R300 the fastest for DoomIII, John Carmack Speaks Again

Don't forget that the Radeon 8500's drivers didn't put it ahead until a fair bit after the release of the Ti 500...closer to the release of the GeForce4's.

And, the Tomshardware benches that I quoted were all after the release of the GeForce4, where the Ti 500 most definitely wins if you either throw out the 3DMark score (which is a bad benchmark...3DMark always has been...and yes, I even said that back when nVidia's cards were trouncing everything), or consider the relative performance of the cards (when the Ti500 loses, it's by a tiny margin...).

 

Nobody's found it yet. If somebody does, I'll say that nVidia is cheating. Until then, nVidia's cards have consistently shown to be more stable and higher-performing than ATI's in non-standard benchmarks. A good example was the Radeon 8500 review over at simhq:

http://www.simhq.com/simhq3/hardware/reviews/atiradeon8500/

In these benchmarks, the Radeon didn't perform badly, but it had some major driver issues (And don't forget...it was compared against the original GeForce3...). This review was done in late March...about six months after the release of the video card.

You'd think that issues like he found in these cards would have been resolved by then...

 

The Radeon 8500 was the fastest card on the market especially with Anistropic filtering enabled up Until the Geforce 4...live with it.

 

Why? I posted a fair amount of evidence to the contrary.

And the Geforce3 was most certainly the fastest card on the market with FSAA enabled...and the Radeon's anisotropic filtering is inferior to the GeForce3/4's.

 

For every example Chalnoth you were proven wrong, at resolutions higher than 1024 x 768. I posted screen shots showing just the contrary to two of the games used for benchmarking UT and Max Payne but conveniantly Ignore those

If the Radeon ansitropic is inferior then (objective opinion there since you don't OWN one ) then the Geforce 3's FSAA is inferior, Quincunx is a blurry mess and doesn't support ALPHA textures..there is lots of flaws in Nvidia hardware, Texture compression etc..you choose to overlook them

 

Oh please, it's so blatantly obvious which card is faster, I can't believe you two can't see it. The GeForce 3 Ti500 is faster than the Radeon 8500, but the Radeon 8500 is faster than the GeForce 3 Ti500. It's so obvious, end of story!

 

Again, your comment was that the 8500 was the fastest until the GF4 came on the scene. The truth is, the Radeon's drivers weren't good enough to put it ahead of the Ti500 until *after* the GF4's were released...and even then the Ti500 still comes out ahead at least half the time.

I guess you haven't known me long. I delight in theoretical arguments. The fact is, the Radeon 8500's aniso doesn't support trilinear filtering, nor does it apply aniso to all surfaces that need it.

I don't use Quincunx on my GF4...even though it doesn't have the blurriness problems that the GF3's does. And alpha textures are a non-issue for me. New games will be using alpha blends anyway (Morrowind already does, UT's screenshots show alpha blending), totally eliminating the problem. And why did you post a Voodoo5 FSAA image?

So, tell me, will you still complain about textures that use an alpha test when ATI finally decides to support multisampling FSAA?

 

Were not talking Geforce 4 here are we, Geforce 3 and Ti500...as for the Voodoo 5 shot we need the comparison showing a supersampling offering.
Alpha Textures are not going away anytime soon.

You simply can't say the Ti 500 is faster than a 8500 now can you, and your bold statement saying Nvidia NEVER lost the peformance crown is wrong.

As for ATI's Ansitropic I refer you here to Matt Burris's comments on his 8500 review...Matt likes Nvidia cards since he is the Webmaster of 3DGPU.

http://www.3dgpu.com/reviews/unitechr8500_1.php

 

And yet we do know that the Voodoo5 has generally superior AA to the Radeon 8500...nice comparison.

No, they're not. The problem with multisampling goes away once an alpha blend is used. As I said before, Morrowind uses alpha blends, and UT2k3 will as well. Those games are very important for me. I've also modified UT's OpenGL renderer to support alpha blends (It was very, very easy...).

 

I don't use FSAA, IMO I thinks its a bunch of Cow Poop...I'd rather have High Poly Models and high resolution gaming with Max Anistropic then a few less jaggies..to each there own.

You seem to like to point out flaws in ATI's Ansitropic without owning one, maybe before you claim its inferior by a few games that show a 45 degree plane, maybe you should actually TRY the card and not rely on other people to do your thinking.

 

One thing I do agree is the Voodoo 5 STILL has the best FSAA quality...never disputed that.

 

well

for you guys with the 8500 ... upgrade to dx9... you will see some improvements... i dunno why but there is a post on rage3d talking about it and when i said what the hell and installed i got anywhere from 1%-15% speed increase...now if only new beta drivers will come out...

 

Um people...

JC is the one that said the 8500 has the perfect feature set for doomIII not me.. where the hell is the speculation? He is also the one who has said more than once that he likes several features of the radeon 8500 better but hesitates making it his development platform due to some lingering driver issues.

Please before you all try to make fun of me, or question what i said, check the information yourself.


Chalnoth,

If this is so "right on" then why did JC design his engine around a scheme that favors every aspect of The 8500 pipeline, and not the Gf series? The only thing that saved even a Ti 4600 from losing to a R8500 is some uknown issue, or bug with the 8500. That does not sound like *designed around nvidia" to me. It has even been stated by JC today, that the 8500 score as it is, is so close to a Ti 4600, that it immidiately wins when you turn on Anisotropic. Thats not exactly a *huge* performance lead now is it...

This is just one example of the complete crap that fills his entire statement.

 

Chalnoth, as for your SimHQ example:

Those drivers are OLD, my friend.

Fact is, RIGHT NOW (and for quite some time) the Radeon 8500 is faster in a VAST majority of games than the GF3 TI 500, and in some games, it beats some of the GF4 family (note: GF4MX is not included in the GF4 family, IMO).

Honestly, WHO CARES how the cards performed with 6 month old drivers? We arent living in the past. ATI's drivers are good now. The card is faster, has far fewer problems than it did.

As for my GF3, i hate the damn thing right now, becasue the latest drivers that are "So wonderful" happen to blow balls in UT. And i mean LARGE BALLS. In UT, my GF3 TI 200 (clocked to 215/515) performs slower than my Radeon 32MB DDR (OEM), even though the radeon is on a SDRAM duron machine (kt133, 1ghz), and the GF3 is on a DDR (kt266a, 1700+).
I hardly call that "great drivers".
I wish you would drop the "ATI drivers Sux0r" routine. Its really old, and quite simply NOT TRUE anymore. Every time you bring it up, you hurt your credibility, which is a shame, because you really do have pertinent comments to make on some things - but based on my experiences with reading your posts on nvnews, i tend to skim your posts because i know they will be full of anti-ATI propoganda, along with a healthy dose of reality-denial.

 

Althornin,

Chalmoth said that the bench were taken 6 month after the realease, that makes in march (no?) and then, yes the drivers are old.

But his point was not that they were new, but that 6 month after the release, the R8500 was not as par. That's how i understand it.

And here is some benches taken in mid-april:
(- Drivers Detonator 28.32
- Drivers ATI 7.68 (6.13.10.6043))






Here is the review (in French )

http://www.hardware.fr/html/articles/lire.php3?article=419&page=1

 

Doomtrooper, your video card costs less than $300. Why get so worked up about something so trivial. Chill out. Get laid. This forum is filling up with these rants.

 

I'm a 13 year old adult, unbiased 3d hardware techie trying to make everyone think I'm older than I really am. As I said before, I'm obviously an adult-minded individual that tries to use "big words", "correct punctuation" and "adult" phrases like "my friend," and "unbiased" over and over and over again, not to mention I use quotes every other word to make people think of me as "witty", "intelligent" and "professional", so you will most obviously see how wrong you are and right I am.

Here are my scores:

Remember, as you will plainly see, I'm not biased at all

My Quake3 Scores 1600x1200x32 :

RADEON 8500 = 479fps

GF3 and 4 = 60fps

As you can clearly see, the radeon 8500 is a better choice than shitty NVIDIA's products even though ATi's drivers SuX0rZ0rzTHorZo0ry (see how cool I am with lame low-life internet dork talk? everyone in real life must think I'm the f0nz himself). I'm not baised, it's just that MY card is better.

 

Ugh, where are you getting this from?

 

Yeah, Hellbinder, can you link this "perfect" talk for me? I keep pretty up to date on JC's 3D wheelings and dealings, and I don't recall a mentin of "perfect" or "developed for," especially since his development platform is a GF card! All I saw was him saying the 8500 should be theoretically faster, but is not in RL. He even mentions that the 8500's extra pixel shader range is not necessary, that D3 won't lose IQ due to clamping. He does say that the 8500's range may allow him to optimize the process and gain some speed, but he said the same of the its texture loopback....

Listen, I'd like ATi to pull off the performance crown myself, as their cards seem better all-around than nVidia's. But I haven't seen anything to indicate that. In fact, I'm pretty worried that the R300 was only slightly faster than a Ti4600 (though I don't know how poorly alpha/beta boards bench, so my fear may be for naught). But based on history, I'm expecting nV's new part to be fastest at launch. I'm eager to see if the next gen has enough new features to put aside pure (benchmark) performance as the main concern.

 

Man I love you guys. The 8500 and the ti500 are essentially equivalent. Not only do they split games they split different RESOLUTIONS of singular games (and im talking 1024x768 and up). For every given bench that I can bring up showing an 8500 being the fastest I can bring up another showing the Ti500 taking top place. And vice versa (lest I be accused of being an 'nvidiot'). Saying that the Radeon 8500 was the fastest card is a fallacy. Saying that the Ti500 was the fastest is a fallacy.