R590 - Will we see this beast?

[Geo]

IT's fairly obvious that Kris didn't have all the facts, or wasn't in full command of them, when he penned that article. (I'm not knocking him in general, just on that one specific piece.) I also think the more obvious comparison for R590 would be R430. If 80m doesn't clock as high as 90nm (I have no idea how low-k or yields figure into this), the safest choice would seem to be to aim for yields at X1900XL speeds.

The reason that article annoys me, Pete, is because I do in fact expect better from them based on historical performance. Talk about a mishmash! What I got was three different strands of rumors mushed into one oatmeal cannon ball and fired against the wall. They have awesome sources; and you, me, and at least 20 or more B3D'ers would have called shens on that article if we previewed it. So, yeah, I'm disappointed and cuffed 'em pretty good.

Re 80nm, what I saw somewhere (Digitimes?) is there are several (3?) 80nm processes of increasing capability. Given the size of R580, it wouldn't surprise me if getting it smaller is more important to them than increasing the clocks right now.

 

[Sunrise]

RV530 is probably replaced by RV535, which I assume is a 80nm RV530? According to the same source the new mainstream chip will be the RV535 while RV505 will be the new low-end chip. Both chips should be out in May.

As for RV560... March first samples. Juli/Aug mass production. Aug/Sep volume in retail.

Your source seems to have AIB contacts. I would hold onto it, because those dates for RV560 seem more like it.

 

[kemosabe]

http://www.hardspell.com/pic/2006/2/21/a5787988-99aa-48c7-bc31-7f626a8eadb3.jpg

 

[Fodder]

This seems to suggest that RV560 and RV570 are one and the same?

Edit: Snap, beaten to it! That'll teach me for taking the time to check something hadn't been posted, only to have it posted while I check!

 

[CJ]

Well guess that kinda confirms everything about RV560/RV570. But ATi is known to change things frequently, so don't be suprised if they make some last minute changes.

 

[neliz]

Well guess that kinda confirms everything about RV560/RV570. But ATi is known to change things frequently, so don't be suprised if they make some last minute changes.

Well, I doubt they change their complete chip after tapeout?

I'd rather say we have a "Pro" and "XT" there..

 

[CJ]

What I meant in particular are the naming-scheme, clockspeeds (and thus performance) and prices, not the core itself.

And I believe there's another slide going around which shows a seperate RV570XT, RV570XL and RV560 including the performancenumbers/percentages that I posted a bit earlier in the thread. I wouldn't be suprised to see it leak its way onto the net sooner or later, just like these 2 slides were already leaked.

 

[neliz]

I wonder if they have the same release date...

 

[Geo]

http://www.hardspell.com/pic/2006/2/21/a5787988-99aa-48c7-bc31-7f626a8eadb3.jpg

Yeah, that's more in line with what I've been expecting too.

 

[Arty]

CJ is right on the mark, Next Gen ATI, On-Chip Crossfire.

 

[Dave Baumann]

By god, these people are slow!

 

[DemoCoder]

You mean ATI took too long to finally put CF on chip?

 

[Tim]

You mean ATI took too long to finally put CF on chip?

No he is talking about the fact that x1300 and x1600, has supported CF without mastercard from the beginning (or from one of the very first Catalyst releases after launch). The x1800 and x1900 could do the same, but in with these cards the bandwidth over PCIe is just to low for them to reach acceptable performance.

Edit: Typo.

 

[Geo]

EB reporting March 23rd launch for X1800GTO and X1900xl.

http://www.elitebastards.com/forum/viewtopic.php?t=13857

Maybe we need a separate thread for these in Graphic Cards by now.

 

[Pete]

I'm in total agreement, geo.

I wonder if doing away with the dongle and compositing chip will impact ATI's SuperAA modes, or if ATI's just going to translate that into a PCB bridge and GPU die space.

 

[NocturnDragon]

No he is talking about the fact that x1300 and x1600, has supported CF without mastercard from the beginning (or from one of the very first Catalyst releases after launch). The x1800 and x1900 could do the same, but in with these cards the bandwidth over PCIe is just to low for them to reach acceptable performance.

There is any chance that we'll be able to enable support for CF without a master card on the x1800 or x1900 using the new RD580? Would those 16 additional lanes be enough?

 

[PurplePigeon]

Gemini mention on digitimes

I assume it's the same "Gemini" that has been mentioned in this thread. From an interview at Digitimes with GeCube http://www.digitimes.com/mobos/a20060222PR202.html

We also have another product we’ll be releasing, a dual GPU graphics card, which we’ve tentatively named “Gemini.â€￾ It’ll be based on the X1000 series GPU from ATI and will have four DVI output ports which can simultaneously display four different DVD streams. Our objective here is to get into the video market. This probably will be available by April after we display it at CeBIT and fine tune the drivers. We’ll probably first launch the X1600XT high-end version for around US$399.

 

[BrynS]

We’ll probably first launch the X1600XT high-end version for around US$399.

Hopefully they can come up with a silent and effective cooling solution. Should be nifty for muli-monitor fans and perhaps prosumer video editing (multiple HD streams, etc).

Some interesting financial specifics were revealed as well:

...This year, about 60% of our revenues from graphics cards will come from mainstream cards, with 5% from the high-end, and the rest from the value segment.
[...]
...In the consumer electronics market you can earn margins of 15% while graphics cards have margins of only 7% since its quite competitive.