PDA

View Full Version : Fuad confirms R800 basic info

Pages : [1] 2 3 4 5
josiahsuarez
05-26-09, 02:19 AM
nothing really new here, basically what was already assumed but now it's "confirmed"

http://www.fudzilla.com/content/view/13832/1/
R800, DirectX 11 is dual chip
Written by Fuad Abazovic
Monday, 25 May 2009 08:35


Comes in late 2009


We just got one interesting news confirmed. ATI's DirectX 11 high-end card, codenamed R800 is made of two 40nm GPUs. This doesn’t really come as a big surprise, but we at least managed to confirm this one. ATI is continuing its X2 roadmap and won't change its direction anytime soon.

RV870 is a single 40nm DirectX 11 capable chip while R800 will be made of two of them. R800 should be competing with Nvidia’s GT300 chip and at this stage, we doubt that anyone knows who should win this performance crown fight but the hopes are high.

GT300 should also be on the schedule to launch in late 2009 but we’ve heard before that ATI hopes to be the first DirectX 11 on the market, but we are quite sure that both next generation chips and cards will launch close to each other.

so apparently Fuad doesn't know if R800 or GT300 will be more powerful. interesting, perhaps it's still anyones ballgame.
nekrosoft13
05-26-09, 02:28 AM
r800 to be dual GPU, while GT300 to be single GPU, and both might be around the same performance.

so my guess is that nvidia might keep the crowd, since they can also go dual GPU.
josiahsuarez
05-26-09, 03:07 AM
I wonder if a standard GT300 would work for a GX2 type card. AFAIK GTX295 dies were around 400mm² @ 55nm, and GT300 is around 500mm² @ 40nm.
Ninja Prime
05-26-09, 07:56 PM
I wonder if a standard GT300 would work for a GX2 type card. AFAIK GTX295 dies were around 400mm² @ 55nm, and GT300 is around 500mm² @ 40nm.

Probably not unless they trim it down, like they did with the GTX295 compared to the original GTX280.
zoomy942
05-26-09, 07:59 PM
wow. thats impressive. and my lowly pc just got an ATI 4670 :(
rage10
05-27-09, 12:22 AM
I wonder if a standard GT300 would work for a GX2 type card. AFAIK GTX295 dies were around 400mm² @ 55nm, and GT300 is around 500mm² @ 40nm.

you sure? I thought that the gtx 295 die was 495mm² at 55nm. the reson that the gtx 295 is not 2 gtx 280 was due to power and cost restraints.
josiahsuarez
05-27-09, 04:17 AM
you sure? I thought that the gtx 295 die was 495mm² at 55nm. the reson that the gtx 295 is not 2 gtx 280 was due to power and cost restraints.

nope I'm not sure. I don't think nvidia ever publicly disclosed the die size, your number might be more accurate.
mtl
05-27-09, 07:36 AM
Nvidia also claims to have dramatically increased clock per clock and shader efficiency on the 300. Could be interesting.
Ninja Prime
05-27-09, 09:31 PM
Nvidia also claims to have dramatically increased clock per clock and shader efficiency on the 300. Could be interesting.

They also claimed that before the GTX280 came out and it turned out to be mostly hot air.
hell_of_doom227
05-27-09, 10:25 PM
They also claimed that before the GTX280 came out and it turned out to be mostly hot air.

GTX280 is superior in every possible way then any ATI product out there, so in my book it wasn't hot air.
Xion X2
05-27-09, 11:35 PM
GTX280 is superior in every possible way then any ATI product out there, so in my book it wasn't hot air.

*snort*

How is it 'superior in every way' to the 4870X2 which is faster or even the single-GPU 4890?

http://img41.imageshack.us/img41/9956/far1920.gif (http://img41.imageshack.us/my.php?image=far1920.gif)

http://img41.imageshack.us/img41/3872/cry1920.gif (http://img41.imageshack.us/my.php?image=cry1920.gif)

http://img41.imageshack.us/img41/2638/fall1920.gif (http://img41.imageshack.us/my.php?image=fall1920.gif)
mtl
05-28-09, 12:48 AM
this review Nvidia's second fastest single GPU card as better than the 4890.
http://www.fudzilla.com/content/view/13094/40/1/2/
Gainward BLISS GTX 275 is a card we’d recommend to anyone craving gaming at 1920x1200 or higher. So let us recap - superior performance, good thermals and noise levels, excellent overclocking potential and for many the most important - great pricing for a high end card is what makes this card a Fudzilla recommended piece of equipment. Don't forget that the HD 4890 is also a great card, albeit a bit slower,
bacon12
05-28-09, 01:00 AM
*snort*

How is it 'superior in every way' to the 4870X2 which is faster or even the single-GPU 4890?



