RE: [GD-Windows] Win2K SP2 debugging slowdown?

[Emmanuel A. <emm...@wi...>]

I have also a strange problem...

My game (it is at a very early stage) runs at 1-2 fps in release mode
when I launch it under the debugger.
When I launch it without the debugger, it runs at 90 fps...

There's no massive deallocations, no use of exceptions...

I know a launch in the debugger should be slower than a real launch, but
not that much ???

Emmanuel



> -----Original Message-----
> From: Javier Arevalo [mailto:ja...@py...]
> Sent: mercredi 3 octobre 2001 14:29
> To: GDWindows
> Subject: Re: [GD-Windows] Win2K SP2 debugging slowdown?
>=20
>=20
> If only it were that easy. :)
>=20
> The situation happens when the game is run under the=20
> debugger, using the
> debug runtime or not doesn't make a difference. The slowdown=20
> magnitude is
> huge, about is a few seconds (5-7) versus a couple tenths of=20
> a second so I
> have ruled out cache, fragmentation and similar issues.
>=20
> Besides, why does it happen only after installing the Win2K=20
> Service Pack 2?
> I guess it has something to do with additional security=20
> features of the SP2
> for processes that run under a debugger, but that's just my=20
> own elaboration,
> and still leaves me with no idea what triggers the slowdown=20
> and how to fix
> it.
>=20
> The hardware configuration doesn't matter, since it's=20
> happened everywhere
> from 2-year-old P3-450's to brand-new 1.4GHz AMDs, 256 or=20
> 512Mb RAM. Heh all
> machines have NVidia cards but I really doubt that has=20
> anything to do, since
> the slowdown happens in areas with zero D3D interaction -=20
> sound and massive
> memory deallocations so far.
>=20
> Javier Arevalo
> Pyro Studios
>=20
> ----- Original Message -----
> From: "Tom Forsyth" <to...@mu...>
> To: "GDWindows" <gam...@li...>
> Sent: Wednesday, October 03, 2001 1:47 PM
> Subject: RE: [GD-Windows] Win2K SP2 debugging slowdown?
>=20
>=20
> > Sounds like the debug runtime is checking every freed block=20
> to make sure
> you
> > haven't scribbled off the ends. When you allocate stuff, it=20
> pads the start
> > and end with some special characters. IIRC, it pads 16=20
> bytes off start and
> > end with FD. When you call free() or delete() or whatever,=20
> it checks that
> > the FDs are still there.
> >
> > I have no idea hy it does this more or less than normal=20
> though - maybe
> > because your memory is heavily fragmented on those runs, so=20
> cleanup is
> quite
> > messy?
> >
> > Tom Forsyth - Muckyfoot bloke.
> >
> > What's he up to now (and can I have a go)?
> > http://www.eidosinteractive.com/downloads/search.html?gmid=3D86
> >
> > > -----Original Message-----
> > > From: Javier Arevalo [mailto:ja...@py...]
> > > Sent: 03 October 2001 08:28
> > > To: GDWindows
> > > Subject: [GD-Windows] Win2K SP2 debugging slowdown?
> > >
> > >
> > > Hi all,
> > >
> > > We have been tracking a weird situation we have, where on
> > > some of our dev.
> > > machines our game takes several seconds to shut down. We
> > > eventually found
> > > out the following:
> > >
> > > - The game must be running in a debugging session under MSVC
> > > (haven't tried
> > > other debuggers).
> > > - The computer must have Windows 2000 _with_ service pack=20
> 2 installed.
> > >
> > > The slowdown happens during a loop that deletes several
> > > thousand dynamic
> > > memory blocks. Most of the time, sometimes it doesn't.
> > > Similar symptoms also
> > > appear in other places, like our streaming music decoding thread.
> > >
> > > Running MSDEV under VTune we found that, when the slowdown
> > > happens, most of
> > > the time during that loop is spent inside the operating=20
> system, spread
> > > half-and-half in two functions: RtlCompareMemoryUlong() and
> > > RtlFillMemoryUlong().
>=20
>=20
>=20
> _______________________________________________
> Gamedevlists-windows mailing list
> Gam...@li...
> https://lists.sourceforge.net/lists/listinfo/gamedevlists-windows
>=20