[llvm-commits] [PATCH] Add MSVC multi processor compilation to build system

[Óscar Fuentes]

Erik Olofsson <Erik.Olofsson at hansoft.se> writes:

> From: Francois Pichet [mailto:pichet2000 at gmail.com]
>> I have a Intel Q6600 which is quadcore without hyper threading.
>> When I do a build my CPU utilization is maxed at 100% (on the 4 core) almost
>> all the time.
>>
>> What value do you have for "maximum number of parallel project build"
>> in your visual studio options?  (here: http://tinyurl.com/5wrqczs)
>
> 16, and as you can see from the graph, during some parts of the build
> the CPU utilization is close to 100%
>
>> I suspect your value is not equal to the maximum you could put. That would
>> explain the difference you get. Put 16 there and redo your tests with and
>> without your patch.
>>
>> I think that for LLVM it is better to configure multiprocessor msvc build using
>> that option instead  of the /MP switch.
>
> That may be true for your processor, but as it doesn't hurt
> performance and the tendency is for the number of cores to increase
> going forward I think it's worthwhile.

I have the same CPU as Francois and see it pretty busy while building
with the IDE without /MP. However, building top-level targets doesn't
work so well for Erik because he has 16 virtual processors and there
are no enough independent libraries/tools to keep them all busy. So I
can understand that /MP makes a difference for him. Erik is also right
pointing out that increasing the number of cores is the trend on the CPU
industry. For those reasons I think that it is a good idea to add /MP to
the MSVC_IDE build and turn it on by default.

[snip]