Because its a single gpu. That graph conveniently is missing sli benches. 4890 was just released what 45 days ago?
Bman212121
05-28-09, 01:15 AM
r800 to be dual GPU, while GT300 to be single GPU, and both might be around the same performance.

so my guess is that nvidia might keep the crowd, since they can also go Tri-GPU.

Fixed.

If R800 and GT300 are around the same performance, then GT300 should be the absolute winner since they can slap 3 of them together. We'll have to wait for jAkUp and Kasu Code to know for sure though. :p
Ninja Prime
05-28-09, 05:28 AM
GTX280 is superior in every possible way then any ATI product out there, so in my book it wasn't hot air.

What does that response even have to do with what I said? You're such a fanboy that you just post random crap pimping NV after people make comments? :rolleyes:
Xion X2
05-28-09, 12:04 PM
this review Nvidia's second fastest single GPU card as better than the 4890.
http://www.fudzilla.com/content/view/13094/40/1/2/
Gainward BLISS GTX 275 is a card we’d recommend to anyone craving gaming at 1920x1200 or higher. So let us recap - superior performance, good thermals and noise levels, excellent overclocking potential and for many the most important - great pricing for a high end card is what makes this card a Fudzilla recommended piece of equipment. Don't forget that the HD 4890 is also a great card, albeit a bit slower,

Did you just quote FUDZILLA!?

:udawg:

You are a trip, MTL. The only time you ever pop your little head out of the woodwork is to challenge something positive said about ATI, and when you finally do, you quote FUDZILLA as a reliable source of information. Too funny.
Xion X2
05-28-09, 12:06 PM
Because its a single gpu. That graph conveniently is missing sli benches. 4890 was just released what 45 days ago?

He didn't say "single GPU"; he said "any ATI product." Also, the 4870X2 is a single card although it is dual-GPU and is considered a "product" of ATI. So according to what he said, he's wrong.

And I said nothing about Crossfire being faster than SLI so you had no reason to bring SLI into the discussion.

Here's some more benches to chew on.

http://img15.imageshack.us/img15/2509/4890vs280crysis.gif (http://img15.imageshack.us/my.php?image=4890vs280crysis.gif)

http://img15.imageshack.us/img15/5672/4890vs280fallout3.gif (http://img15.imageshack.us/my.php?image=4890vs280fallout3.gif)

http://img15.imageshack.us/img15/7067/4890vs280fear2.gif (http://img15.imageshack.us/my.php?image=4890vs280fear2.gif)

http://enthusiast.hardocp.com/article.html?art=MTYzNiw3LCxoZW50aHVzaWFzdA==
mtl
05-28-09, 01:18 PM
Did you just quote FUDZILLA!?

:udawg:

You are a trip, MTL. The only time you ever pop your little head out of the woodwork is to challenge something positive said about ATI, and when you finally do, you quote FUDZILLA as a reliable source of information. Too funny.

Spoken like a liberal. When the facts don't support their argument, They start namecalling and scream louder. Pathetic.
Xion X2
05-28-09, 03:50 PM
Spoken like a liberal. When the facts don't support their argument, They start namecalling and scream louder. Pathetic.

Irrelevant. The 'facts,' as you call them, are not in your corner. You quoted Fudzilla as a reliable source. That's what's 'pathetic.'
ChrisRay
05-30-09, 08:00 AM
They also claimed that before the GTX280 came out and it turned out to be mostly hot air.

Not really. Nvidia claimed the GT200 had far superior geometry shader performance compared to the G92/G80. And it does. The difference between GT200 and G92/G80 when doing Geometry shader tasks is rather significant. The only other claims to improved shader efficiency Nvidia ever claimed was that register usage was better on the GT200 which could result in 10% increase in a shader to shader/clock per clock scenerios. The benefit was about 2-8% in reality when you took a GT200 and cut its shader/ROPs/TMUS to about 8800 GTX levels. Such comparisons are really difficult to make though. Since beyond very closed testing. Its impossible to make a GT200 card behave like a G80 card due to the different way TMUS/ROPS are balanced between the two cards. GT200 has a higher shader to texture ratio and the texture map units are linked to the shader core.
Lfctony
05-30-09, 12:42 PM
Unless the GT300/R800 offer double the performance of my single GTX280 on one chip, I'm not even bothering. I already went SLI and came back, single chip or no upgrade for me...
Ninja Prime
05-30-09, 06:42 PM
Not really. Nvidia claimed the GT200 had far superior geometry shader performance compared to the G92/G80. And it does. The difference between GT200 and G92/G80 when doing Geometry shader tasks is rather significant. The only other claims to improved shader efficiency Nvidia ever claimed was that register usage was better on the GT200 which could result in 10% increase in a shader to shader/clock per clock scenerios. The benefit was about 2-8% in reality when you took a GT200 and cut its shader/ROPs/TMUS to about 8800 GTX levels. Such comparisons are really difficult to make though. Since beyond very closed testing. Its impossible to make a GT200 card behave like a G80 card due to the different way TMUS/ROPS are balanced between the two cards. GT200 has a higher shader to texture ratio and the texture map units are linked to the shader core.

They claimed 30%, by your own words, it was 2-8%. Mostly hot air.
nekrosoft13
05-30-09, 06:49 PM
He didn't say "single GPU"; he said "any ATI product." Also, the 4870X2 is a single card although it is dual-GPU and is considered a "product" of ATI. So according to what he said, he's wrong.

And I said nothing about Crossfire being faster than SLI so you had no reason to bring SLI into the discussion.

Here's some more benches to chew on.

http://img15.imag eshack.us/img15/2509/4890vs280crysis.gif (http://img15.imageshack.us/my.php?im age=4890vs280crysis.gif)

http://img15.ima geshack.us/img15/5672/4890vs280fallout3.gif (http://img15.imageshack.us/my.php?ima ge=4890vs280fallout3.gif)

http://img15.image shack.us/img15/7067/4890vs280fear2.gif (http://img15.imageshack.us/my.php?image=4890vs280f ear2.gif)

http://enthusiast.hardocp.com/article.html?art=MTYzNiw3LCxoZW50aHVzaWFzdA==

you do realize you are comparing a card that is what 10+ months old to someting that is little over a month old?
sammy sung
05-30-09, 09:54 PM
you do realize you are comparing a card that is what 10+ months old to someting that is little over a month old?

Why shouldn't he ?

since hell of doom said this and pay attention to the is,not was :

GTX280 is superior in every possible way then any ATI product out there, so in my book it wasn't hot air.

He is claiming that the 280 gtx is superior in every possible way to ANY ATI product out there (wich is untrue if that was what he meant to say) Xion is just proving him wrong,dunno why you can't see that.Sure ,the 280 is 10 months old and was superior to every single chip sollution ATI had at that time,it sure as hell isn't superior to every damn ATI product out today.
ChrisRay
05-31-09, 12:23 AM
They claimed 30%, by your own words, it was 2-8%. Mostly hot air.

No they didn't. I was actually at editors day where they explained it to the editors. Were you? Nvidia never tried to make the GT200's new shader efficiency any higher than it was. They were neither hyping it nor were they making a big deal out of it. They were noting dual issue and the fully functioning Mad + Mul unit and moved on. The geometry shader performance. They did hype however. I happen to have documents of most of all Nvidia's launches and editors day in PDF format so I know what they were claiming. Heres Nvidia demonstrating the dual issue registry improvements. They showed a heavy register usage situation and made a comparison of what the fully exposed Mul can do ((hell if you know how the Mad + Mul works you'd know that they could never claim a 30% increase from it)). Nvidia's own reviers guide and editors slides never made such a claim. Again the only place they claimed much better performance was in the geometry shader. In Which case the GT200 series is quite efficient compared to the Geforce 8/9 series. Anyone who claimed 30% increase either didn't understand the documentation they were given or just wrote whatever they felt like.


Sorry but your information is inaccurate and wrong.



http://chrisray.soterial.com/chrisraymisc/200register.png