Re: [boost] serialization performance

From: Martin Slater (mslater_at_[hidden])
Date: 2005-11-12 21:24:49

> Hmmm - then the fact it showed up on the profiler suggests that
> the program wasn't compiled with full optimisation?  You might
> want to expand upon this.

This may well be out of our control, IIRC it was coming from the crt 
library provided with VC, in user code you can control it via the 
necessary pragma but asking users to rebuild the crt library is a bit 
unreasonable;) Anyway if this test has been reduced to pointer / pointer 
comparision then all should be good.

>>If your interested in vtune they do an evaluation verion at
>>https://registrationcenter.intel.com/EvalCenter/EvalForm.aspx?ProductID=319
>>It is simply the best profiling tool I have ever used.
> 
> 
> At one time I had the intel eval compiler installed and it was very good.
> My license expired and I just didn't have the incentive to actually pay
> for it.  Too bad I would have liked to have it my test suite.

This is just an eval for vtune, still not cheap though.

> 
> 
>>I'd be more than happy to help out with any profiling and
>>optimisation I can.
> 
> 
> Well, you'll get your chance pretty soon.  Soon I'll be checking in my
> test_overhead program into the development tree.  I think I can
> pass the compiler switches to get it to generate a profile - at least
> for gcc - but I'm struggling to figure out how to get bjam to invoke
> gprof to display the profile in the output and to make sure I can
> see it in the test matrix.  So you may get your chance to make this
> work for vtune.  I'm surprised that profling / bench marking isn't
> commonly part of the test suites of boost libraries.

Cool,  i'll jump in as soon as you have it uploaded.

cheers

Martin

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.362 / Virus Database: 267.13.0/167 - Release Date: 11/11/2005