>>> use //build switch and if necessary redirect stderr to
>>> stdout by //stderr:1
>> I know these, but couldn't we make it easier?
>> I may look into this later. memtest output can
>> only be screen grabbed.
>So probably we should follow Massimo suggestion and use
>SET ALTERNATE for logging out
On Tue, 23 Sep 2008, Szak�ts Viktor wrote:
Hi Viktor,
>> use //build switch and if necessary redirect stderr to
>> stdout by //stderr:1
> I know these, but couldn't we make it easier?
> I may look into this later. memtest output can
> only be screen grabbed.
So probably we should follow Massimo
Hi Przemek,
I've done the tests, MT speed got ~5% better.
The MT overhead for MSVS 2008 on my system is 41%.
How much do you think this could be further enhanced
in case we'd pass the HVM context to all Harbour
API functions?
ST ST
MT
Hi Przemek,
Okay, I'll redo the tests using DLALLOC only to
keep it simple.
BTW, can we change memtst to output to stdout,
not require key presses, and also for both tests
to include build-time switches in their outputs.
These would ease testing.
use //build switch and if necessary redirect s
On Tue, 23 Sep 2008, Szak�ts Viktor wrote:
Hi Viktor,
Many hanks for your tests.
Now the MT overhead is much smaller. I'll commit last
modification with TLS buffering and this will be the
final version. Probably we can improve the speed
yet a little bit but it will not be anything hardly noticeab
On Tue, 23 Sep 2008, Szak�ts Viktor wrote:
Hi Viktor,
> Pls find attached some speedtst/memtst outputs.
> Looks like dlalloc is well suited for MSVC with
> a significant speedup in ST mode, and mostly
> better results in MT mode.
> Tested with MSVS2008 SP1, XPPro SP3, Intel 865 + P4HT 2.6, 3GB RA
On Mon, 22 Sep 2008, Szak�ts Viktor wrote:
Hi Viktor,
> Many thanks, that solved the problem.
> Shouldn't we make USE_DL_PREFIX the default
> for all compilers/platforms, to avoid such
> possible problems, and also to make this matter
> unified across all our builds?
In fact it's a little bit le
Hi Przemek,
Many thanks, that solved the problem.
Shouldn't we make USE_DL_PREFIX the default
for all compilers/platforms, to avoid such
possible problems, and also to make this matter
unified across all our builds?
Brgds,
Viktor
On 2008.09.22., at 21:17, Przemyslaw Czerpak wrote:
On Mon, 22
On Mon, 22 Sep 2008, Szak�ts Viktor wrote:
Hi Viktor,
> I've now retried after a full rebuild (with same flags
> as in prev message) and I'm getting the same error and
> GPF results with small or even empty test program,
> like "FUNC MAIN()".
> Do you have any tips what could this be? Or how coul
