On Monday, 24 June 2013 at 15:04:37 UTC, dennis luehring wrote:
>
> i know - my question was - how does that look using msvc...
>


I just did a very quick test using the latest DMD source:

Using the command line

-release -unittest -c D:\DTesting\dmd.2.063\src\phobos\std\algorithm.d


DMD built with DMC takes ~49 seconds to complete, but DMD build with VC2008 only takes ~12 seconds. (Need to get a proper VC build done to test it properly).
Looks like the DMC build spends far more time allocating memory, even though the peak memory usage is only slightly lower in the VS version?

Am 24.06.2013 18:15, schrieb Richard Webb:
> On Monday, 24 June 2013 at 15:04:37 UTC, dennis luehring wrote:
>>
>> i know - my question was - how does that look using msvc...
>>
>
>
> I just did a very quick test using the latest DMD source:
>
> Using the command line
>
> -release -unittest -c
> D:\DTesting\dmd.2.063\src\phobos\std\algorithm.d
>
>
> DMD built with DMC takes ~49 seconds to complete, but DMD build
> with VC2008 only takes ~12 seconds. (Need to get a proper VC
> build done to test it properly).
> Looks like the DMC build spends far more time allocating memory,
> even though the peak memory usage is only slightly lower in the
> VS version?

so it could be std library implementation related - can DMC use the msvc libs? (just for the compare)

and you should also try 2010 - or better 2012 msvc (it still gets speedier code out)

On 6/24/2013 6:19 AM, dennis luehring wrote:
> how does that look using msvc compiling the dmd compiler
> as it turns out that msvc make dmd much faster

The profile report was done by gcc/gprof.

And besides, better compilers shouldn't change profile results.

On 6/24/2013 8:57 AM, dennis luehring wrote:
> what is the sense of doing profiling in BUILD=debug?

Non-debug builds may strip out the symbols, rendering the profiling report useless.

On 06/24/2013 05:57 PM, dennis luehring wrote:
> what is the sense of doing profiling in BUILD=debug? or is the optimizer
> still active?
>
That is compiling phobos in debug (non-optimized) mode not dmd. As I said building phobos in release didn't succeed for whatever reason.

I can try to install kernel debuginfo that 12% might contain some useful information.

On 6/24/13, dennis luehring <dl.soluz@gmx.net> wrote:
> i know - my question was - how does that look using msvc...

The recent speed improvements in git-head have improve both the DMC-based and MSVC-based DMD executables. It shaved off from 700msecs on small projects up to 4 seconds on large projects for clean builds (for me).

That being said the MSVC built compiler is usually twice as fast as the DMC built one. It goes to show that whatever is in that DMC backend is long outdated.

On 06/24/2013 08:43 PM, Martin Nowak wrote:
>
> I can try to install kernel debuginfo that 12% might contain some useful
> information.

http://codepad.org/gWrGvm40

On Mon, 24 Jun 2013 11:44:24 -0700, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:

> On 6/24/13, dennis luehring <dl.soluz@gmx.net> wrote:
>> i know - my question was - how does that look using msvc...
>
> The recent speed improvements in git-head have improve both the
> DMC-based and MSVC-based DMD executables. It shaved off from 700msecs
> on small projects up to 4 seconds on large projects for clean builds
> (for me).
>
> That being said the MSVC built compiler is usually twice as fast as
> the DMC built one. It goes to show that whatever is in that DMC
> backend is long outdated.

I asked Walter about that recently, he says that with all the focus on D there hasn't been enough time to seriously develop the backend in a LONG time. IIRC he said it's been something like 15 years since it's seen any serious development work...

NOTE: It came up because I was asking about getting things like auto-vectorization and Haswell's new Transactional Synchronization Extensions and the related instructions into the backend.

-- 
Adam Wilson
IRC: LightBender
Project Coordinator
The Horizon Project
http://www.thehorizonproject.org/

On 6/24/2013 12:27 PM, Adam Wilson wrote:
> I asked Walter about that recently, he says that with all the focus on D there
> hasn't been enough time to seriously develop the backend in a LONG time. IIRC he
> said it's been something like 15 years since it's seen any serious development
> work...

It's also possible that a faster malloc() is responsible for much of it.

Am 24.06.2013 20:01, schrieb Walter Bright:
> On 6/24/2013 6:19 AM, dennis luehring wrote:
>> how does that look using msvc compiling the dmd compiler
>> as it turns out that msvc make dmd much faster
>
> The profile report was done by gcc/gprof.
>
> And besides, better compilers shouldn't change profile results.
>

different optimization strategies can shuffle the results totaly - or do we talk about different things