rec.autos.simulators

Thief                   - dual chip - 70 fps
                            - single chip - 73 fps

(tested using Fraps)
Both D3D games which use DX6.  Obviously I understand the factors which make
up how a game runs are extremely complex, and every game is different, but
this is still pretty strange.  Surely the dual-chips should be faster in
every circumstance.
 I would guess there is an overhead involved in having the dual-chip setup,
and when a game isn't graphically demanding, this overhead becomes more of a
bottleneck than the actual rendering.  So the dual-chip is mainly of
advantage with visually demanding games like Q3.

The only other possibility I can think of is that the dual-chip isn't
working right yet in some circumstances.  We already know it doesn't work in
D3D at res's over 1280*1024 yet, and that 3dfx are working on a fix for the
next drivers.  Maybe theres a few more problems they haven't found yet.

Proud Voodoo5500 owner

I'll try to work up a FPS matrix next week - I'm taking my PC to a
shooter LAN party this weekend and I'm building a win2000 server for
it, so no GPL til Sunday.

Kendt

Sent via Deja.com http://www.deja.com/
Before you buy.

  This jibes with the notes on the 3dfx V5 newsgroup.  With FSAA off the 2
chip function doesn't seem to be working.    It may only be at higher rez's
or just more noticable at higher rez.  With FSAA enabled you should see the
performance of both chips, but with the added load of the fsaa.
dave henrie

I think it's only at above 1280x1024 - I was benching my card with
3DMark, and there was a clear difference between single-chip and
fastest performance at 1024x768, no FSAA.

Kendt

Sent via Deja.com http://www.deja.com/
Before you buy.

Sent via Deja.com http://www.deja.com/
Before you buy.

Glide - 1024*768 - single chip - 28.3 fps
                                 fastest performance - 26.3 fps

          1280*1024 - single chip - 25.7 fps
                                 fastest performance - 24.9 fps

D3D - 1024*768 - single chip - 23.3 fps
                                fastest performance - 23.9 fps

           280*1024 - single chip - 21.7 fps
                                fastest performance - 23.0 fps

"Fastest performance" is faster than "single-chip" in D3D, but the other way
around in Glide.  It's very strange.  These tests are all done sat on the
Monza grid with 19 opponents in front of me with the in-car view, and the
slider bar on full, and all details turned on.  Basically, the most
demanding test imaginable for GPL.  If it's just me, I get 36 fps all the
time.

I get the same thing on my system, P3-550, 128 MB, V5 5500. Single Chip mode
is about 2 fps faster than Fastest Performance mode in every resolution and
I also believe that GPL runs about the same as my old V3 or a bit slower.
The V5 performance is not bad in the chase view, but I only race in the
***pit view. And what really sucks is after seeing GPL with FSAA it is hard
to go back to not having it, but the performance with FSAA is not acceptable
to me in the***pit view and it should run faster. This has to be a driver
bug of some type along with 2X FSAA crashing the game.

Kyle

I can't help thinking that GPL should be running faster on the V5.  I
certainly expected it to run faster in all cases with both chips, than it
should with just one.  I also expected it to be comfotably faster in GPL
than my old V3 was, but it's the same speed or a litte bit slower.

Simon.