gem5-dev>
>>
>>
>
> ______________________________**_________________
> gem5-dev mailing list
> [email protected]
> http://m5sim.org/mailman/**listinfo/gem5-dev<http://m5sim.org/mailman/listinfo/gem5-dev>
>
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to