R300 - Final Specs?

[alexsok]

Taken from Nvnews:

R300 Specifications - 7/15/02 10:33 am - By: Typedef Enum - Source:
The Inquirer has posted what appears to be the final specs on the R300. In addition, it now seems like a virtual lock that ATI is, indeed, going to unveil the R300, along with the RV250 on the 17th of July.
The basic specs look like this:

315 MHz core clock

8 Pipelines

107 million transistors

.15u technology

External power (a-la 3dfx)

Four Vertex Shader units

Pixel/Vertex Shader 2.0 Support

VideoShader (accelerate MPEG video)

16X Anisotropic Filtering

~15,000 3DMarks (yeah!)

400 million poly/sec. throughput

If ATI is true to their word, the R300 should become available within a minimal amount of time after being launched. The speculation is that late August to early September would be a good ballpark.

The announcement July 17th?! (that was already noted in one of the posts here, but still...)

Exciting times are ahead!!

Edit:
For anyone intrested, full news here:
http://www.theinquirer.net/?article=4431

Intresting Notes:

The card will have support for 16X anisotropic filtering and will be able to deliver 400 million polygons/s, which, even to a spod like me, seems to be bverging on overkill. If we assume that we need 14 FPS you will be able to get 16,6 million polygons per frame. That is amazing number, if we may say so. Today's top fancy demos use less than one million, so you can imagine how good this card will be.

Performance-wise, as we said you will be able to see 15,000 3DMarks on a P4 platform, where Geforce 4 would not give more than 12,000. We are talking about non-overclocked systems, and recognise you can get even more with faster Pentiums and overclocked cards.

 

[Guest]

400 MILLION triangles?!!??? Holy crapola...... The NV25 only does like 125 Million. That does seem like an awfull pile of triangels doesn't it?

 

[MuFu]

Yes... could be effective throughput after tesselation though.

This is straight from the source: The R300 is easily capable of 300MHz+. In fact, some samples are running at 325MHz without any problems. The A11 revision had a bug in the memory interface that seriously hindered its operating frequency. A12 does not have this problem and they have ramped using the A12 mask.

The AIW will most definitely run at the same speed as the standard card. This should be available near the end of August, while the AIW version will arrive in September sometime, although the dates are unclear to me.

It's a pleasure to be bringing you this 100% confirmed SWEETNESS (courtesy of nobody in particular, hehe).

MuFu.

 

[Guest]

Yes... could be effective throughput after tesselation though.

This is straight from the source: The R300 is easily capable of 300MHz+. In fact, some samples are running at 325MHz without any problems. The A11 revision had a bug in the memory interface that seriously hindered its operating frequency. A12 does not have this problem and they have ramped using the A12 mask.

The AIW will most definitely run at the same speed as the standard card. This should be available near the end of August, while the AIW version will arrive in September sometime, although the dates are unclear to me.

It's a pleasure to be bringing you this 100% confirmed SWEETNESS (courtesy of nobody in particular, hehe).

MuFu.

A card running at 325 MHZ at the .15micron process and 107 million transistors has got to be "the hottest PC hardware available". I wonder if they have adopted some sort of way to keep it cool?

I would imagine that this card will have some honking fan a la geforce 4 ti cards do. If not ATI has certainly managed to do something that nvidia can't do. Create a 100 million + transistor that runs at .15 micron that still runs relatively cool. I suspect that they will have extra cooling attached given the incredable mass of transistors at .15.

325MHZ is a higher clock rate then the geforce 4 ti cards default, for some reason ATI manages to create cooler running hardware compared to what nvidia can muster. I would suspect that this is why nvidia is insisting on the .13 micron process for the NV30.

 

[Geeforcer]

If R300 is indeed 107M/.15 @ 315MHz, one has to wonder WTH did Matorx do wrong? Its not like there has been a major breakthrough with .15 process in a month following Parhelia's release.

 

[Bambers]

Ati chips do seem to be much cooler, I was suprised with my 8500 against my old gf256. The heatsink in those released photos is pretty big though.

 

[Guest]

Ati chips do seem to be much cooler, I was suprised with my 8500 against my old gf256. The heatsink in those released photos is pretty big though.

Yeah they do look big but in a different sort of way it is almost as though they are cooling something on the perimeter of the chip. Any guesses as to what that would be? I heard a lot of talk some time ago now that ATI was considering embedded SRAM. I know that ARTX has experience in this area a la the NINTENDO gamecube. But I wouldn't pretend to know anything about this sort of thing it is just a guess at what is on the perimeter of the actual chip. It could be a lot of things, no real way to know why the fins on the heat sink seem to cover a larger area then normal, is the chip that big?

 

[Ascended Saiyan]

400 MILLION triangles?!!??? Holy crapola...... The NV25 only does like 125 Million. That does seem like an awfull pile of triangels doesn't it?

