TurboGL Musings

[Gurm]

Well, it wasn't the boost they claimed, but I was only playing at 1024x768 and 800x600 so... overall I'm happy. It DOES seem to have brought up the low end of the average, which is what's important. And they seem to have fixed that texture slider problem in Q3.

So now, instead of mostly getting 50fps with some spots dropping down to 2fps, you mostly get 50fps with some dropping down to 25fps... so overall the average is around 40, very playable considering we're talking about 32-bit, highest texture quality, 1024x768. (Quake3, folks).

In case someone missed it, those of us who have begged, kicked, screamed, or actually coded our way into the developer relations program now have drivers version 5.30.006 available to us. The difference from 5.30.003 is simply TurboGL (and a few little bugfixes).

Oh happy day. Now I can reinstall and let Windows Millenium Beta be the reason for all my crashes.

- Gurm

------------------
G. U. R. M. It's not hard to spell, is it? Then don't screw it up!
The word "Gurm" is in no way Copyright 1999 Jorden van der Elst.

[troop]

That's a positive report I suppose though not an enthusiastic one . My take on the "Turbo" is that while increased speed is nice, texture management (TM) is everything. Since the Turbo supports SIMD as the source of its speed, this driver's speed has apparently got ****'all to do with improved TM. Is that an accurate assessment ?
How does the driver you are testing handle TM compared to the released versions we are using ( beta icd ) ?
Can you compare a HL multiplayer fps for me ?
That's the test I care about.If you have HL/MP , do you still see poor colours , blinking lights , lousy TM ? Until TM is perfect , I couldn't care less about SMID. And ,I imagine , neither could the majority of MATROX gamers who use non PIII machines.

One other question--why is the development of
an OpenGL driver apparently a problem here when other chipsets have decent drivers OOTB ?

Thx,

------------------
In Harm's Way

[Rags]

One other question--why is the development of
an OpenGL driver apparently a problem here when other chipsets have decent drivers OOTB ?

Let's see, the nVidia cards have decent drivers, 3dfx still has yet to build an ICD that is near as complete as either nVidia or Matrox. So where are all of these consumer chipsets (note: 3dlabs does not count) with all of these wonderful ICD's??? An ICD is MUCH harder to code as opposed to coding a D3D driver.

Rags





------------------
FedEx Sucks!

[Wombat]

I heard this nasty rumor from someone on this board that the 5.3 drivers also got rid of that nasty texture corruption/ rainbow effect present in HL, and other Quake-based games. Did you ever have this problem? If it's gone, I may have to go boot into Windows.

-Wombat

[Kruzin]

I have run 6 beta versions of 5.30.xxx so far, and not seen a single texture problem in HL. However, I am on a P2, so not using the TurboGL MCD...

This latest build had more HL bug fixes in the full ICD than anything else. It's looking better than ever (I never had all that many texture problems to start with in 5.2x), and running quite smoothly for me in my TFC games...

[Tumu]

Wombat, I have seen that texture corruption since PD4.51 and ICD beta 2. Maybe PD5.30 will fix it at last.

Kruzin, have you seen anykind of stuttering or stalls in the past releases of PD5 (or in the betas)? I am having them in any game or application that uses either DirectX, OGL or D3D. The stalls happen in windowed mode too if it's available. Windows does work normally if no DX, OGL or D3D app is running. Usually the stall lasts for about 5-10 seconds. I didn't have any stalls with 4.xx series of PD. Though it can be some problem with my installation or machine, but I haven't done any major configuration changes since changing from PD4 to PD5 and the occurances are quite strange.

-Tumu



------------------
Celeron 333 (not overclocked), GigaByte 6BXE Intel 440BX AGPset (bios v3.3),
64MB PC100 RAM, G200 AGP 8MB SGRAM (not overclocked, a fan, bios v2.6),
PD5.13, SB16 Value, 3COM Fast EtherLink XL, Sony Multiscan 100ES,
Win95OSR2.1 finnish

[tish beta2]

Tumu, I too have those stalls.

It was there after a 101% clean install of win98se.

those of us who have begged, kicked, screamed, or actually coded

So it IS possible?! Well then, let the screaming begin.

