Anand talk R580

G72 is nvidias next card ? what will it be ? 90nm G70 ?
When is G80 expected and is there any info at all about it ?
G80 and R600 is the "vista" chips ?
Assume R580 will use GDDR4 so it cant be launched soon, cuz if they do they have to stick GDDR3 on it and people might be upset when 2 month later same card comes with GDDR4.
 
I think the argument that R580, with the same memory and core clocks as R520, wouldn't be dramatically faster than R520, is the crux.

First you have to find games with the heavy duty shaders that are going to use that extra ALU power.

All the evidence we have so far is that R520's texturing, rendering, render-target blending and AA are bringing performance improvements of 10-50% more than an equally clocked X800XT. The performance of all of these isn't going to change in R580, I expect, because they're based upon improvements in texture caching and the memory interface/ring bus.

Texturing may get more efficient, I suppose, because with 3 shader quads sharing a texturing quad the texturing quad's idle time may fall to nearly zero. Depends, really, on the scheduler. Maybe the texturing quads in R520 already have near-zero idle time for the same shader code. I wish we knew more about this...

The lack of the new memory controller in RV515 makes comparisons between RV515 and RV530 pretty awkward - and therefore hard to use as a basis for guesstimating R580 on the basis of R520. And the double-Z ROP capability in RV530 further blurs things. But, I hope Dave's going to do an analysis of RV515 v RV530 at some point, anyway...

Jawed
 
Dave Baumann said:
I think there is a greater chance of R580 + GDDR3 and then the refresh being R580 with GDDR4.

That would indicate to me then that ATI have some confidence now in being able to ultimately get some very good clocks out of 580.

It'll be interesting to see what the first R580 is clocked at....
 
kemosabe said:
A devil's advocate could argue that "in house and working" was exactly the same description used (on the same slide) for R520, RV530 and RV515, all of which are currently in production.

How long before have IHVs prototypes "in house and working" until they reach production-ready silicon? Since when exactly were R520, RV530, RV515 also "in house and working". The second devil's advocate would tell you that major ISVs had pre-production prototypes of R520 from as early as Spring 2004. How long before that did ATI have samples "in house"?
 
Joe DeFuria said:
That would indicate to me then that ATI have some confidence now in being able to ultimately get some very good clocks out of 580.

It'll be interesting to see what the first R580 is clocked at....

Or that it'll end up more bandwidth limited in it's first incarnation then they would have expected. Memory availability is kind of a gamble when you're forecasting IMHLO at least.
 
Subtlesnake said:
Could you come to a rough idea of the performance of the R580, by quadrupling the performance figures for the R530?
You mean tripling (3x the shader units and assuming similar clocks), and then only in synthetic tests, maybe. Future, even more shader-oriented games and benchmarks should tilt in its favor.

The consensus seems to be that current games aren't shader-limited enough for the extra shader power to be of much immediate use. RV530/X1600 is an example of that. It's got gobs of theoretical shader power (almost twice as much as a 6800, AFAIK), but more mundane aspects (TMUs, mainly, AFAIK) are holding it back. It'll probably shine with shader-oriented UE3-gen titles.
 
Pete said:
It's got gobs of theoretical shader power (almost twice as much as a 6800, AFAIK)
They can't really be compared directly, because the usage statistics of the pipelines are so wildly different.
 
Last edited by a moderator:
Samsung Electronics has announced that it had created the world’s first GDDR3 memory chip capable of working at 2.0GHz effective clock-speed. The company has also said its 1.60GHz GDDR3 memory products were available for customers.

Samsung’s GDDR3 memory rated to run at 2.0GHz features 512Mb density and was made using 90nm process technology. The new GDDR3 incorporates a JEDEC standard 136-ball package. Samsung did not reveal when such device is to be available, but indicated that it had initiated mass production of its 1.60GHz 512Mb GDDR3, which was developed in December 2004.

“The 1.6Gbps GDDR3 is available in graphic cards with a maximum density of 1GB by combining sixteen monolithic 512Mb GDDR3s together,â€￾ Samsung said in a statement.

http://print.pcvsconsole.com/?news=3074

Samsung Electronics today unveiled what it claimed to be the industry's first 512Mb DDR2 SDRam fabricated using 70-nanometre processing, the smallest process technology yet applied to a DRam device.

According to the Korean giant, the 70nm technology maintains continuity with the existing 80nm and 90nm processes its currently uses in most DRam production today.

However, the number of chips yielded per wafer will be "at least 100 per cent higher" than could be obtained with 90nm technology, the firm reported.

http://www.vnunet.com/vnunet/news/2143948/samsung-claims-dram
http://www.techworld.com/storage/news/index.cfm?NewsID=4574&inkc=0
 
Last edited by a moderator:
Subtlesnake said:
Could you come to a rough idea of the performance of the R580, by quadrupling the performance figures for the R530?
Interesting idea.To get reasonable results available bandwidth should be scaled back aswell. Does the RV530 have 2 z-units capable of operating on two z-pixels at a time (like NVidias architecture) vs. R520 with 2 z-units operating on 1 pixel? If so then atleast 2xMSAA should be used to make the results more useful.
 
Ailuros said:
How long before have IHVs prototypes "in house and working" until they reach production-ready silicon? Since when exactly were R520, RV530, RV515 also "in house and working". The second devil's advocate would tell you that major ISVs had pre-production prototypes of R520 from as early as Spring 2004. How long before that did ATI have samples "in house"?

ISV did not have R5xx prototypes in 2004 at all, actually Ati was not able to ship working samples before the soft ground issue was fixed - pre-fix chips was simply to flaky.

From the day where they decide to start volume production to the day they are actually ready to ship, it is about three months. It is very hard to say how long it takes from a “working” prototype until release, if they need a another silicon respin it could take close to six months before they are able to actually ship in volume, but a new metal layer steppings might have very little impact on the time table.
 
Dave Baumann said:
I think there is a greater chance of R580 + GDDR3 and then the refresh being R580 with GDDR4.

I was thinking about that and time frame. If there is a 580 +GDDR3 then a refresh with GDDR4, when do you think that refresh is scheduled?

For now it seems that cards with the R580 will not be out before end of Q1 and I was under the impression that R600 was scheduled for Q4 2006. So there is little room for the GDDR4 refresh no ? Unless one of theses chips is coming out as the top end card without replacing a previous card ?
 
PatrickL said:
I was thinking about that and time frame. If there is a 580 +GDDR3 then a refresh with GDDR4, when do you think that refresh is scheduled?

For now it seems that cards with the R580 will not be out before end of Q1 and I was under the impression that R600 was scheduled for Q4 2006. So there is little room for the GDDR4 refresh no ? Unless one of theses chips is coming out as the top end card without replacing a previous card ?


I think what Dave meant was there's two possible situations for the refresh part. Situation A is R580+GDDR3. Situation B is R580+GDDR4. He indicated Situation A is more likely than Situation B.
 
BRiT said:
I think what Dave meant was there's two possible situations for the refresh part. Situation A is R580+GDDR3. Situation B is R580+GDDR4. He indicated Situation A is more likely than Situation B.
Considering situation B would be set for late next year, yeah, I'd say situation A is a bit more likely.
 
Back
Top