I think you mean 400 million Polygons/sec actual triangles/second would be about much much lower than that,~100 Million triangles/sec which is still high if you ask me.

 

[Guest]

400 MILLION triangles?!!??? Holy crapola...... The NV25 only does like 125 Million. That does seem like an awfull pile of triangels doesn't it?

I think you mean 400 million Polygons/sec actual triangles/second would be about much much lower than that,~100 Million triangles/sec which is still high if you ask me.

Yeah I got a little too excited there. Doe!!!!

 

[Gollum]

polygons and triangles are basically the same thing, both are 2-dimensional shapes, only that a polygon can have more than 3 points. what you were getting at was probably points or vertexes/sec. 400mil vertextes/sec would be the same as 133mil triangles/sec...

 

[Guest]

polygons and triangles are basically the same thing, both are 2-dimensional shapes, only that a polygon can have more than 3 points. what you were getting at was probably points or vertexes/sec. 400mil vertextes/sec would be the same as 133mil triangles/sec...

Yeah I thought that a triangle was a primative poly or something like that gesh. I think I better take a course in comp graphics before I post something. Man... I am always it seems second guessing what it is that I know and just when I think I got something figured out it turns out backwards or something. CRAP. Anyone have an idea as to why the R300 heat sink seems to cover such a large area?( Other then the obvious posibility that the chip is that freaking big.)

 

[Steve]

400mil vertextes/sec would be the same as 133mil triangles/sec...

Not really. Most of the vertices are shared in a highly optimised mesh. I'd wager that the real time number would be somewhere between 200 and 300 million triangles per second.

 

[Edward]

hope trilinear and aniso will be supported now.

And I would like to know if 60+ fps at 1280x1024x32 with16x aniso in UT2003 is possible

 

[Pete]

I'd say the heatsink is so big for the simple reason that the chip will be FREAKING HOT, and, as thus, a larger heatsink will help dispel the heat more quickly.

Is there a good reason why video card makers haven't moved AGP card cores to the top of the chip, for better cooling? The CPU and video card fans are blowing in different directions, and the CPU is usually well cooled, so I wouldn't think heat rising from the video card would be an issue. It does so anyway, and the PSU pulls (hot) air over the CPU either way.

 

[Thowllly]

Is there a good reason why video card makers haven't moved AGP card cores to the top of the chip, for better cooling?

Coz they aint allowed to, according to AGP specs, or sumthin...

 

[Bambers]

Can't understand the ATX mobo spec personally. It seemed much more sensible with the old AT spec where the cpu was in the bottom corner having COLD air blown over it from the system fan on the front of the case.

Now its stuck inbetween a hot PSU and a hot gfx card with hot air floating in from all the drives

 

[Geeforcer]

Not really. Most of the vertices are shared in a highly optimised mesh. I'd wager that the real time number would be somewhere between 200 and 300 million triangles per second.

Exactly. Below in an example of a simple, poorly drawn mesh, that nonetheless uses less then 2 vertices per triangle (26/14).

 

[David G.]

The cooling system is so big because the chip is @ 0.15 micron !

What's so hard to understand ?!

Sure ATi has done an amaizing job here but I didn't expect this chip to come with passive cooling ....

And also I don't understand why are you so amaized about ATi chips being cooler ... sure R300 is a wonderfull exception but isn't ATi the best specialist in low consuming mobile chips ? Isn't ATi the one with 86% market share in mobile systems ? ( actualy now it's around 60-70% ... but it was 86 last year )

 

[Guest]

So the triangle count is somewhere in between 200-300 million(WOW!) where the geforce 4 ti is about 125 million. Right?

 

[Guest]

The cooling system is so big because the chip is @ 0.15 micron !

What's so hard to understand ?!

Sure ATi has done an amaizing job here but I didn't expect this chip to come with passive cooling ....

And also I don't understand why are you so amaized about ATi chips being cooler ... sure R300 is a wonderfull exception but isn't ATi the best specialist in low consuming mobile chips ? Isn't ATi the one with 86% market share in mobile systems ? ( actualy now it's around 60-70% ... but it was 86 last year )

I am not "confused" boss it is just that the heat sink covers a wider area then others. I was thinking that there was actually something under the wider area possibly. But if I understand anothers suggestion correctly its just wider to dispurse the heat better and not nessesarily that there is something directly underneath the fins themselves found on the perimeter of the GPU. I was just asking why the heatsink covered such a large area and someone gave me a logical answer... no confusion, just curious.

Yes ATI does have the lower power consumption technology. Nvidias solution ("power miser" or something like that.) is a mode that you put it into to keep it from using more power, but IIRC it also takes a drop in performance when doing this.(I think.) Where ATIs technology just straight out uses less power resulting in cooler running GPUs.