From: Paul A. Bristow (boost_at_[hidden])
Date: 2002-06-25 04:07:59
Yes, I have finally managed to compile all the tests and bench files (and
concept) with MSVC.net Standard edition (no otpimisation) and all run OK and
debug and the bench tests in release. I am still working on the tests using
release mode. There is a modest speedup in release, factor of a few say for
some benchmarks, no change for others.
But I had to change several test properties and had great trouble with
/Zm1000 is enough for all.
However I suspect a fault in the compiler /Zm setting business.
Once the heap overflow is triggered, it seems to fail whatever the /Zm
(it doesn't start compiling, but immediately bombs out).
I got rather confused but eventually suceeded.
I think that restarting the session after changing to /Zm1000 is needed.
Perhaps a clean does it too - I was puzzled.
No warnings at level 3, but need to enable MS language extensions.
Prpper for conformance can be on.
I am still working on this.
Dr Paul A Bristow, hetp Chromatography
LA8 8AB UK
+44 1539 561830
Mobile +44 7714 33 02 04
> -----Original Message-----
> From: boost-bounces_at_[hidden]
> [mailto:boost-bounces_at_[hidden]]On Behalf Of Martin Schuerch
> Sent: Tuesday, June 25, 2002 6:41 AM
> To: boost_at_[hidden]
> Subject: [boost] Problems with uBLAS and /Zm Option (VC7)
> I tried to run the test programs of ublas under VC7. I used the converted
> workspace from the workspace of VC6.
> -> Compiler errors due memory limitations (1GB installed).
> The suggestion of the online help is to set a bigger number in the /Zm
> In the included projects a number of 1000 (2000 is maximum) is
> already set.
> Increasing the number doesn't help.
> Deleting the already set option does help for most test-files.
> For the other
> files a number of 500 will succeed.
> I would like to replace our home made matrix class with the uBLAS one
> (during the review). But have a little bit a bad feeling, because
> I wouldn't
> like to play around with the /Zm option in every file.
> Has anybody experience with this?
> Best regards
> Unsubscribe & other changes:
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk