[JonWoodruff]

Comparison

Is it just me, or is there new (or maybe guessed) info in here? Some of the info is pretty specific, so it might be NDA breaking stuff. Have you guys seen this stuff before? Especially the Nvidia stuff. (Glide?)

[Geeforcer]

Ahh, good old INQwell, the worst pseudo-news site on the net.

Manufacturer: Nvidia || ATI SIS || Matrox || 3d Labs/Creative ||
Anti Aliasing: Anisotropic filtering || ATI Smoothvion Supersampling || Proprietary SIS || FSA || Edge AA Programmable

Wow, Nvidia new AA is called ...Anisotropic filtering.

[JonWoodruff]

Yes, of course I noticed that. I wonder if the info comes from some small time rumor site. (not theinquirer. That's a big time rumor site.)

It claims 256bit interface for NVidia, along with dual z-buffer, NV AutoShaper(?smart n-patches?), "True Time" processor(?), T&L companion chip.

[Brimstone]

I think those are very false. The R300 shipping next year? I don't think so.

Two chips for the NV 30? I doubt that.

The whole thing comes across as bogus.

[mboeller]

Originally Posted by Geeforcer

Ahh, good old INQwell, the worst pseudo-news site on the net.

Manufacturer: Nvidia || ATI SIS || Matrox || 3d Labs/Creative ||
Anti Aliasing: Anisotropic filtering || ATI Smoothvion Supersampling || Proprietary SIS || FSA || Edge AA Programmable

Wow, Nvidia new AA is called ...Anisotropic filtering.

Maybe they ment an highend filtering algorithm like the FAST-filtering. This form of filtering works better than aniso, and together with MSAA the result could look really good.

[Humus]

More false info: Matrox Parhelia AGP 8x .... it's AGP 4X only.

[Ichneumon]

That table is a bunch of hooey... Oh, there are elements of stuff that's correct in there, but the rest of it is just made up.

For one thing, unless something drastically changes between now and then, r300 will be here in the fall.

[Randell]

For some reason, I've seen it on gaming sites, people have convinced themselves that DX9 will go the same route as DX8. i.e. nVidia launch first Q3 2002 with a DX9 chip and ATI launch second in Q1/2 2003 with a DX9.1 chip. Again I think the Inquirer posted this rumour.

Which goes against everything ATI have ever said about the R300.

[Dave Baumann]

Here's an interesting related snippet from a Clud3D (board vendor) interiew over at Reactor Critical:

Janita Willems: New solutions powered by ATI, such as the RV250 and the R300 will be announced during Computex 2002. Mass production is expected in August 2002, R300 will be announced a week after RV250. The RV250 GPU features integrated Hardware MPEG in the chip and also VIVO functionality will be implemented. R300 is expected to have three times higher performance compared to the competitors.

Somehow I dount those 3X performance claims though...

[Ichneumon]

My guess is that quote comes from the plan to have the r300 out before the NV30... I expect the r300 will be a good 3x faster than GF4s and 8500s for most things.

[Oompa Loompa]

Originally Posted by Ichneumon

I expect the r300 will be a good 3x faster than GF4s and 8500s for most things.

That is not a realistic expectation. But we'll see...

[Ichneumon]

yea... 3x is a lot... perhaps too high. But it isn't unrealistic if it ships at or near the clock speeds/configurations i've heard for the pro version.

Sure it prolly won't be 3x the speed in todays games, but it'll be 3x the speed with effects and features enabled. Everything about the r300 is way over the top compared to the Gf4 and r200. Whether or not the consumer version ships in a config that will perform like that remains to be seen however.

[Hellbinder]

Somehow I dount those 3X performance claims though

I dont think so at all. Ati said months and months ago, that the R300 was designed to be 3x faster than the R200. At least 3x more powerful. Now we are hearing little birdies like this one confirming it. I fully and completely expect the R300 to be 3x faster.

How could it not be with its rumored specs? Plus other "stuff" that Artx added we dont even have rumors about? R200 does 75million pollies peak, does 225 million sound that unreasonable? If it does indeed have 4 programable Pixel shaders and 8 pipes with 4 tmu's each. its going to crush everything in its path.. hands down

I know that there are many of you who will line up to disagree with me. But... Look at the mounting evidence... Even JC chose the R300 to display DoomIII at E3..

[muted]

hey , maybe on paper ..

just like the parhelia .. is 2x faster than the 8500 and geforce 4..


if i take your previous comments about the parhelia , then i guess the r300 is gonna suck

[Dave Baumann]

Originally Posted by Hellbinder[CE

]At least 3x more powerful. Now we are hearing little birdies like this one confirming it.

Confirming or spouting the same party line?

Originally Posted by muted

if i take your previous comments about the parhelia , then i guess the r300 is gonna suck

Who's that aimed at?

[Hellbinder]

Dave,

It was aimed at me i think... But im not sure why... The R300 is going to ROCK based on paper stats. i have been cold on Parhelia from nearly the begining..

Id say spewing... except for the little detail that an internal Ati developer *reportedley* told his friend that the R300 is going to "destroy" everything comming out.

Sure its all speculation, and conjecture at this point.. but.. its more fun that thinking about all the crazy crap going on in the world right now.. like india and Pakistan on the brink of Nuclear war.

[Brimstone]

I tend to think the R300 is going to be very impressive. While obviously it will dwarf a Geforce 4 or Radeon 8500 in raw power, it may end up being the right part with good timing. Of course the NV 30 will be in the same class, but I've read nothing of developers getting early NV30 hardware unlike with the R300. Numereous times I've read that ATI is just waiting for Direct X 9 to be released which to me suggests a chip that has had plenty of time to mature engineering wise.

Nvidia has lots of talented engineers along with a stellar track record, but I'm unsure if the integration of 3dfx may have caused them to lose some of their focus. Anything about the R300 and NV 30 is a lot of guess work right now, so only time will tell.

[JF_Aidan_Pryde]

3x Pipelines of any sort (pixel, vertex) does not mean 3x the performance. Your ATI friends are probably talking about hardware, not real world performance.

And people, I think it's time to wake up from the "blow everything away" dream. It didn't happen with 3dfx, it didn't happen with Bitboys/Matrox, it's certainly not going to happen with ATI. I just don't see David Kirk and Gary Tarolli sitting there, twidleling their thumbs and watching ATI produce something 3x their "most significant contribution since they entered 3D."

In a nutshell - Be realistic; Everyone is cooking something with the same possible ingredience (EDRAM, 256DDR, TBR) . The same resources are avaliable to all players, there wont' be something that the competition "never thought of."

[MDolenc]

If you ask me I'd say the specs on The Inquirer are just a piece of crap...
There is quite some things known about DX9 (even for general public) and if you have not yet seen this presentation from Meltdown last year I suggest you take a look at it. There is clearly said: "32 address ops, 64 math ops" and now R300 "is" going to make 4 such pixels per clock (and NV30 with 2 doesn't sound much more realistic)?? Will each pipeline be able to do 32 math ops? Don't think so...

[Ailuros]

Originally Posted by JF_Aidan_Pryde

3x Pipelines of any sort (pixel, vertex) does not mean 3x the performance. Your ATI friends are probably talking about hardware, not real world performance.

And people, I think it's time to wake up from the "blow everything away" dream. It didn't happen with 3dfx, it didn't happen with Bitboys/Matrox, it's certainly not going to happen with ATI. I just don't see David Kirk and Gary Tarolli sitting there, twidleling their thumbs and watching ATI produce something 3x their "most significant contribution since they entered 3D."

In a nutshell - Be realistic; Everyone is cooking something with the same possible ingredience (EDRAM, 256DDR, TBR) . The same resources are avaliable to all players, there wont' be something that the competition "never thought of."

No particular reason to quote it; just showing that I did take notice

[Ichneumon]

Originally Posted by JF_Aidan_Pryde

3x Pipelines of any sort (pixel, vertex) does not mean 3x the performance. Your ATI friends are probably talking about hardware, not real world performance.

And people, I think it's time to wake up from the "blow everything away" dream. It didn't happen with 3dfx, it didn't happen with Bitboys/Matrox, it's certainly not going to happen with ATI. I just don't see David Kirk and Gary Tarolli sitting there, twidleling their thumbs and watching ATI produce something 3x their "most significant contribution since they entered 3D."

In a nutshell - Be realistic; Everyone is cooking something with the same possible ingredience (EDRAM, 256DDR, TBR) . The same resources are avaliable to all players, there wont' be something that the competition "never thought of."

I was never suggesting it'd be 3x the performance of next-gen cards, and I agree that would just be silly for someone to think that. But 3x the 8500 or 4600 is not entirely unrealistic. Everything about the R300 is (at least in one expected incarnation of it) far over the top in all areas of design over the currently available high-end cards.

[Matt Burris]

A site that threatens me with libel because of a post that stated an opinion that they post rumors, and not to take anything they say without a big grain of salt certainly speaks volumes. Anyone who is that defensive over their site isn't the type of site I visit. I still consider the site and the things they post a joke.

[noko]

Well there are a few things we do know about. The gf3 should be able to do 30FPS at max settings in DoomIII, that is if Carmack prediction or goal was ment. This supposenly was at 640x480x32. Now what resolution and speed was the R300 pulling at E3? This should indicate relative performance difference between a GF3 and the R300. That is if the R300 configuration being used is the one that we can get. The initial developer R300 is probably clocked somewhat slower then the production model which would probably be a newer stepping in addition to a more refined process in chip production. 3x the performance in this case does pan out as far as I can tell.

[pascal]

My guess it was 30 fps at 1024x768x32.

[Captain-Caverne]

Both R300 & NV30 will be three times faster than current top products... but in a very specific game... or should i say "engine"?

At least this is PR claims..., but considering what they (Nvidia & Ati) expect to include into their new chip, it could be true (and what was showed at E3 a few days ago tends to confirm that).