[IceStorm]

Hey, I think if Creative made a consumer board based on the Permedia 2, then it counts as an average-joe chipset. :-)

So S3's got some sort of OpenGL thingamabob (either that, or no one can play Q3A with it), and ATI's Rage 128 has one. And 3dfx has to have one for Q3A to work, so it looks like there are six OpenGL ICDs out there for the various cards - nVidia, 3dfx, ATI, and S3, Matrox, and Permedia.

Eh, I figure Matrox's OpenGL blows goats, perceptually speaking, because all I've been playing lately is HL and UT, and both are geared/designed/made/fit-like-a-glove with 3dfx's miniGL or GLide. :-)

Just my $0.02.

------------------
Primary System: PIII-540 (450@4.5x120), Soyo 6BA+ III, 256MB PC100 ECC SDRAM, G400 MAX in multi-monitor mode. V2 SLI rig. Two Mitsubishi Diamond Pro 900u monitors, 3Com 3C905, SoundBlaster Live!, DeskTop Theater DTT2500 DIGITAL Speaker System (Sweeeeeet!), 2nd Parallel Port, WD AC41800 18GB HD, WD AC310100 10GB HD, Toshiba SD-M1212 6x DVD-ROM, HP 8100i CD-RW, Epson Stylus Pro, Sharp JX-9400 LJ-II compatible, OptiUPS PowerES 650, MS SideWinder Precision Pro USB joystick, Logitech 3-button mouse, Mitsumi keyboard, Win98 SE, Belkin OmniCube 4-port KVM

Secondary System: PII-266, Asus P2B BIOS 1008, 128MB PC100 ECC SDRAM, Millennium II, 3Com 3C590, ADSL Modem 640kbit down/90kbit up, 3Com 3C509, Mylex BT-930 SCSI card, Seagate 2GB Hawk, NEC 6x CD-ROM, Linux distro S.u.S.E. 6.1 (IP Masquerade works!)

Tertiary System: DFI G568IPC Intel 430HX chipset, P200MMX, 96MB of non-parity RAM, Millennium II, Intel Pro/100+ client NIC, SoundBlaster 16 MCD, Fujitsu 3.5GB HD, WD 1.2GB HD, Creative Dxr3 DVD decoder card, Hitachi GD-2500 6x DVD-ROM, Win98 SE

All specs subject to change.

[Rags]

So S3's got some sort of OpenGL thingamabob (either that, or no one can play Q3A with it), and ATI's Rage 128 has one. And 3dfx has to have one for Q3A to work, so it looks like there are six OpenGL ICDs out there for the various cards - nVidia, 3dfx, ATI, and S3, Matrox, and Permedia.

S3's "ICD" allows very little use with OpenGL games/Apps, my G200 does better, I would hardly call their ICD an ICD. ATI's ICD is not really that terrible, but is a LOOOONG way off from providing the completeness of Matrox or nVidia, and it is more like a miniICD, anyhow. 3dfx has a beta ICD, and for it's official releases it has MiniGL, which is NOT an ICD. Permedia, hehehe, yah you can use it in normal apps, but in games, watch it squirm, and it's performance is on a par with the likes of the G200. SO, I guess that leaves us with the 2 I have given you before: Matrox, and nVidia.

Rags



------------------
FedEx Sucks!

[Gurm]

Rags:

Actually, 3dfx's recent ICD passes more tests on my wife's machine than my G400 ICD does on my machine. The 3dfx ICD runs Glaze and 3D Exercizer, whereas the G400 ICD does neither (Glaze shows a blank screen at a healthy 50fps, and 3D Exercizer crashes hard).

ATI's ICD is relatively feature-complete, but slower than cold molasses. S3 is, I agree, a big ol' joke.

Has anybody tried out a Permedia3 yet? They've been available for months but nobody seems to really care... 8-)

- Gurm

------------------
G. U. R. M. It's not hard to spell, is it? Then don't screw it up!
The word "Gurm" is in no way Copyright 1999 Jorden van der Elst.

[Thorn]

The Permedia3 is VERY slow in games...

But seems to be a great card for professional OpenGL apps. Just another typical 3DLabs product... all work and no (or very little) play.

http://www.anandtech.com/html/review...m?document=999

-Thorn