gffx benchies?

[Rambler]

sorry if this link was mentioned somewhere, i didn't notice in that case

http://www.3dcenter.org/artikel/2002/11-19_b.php

they say nvidia provided them the benchmark results - do they sound resonable or did nvidia just took the best case scenario?

edit: now i see that article is from 11/19, nm then

 

[KimB]

Since most of the benchmarks were with antialiasing and anisotropic filtering, they most definitely seem plausible. One interesting thing of note is that the DOOM3 benchmark was not with antialiasing. This is the comparison that I really want to see: GF FX AA vs. Radeon 9700 AA. Just given the hardware, ATI should win this one. But, nVidia has put out more efficient architectures in the past, so we'll have to see.

 

[Mintmaster]

Just given the hardware, ATI should win this one. But, nVidia has put out more efficient architectures in the past, so we'll have to see.

I'm surprised that you think ATI should win this one. Doom 3 depends largely on clock speed because it has both Z-only and stencil-only passes, as well as heavy multitexturing in the final pass which can spread out the bandwidth requirements. In fact, I think Doom 3 is where NV30 will shine brightest.

I know you're talking about AA, but it seems like ATI's compression works quite well given the 9500 Pro benchmarks, where the card only has a 128-bit bus. I think clock speed will still rule with Doom 3, even with AA.

As for more efficient architectures, the Rage128 and Radeon were both significantly more efficient than their NVidia counterparts (in 32-bit, obviously). Unfortunately, they didn't have the clock speed they needed. The Geforce3 was more efficient than the 8500, but not by much. Now, we see the 9500 Pro outperforming the Ti4600 in most benchmarks even without AA (except Q3 and 3DMark), yet the Ti4600 has 20% more bandwidth and 10% higher texel rate, so I would definately say ATI has got their efficiency up.

 

[KimB]

Well, obviously it does shine brightest in DOOM3, without FSAA. We'll see how it does with FSAA. Of course, given nVidia's past, I still think that it will do better. With the Radeon 9700 I currently have, well, the drivers are pretty crappy, and ATI has a long way to go to match nVidia's drivers.

 

[martrox]

Well, obviously it does shine brightest in DOOM3, without FSAA. We'll see how it does with FSAA. Of course, given nVidia's past, I still think that it will do better. With the Radeon 9700 I currently have, well, the drivers are pretty crappy, and ATI has a long way to go to match nVidia's drivers.

No suprise there! I can't help but wonder why I have had so little problems with my 9700's drivers.........[/b]

 

[gkar1]

Thou shalt not heed the f@nboys' drivel

 

[John Reynolds]

Well, I've had considerably more problems with ATi's drivers than I've seen with Nvidia's. NFS:HP2 wouldn't work at all, flashing shadows in Mafia, no AA in some 32-bit games (Arx Fatalis), Sacrifice is a mess (unplayably so), etc. This doesn't mean NV's drivers are perfect either (HoMM 4 wouldn't work at all this spring when using the latest Detonators on my Ti500) and the Detonator 40.00s seem to be generating quite a # of complaints.

 

[martrox]

I had far more problems with my GF3 drivers when I first got the card, it was months before they got them right. And, I still have a GF4 TI4600, and the newest drivers are no better - in fact I'd say worse, than the newest ATI drivers. And I bet the drivers for the NV30 won't be all that great when that card is introduced. This is no abuse on nVidia, as the drivers are still decent.... but it's about time the driver BS about ATI stopped. Why not compare how the drivers were for the GF3 (new tech at the time) with the 9700 when both were introduced, ok? How bout comparing them 4 month after introduction...... To compare 4 month old drivers on new technology with drives that have been around for going on 2 years is bogus, to say the least.

 

[Saem]

I'm pretty sure Mafia's shadow problem was Mafia's shadow problem. ATi released a patch to fix their problem.

 

[T2k]

With the Radeon 9700 I currently have, well, the drivers are pretty crappy, and ATI has a long way to go to match nVidia's drivers.

This is guy is just funny, I think...

I never had any problem w/ latest drivers. Only minors - solved by the right detail settings...

PS: What do you think about the latest 40Dets, huh?

 

[DeathKnight]

What do you think about the latest 40Dets, huh?

The 40.72's suck as do the 41 betas. The 40.41's are by far the best in my experience (they are what drove my 3DMark2K1 score up nearly a thousand points to over 9000 with my 4200).

 

[WaltC]

I had far more problems with my GF3 drivers when I first got the card, it was months before they got them right. And, I still have a GF4 TI4600, and the newest drivers are no better - in fact I'd say worse, than the newest ATI drivers. And I bet the drivers for the NV30 won't be all that great when that card is introduced. This is no abuse on nVidia, as the drivers are still decent.... but it's about time the driver BS about ATI stopped. Why not compare how the drivers were for the GF3 (new tech at the time) with the 9700 when both were introduced, ok? How bout comparing them 4 month after introduction...... To compare 4 month old drivers on new technology with drives that have been around for going on 2 years is bogus, to say the least.

