rec.autos.simulators

> > > Knocking detail down to how I ran on the old 1800/SiS735/Ti4200
machine
> > I'm
> > > getting at best equivalent performance.  Definately noticably worse on
> > > 2400/nForce2/9700Pro than on 2400/SiS735/Ti4200.

> > I'd say your cpu is the bottleneck.

> > I run an Athlon 2800 with a 9700 Pro and my cpu is the bottleneck on my
> > system for nr2003.

> > -Tim

> What kind of performance are you shooting for?
> My Athlon 1900XP (A7V333mb) with GF4 4400 runs satisfactorily (50-80fps)

> I'd try D3D. For me, with the Radeon, D3D is a tad faster in N2k3,
> especially when the fancier graphics features are ON. Or at least, D3D's
> minimum fps is higher than OpenGL's.

> Good luck with your mobo, I hope it's not defective.

> Achim

> > > How far have you gotten meanwhile following the advice you've received
> in
> > > terms of mobo drivers, installing the proper VGart, etc.?

> > Nowhere as yet, but I noticed last night that there maybe a mobo
> > problem - I can only see half my RAM!

> > > Your RAM is in the right slots, and if your CPU is recognised properly
> then
> > > you can't lose that much speed there anymore. You might want to make
> sure
> > > your AGP speed is set to 66, your CPU Interface is set to aggressive,
> and
> > > your RAM frequency has been set to Sync (or 100%, nomenclature differs
> in
> > > some BIOS versions) with your FSB (identical speed - this is faster
than
> > > using asynchronous speeds). You might also want to check in the BIOS'
> > > Advanced Chipset settings whether FSB and CPU Multiplier are set
> correctly
> > > despite them being shown correctly at bootup - just to be sure.
> > > You might have to set System Performance and Memory Timings to 'User
> > > Defined' to be able to adjust the parameters manually. For all I know
> the
> > > 2400 comes with different FSB flavours (133 or 166 I think), so there
is
> a
> > > slight chance of a potential mistake in this area and you should make
> sure
> > > it's all set correctly. I know it isn't very likely, but I'd check it
to
> be
> > > sure.

> > I'm pretty sure my BIOS setting are sound but may have a hardware
> > issue as I noted above :(

> > > I assume you've made sure the in-game settings are identical to what
> they
> > > were under your GF4 when comparing the speeds? The Radeon loses a lot
> more
> > > fps with more cars displayed than the GF cards, and while the end
result
> > > should still be faster than the GF's, this might be worth looking
into.

> > Now THIS is interesting!  I was unaware of this and am gutted if true.
> >  I do all my tests at back of grid at Daytona with *all cars ahead*
> > drawn.

> > N2002: On the 1800/Ti4200 with no reflections min fps is 30.  Putting
> > the XP2400 in the same rig allowed me to turn on relections and not
> > drop below 29fps.  This is running OpenGL.
> > Without reflections on the new kit (2400/nForce2/9700Pro) min fps is
> > just below 40 with same settings.  Switch on reflections and I drop to
> > 24fps!

> > If I can run N2002/2003 all cars ahead, 5 cars behind, full detail
> > cars (with reflections) and world, medium mirrors, 12 sounds heard,
> > 100% draw distance at 1280x960 and *never* drop below 30 fps I am
> > happy.
> > I was so nearly there with the 2400/SiS/Ti4200 I thought I'd be happy
> > as a pig in sh!t moving to nForce2 and 9700 Pro...

> > I can live without the fancy shadow business in N2003, but thought
> > that I *might* have gotten to add shadows onto (but not into) cars.

> > > As a last resort - your 3dMark2001 IMHO looks very low overall. Hence,
> even
> > > if it's faster than your previous one, IMHO it should be around 15000
> rather
> > > than around 12000. So, I'd hazard a guess that there is a slim chance
> your
> > > graphics card doesn't perform as it should. Have you checked the old
> > > graphics card in the new system, to see if the new system is also
slower
> > > with the old graphics card?

> > I'm tempted.

> > > I wouldn't be too worried about clean installs and such (I've used the
> same
> > > Windows installation for 3 years, and even across motherboards with
> > > different chipsets, simply deleting the mobo branch in the system
> manager to
> > > force a re-recognition). If you deinstall the previous drivers
properly
> in
> > > safe mode (does XP have that?) using Window's deinstaller feature, and
> > > explicitely make the graphics card a Default PCI VGA card before
> rebooting,
> > > you should be fine.

> > > You _were_ using WinXP on the old system as well, right? Is it
possible
> that
> > > you've set up WinXP in a different way from before? That you've not
used
> > > optimisations this time which you'd been using in the previous
install?

> > Yes I was using mildly teaked XP home but I found that any tweaks I
> > tried had precisely zero noticale effect.  Perhaps the same tweaks
> > *will* be noticable on the new kit...

> > This whole area is moot until I find out whether I have a hardware
> > fault - which sadly I think looks likely...

> > Regards, and thanks for the in-depth, considered response,

> > Mark

> nForce2 boards are indeed picky when it comes to RAM. Luckily RAM prices
> aren't too high right now.

> As for the driver install order, I once gained 2 or 3 fps by deinstalling
> the Nvidia stuff in Windows' Safe Mode instead of in normal mode, so I'm
> inclined to believe that by doing a perfect install you can speed things
up
> a bit. But I don't think someone could get a boost of 2000 in 3dmark
unless
> something had gone seriously wrong during the previous install. OTOH, if
he
> says so, it may be possible.

> As for your system, you've compared the two systems under the conditions
> most unfavourable for the new system. If you'd compared them with 4xFSAA
and
> 8xAniso enabled, and with maybe just 20 cars ahead, or maybe with 42 cars
> displayed ahead but during the race instead of on the grid, things would
> have looked much different. But, the ATI clearly has its weaknesses as
well,
> this cannot be denied.

> I don't know why OpenGL is slower in N2k3 than D3D, as in IL2-FB for
example
> they're identically fast for me. Maybe it has to do with Papy's
> implementation as well. But, I think that indeed Nvidia's OpenGL
> implementation is very mature and helps make their older cards look
somewhat
> better (faster) as compared to ATI than they really are.

> The ATI drivers in general I think cause a somewhat higher CPU hit than
the
> Nvidia drivers, so ATI gets better with faster CPUs.

> It has to be noted, though, that ATI seems to really take the driver
> development serious at the moment. They keep releasing new driver versions
> about every 6 weeks, a clear indication that they're working ***
> optimising their drivers. I'm sure with time ATI drivers will reach the
> maturity of Nvidia's drivers, and they aren't too far from it right now
> either.

> Maybe also check the forums at http://www.racesimcentral.net/. This seems to pretty
> much be _the_ ATI knowledge ressource on the net right now.

> Achim

> ...
> > Running 2x/2x on 9700Pro is a nice image improvement for a 1fps hit - so
> > that's a positive!

> > My hardware fault is definately a bad DIMM.  It would seem that my old
> > system is less fussy with borderline memory as it can see it, but the
new
> > rig can simply not recognize it at all.  Will not boot up if I use only
> the
> > bad DIMM.

> > I got a private mail from someone who said that *without changing
> [i]any[/i]
> > kit* they boosted their 3Dmark scores by 2000 by changng the order in
> which
> > they installed drivers and directx.  Anyone heard of this?

> > It certainly would appear that there's something going on in these games
> > that benefits the nVidia cards.  Either that or nVidia's openGL
> > implimentation is far superior.

> > Oh well - I'm feeling a little less desperate than I was.