I agree with this, and prior to my purchase and subsequent decision to keep a 9700 Pro (replacing my GF4 Ti4600), I was pretty much an ATI nay-sayer and had vivid memories of Rage Fury driver problems galore (a card which I did return for lack of useability.) I've used the catalysts 6143's (on the install CD)-the current DX9 beta 6228's and was frankly impressed by how "mature" the drivers seemed for a brand-new architecture product. In my system I've had just about 0 problems with this card or its drivers (obvious but not-fatal z-buffer problems in Morrowind is about the extent of the problems I've had.) Testing with some 35-40 3D games since the purchase, I've yet to have one crash and refuse to run, or run in such an unstable fashion that I couldn't play the game. In fact, in my system running my software, if I rated the GF4 Ti4600 as a "100" (I never had problems with it, either, and considering its years of driver development) I would rate the ATI Catalysts overall as a 95 (in many specific cases 100)--but overall 95. For a brand new architecture, not only do I find that remarkable, but considering this is ATI's new architecture we're talking about--almost miraculous...

If ATI keeps this up soon no one will remember how bad their drivers used to be. The last stable drivers I used with the Ti4600 were the 30.82's. I tried the beta 40.42's that nVidia released and they were so bad--so buggy--I went back to the 30.82's (shortly before picking up the 9700 Pro.) It's kind of interesting to speculate, but it's very interesting to see what competition does to companies which are not accustomed to it in their recent histories. I would almost call the 30.82 Dets nVidia's "pre-competition" drivers, and all of the 40's thereafter nVidia's "competitive period" drivers...Sure looks like the competition has 'em pretty rattled as they seem to be making a number of mistakes in their haste to do...something...relating to their drivers. Who knows--maybe the driver guys got an ultimatum from the boss which said: "FIND the performance improvements to propel the ti4600 to the level of the 9700 Pro, or we'll find some new driver programmers"....!... WHATEVER has happened at nVidia, I can't recall seeing this many consecutive buggy driver releases from them--in quite a while, I think.

 

[alexsok]

WaltC:While I mostly agree on your points, I'd still have to disagree about NVIDIA drivers... they were always stable for me, and the situation further improved with the introduction of 40.xx drivers (and 41.xx now). I never had a single problem with their drivers! On all other points I agree with you 100%.

 

[radar1200gs]

I think what you are seeing with nVidia detonators 40.xx and above is preparation for NV30 - for instance in the 41.xx drivers the performance/quality slider is present (adaptive texture filtering) which is part of NV30's intellisample technology.

 

[NV25]

My 9700 Pro has given me almost no problems, the only ones I can recall are...

1) nature test in 3dMark2001 -- this still crashes for me no matter which catalysts I use (but it's not a game so no big deal).

2) Medieval: Total War -- strategy map would not load - however, that has been corrected in the latest drivers.

And... well, that's all I can think of really.

It's fair to say that nVidia's drivers are stable also, but image quality was seriously degraded in 40's dets to increase performance, filtering was set below it's previous level. I didn't like that at all. ATi's quality is so much better I can't believe it. 6xAA + 16xAF makes a game like Morrowind look fantastic, and it still runs faster than what I had before.

In fact, ATi has impressed me so much with the 9700 Pro I'll probably skip the GeForce FX and buy ATi's next product.

 

[alexsok]

It's fair to say that nVidia's drivers are stable also, but image quality was seriously degraded in 40's dets to increase performance, filtering was set below it's previous level. I didn't like that at all.

Degraded? there was a bug in the first release of the 40.xx detonator set that was later fixed in the next detonator set (basically, if you moved the aniso slider to some setting and then back to the original setting, you enabled "point sampling").

Frankly, the quality and perfomance using the 40.xx detonator sets increased over the 30.xx detonator sets!

 

[Reverend]

Well, I've had considerably more problems with ATi's drivers than I've seen with Nvidia's. NFS:HP2 wouldn't work at all, flashing shadows in Mafia, no AA in some 32-bit games (Arx Fatalis), Sacrifice is a mess (unplayably so), etc. This doesn't mean NV's drivers are perfect either (HoMM 4 wouldn't work at all this spring when using the latest Detonators on my Ti500) and the Detonator 40.00s seem to be generating quite a # of complaints.

John, I've seen you post problems with games using the R300 and I am completely stumped. Granted, the shadows in Mafia is a known problem but other games all run fine for me (except, also, for the TruForm bug in UT2003). NFSHP2 runs fine for me ootb.

Have you checked around if the problems you described occurs on systems similar to yours (in particular mobo, type of CPU and soundcard)?

 

[NV25]

Degraded? there was a bug in the first release of the 40.xx detonator set that was later fixed in the next detonator set (basically, if you moved the aniso slider to some setting and then back to the original setting, you enabled "point sampling").

I don't think you understood my point.

Those dets promised an increase in performance, however, the driver "bug" - either by accident or design - improved performance at the expense of quality. Once you moved the slider back into its "original" position, I saw no extra speed at all. Conclusion: nothing gained.

I still see people over at MadOnion praising the 40.xx dets for adding points to 3dMark even though image quality was lousy. And what if the 'bug' wasn't so obvious? Would it even be called a bug, or a feature? imo, it was simply an attempt to boost 3dmark scores.

Anyway, that's history now and I have a 9700 Pro.