The NEXT LAST R600 Rumours & Speculation Thread [Archive] - Page 5

Bouncing Zabaglione Bros.

21-Mar-2007, 15:42

That's it. I'm neg-repping the next person to link to FudZilla. Who's with me???!!! :twisted:

I've actually been thinking about this for a while, but I suspect I'll get spanked for misuse of the rep system. I mean TheInq is a well known gossip site and all, but most of the criticism it got from this board was down to Faud's inane and incorrect rambling. Now Faud's gone, people have just followed him and keep linking to the same old rubbish.

It's not like this is insider info, this is just total rumours that he keep splurging out. They are just unfounded guesses pretending to be based on some kind of secret knowledge. Heck, Faud's new site makes TheInq look like a paragon of objective and accurate reporting.

Take a look back at Faud's old R600 news. He swings from one extreme to another, completely contradicting his previous stories. If he knew anything, he wouldn't have to completely backtrack over what he said last week.

1BFC

21-Mar-2007, 15:46

You know I must say that all the delays and lack of any type of credible benchmark info for the R600 is causing me to delay upgrading my very aging Athlon XP + 6800GT system for months now. I understand that R600 is not released and what not, but at this point I think it's starting to hurt AMD that they don't have a timely "leak" or SOMETHING.

I have, for the first time I can remember, 5 people all waiting with me basically telling me "When you build your system, build mine too because we are tired of waiting". Everyone has their own personal point where they are no longer willing to wait for something. I think a lot of folks are close to that point and AMD will miss their opportunity for a sale with a lot of the upgrade market if they don't have some credible and meaningful info come out soon that will give the folks waiting a good to reason to wait.

Personally, I say a week or two more is all they got. At least within my circle of people.

IMO of course.

-BFC

vertex_shader

21-Mar-2007, 15:50

Fuad says the ATI partners have "lost the fate".

http://www.fudzilla.com/index.php?option=com_content&task=view&id=224&Itemid=1

How about you? You still think R600 is worth waiting for? I do. But it's getting harder and harder to resist buying an 8800. They're almost giving the 320 MB GTS away for free now. :???:

Buy the GTS, if the R600 a beast you can still sell your gts for 98$ :lol:

Launch coming day to day nearer, the FUD rate will be higher day to day, almost all golden classic come out already :smile:

vertex_shader

21-Mar-2007, 15:56

Take a look back at Faud's old R600 news. He swings from one extreme to another, completely contradicting his previous stories. If he knew anything, he wouldn't have to completely backtrack over what he said last week.

Hexus/Vrzone is the same, but the fun factor still high :smile:
Someone need to check what Dave Orton doing in his office, packing his things, or drinking champain, this say more than 1000 words about R600 :lol:

Razor1

21-Mar-2007, 16:02

Hexus/Vrzone is the same, but the fun factor still high :smile:
Someone need to check what Dave Orton doing in his office, packing his things, or drinking champain, this say more than 1000 words about R600 :lol:

if he is drinking in his office he would probably have to pack his things anyways :lol:

_xxx_

21-Mar-2007, 16:03

Bouncing Zabaglione Bros.

21-Mar-2007, 16:10

Of course people are following Fudo. Ever heard of boredom? He's the guy chirping in with new rumours on a daily basis (founded or not doesn't matter) and people take a look.

Hell, we're no better ourselves here in this forum, we just don't have a website with "-zilla" ending.

What's said here is clearly speculation. No one is dressing it up as fact and putting it on the front page - which is what Faud is doing. I wouldn't be surprised if most of his "scoops" come straight from here, with the rest coming from his fertile imagination.

Tim Murray

21-Mar-2007, 16:15

Hell, we're no better ourselves here in this forum, we just don't have a website with "-zilla" ending.
Some of us are! :runaway:

Kaotik

21-Mar-2007, 16:22

Some of us are! :runaway:

*Runs to register Beyond3DZilla.com*:shocked:

_xxx_

21-Mar-2007, 16:22

Some of us are! :runaway:

Many are, but still we have quite a few such people here.

jamis

21-Mar-2007, 16:31

That's it. I'm neg-repping the next person to link to FudZilla. Who's with me???!!! :twisted:

I've actually been thinking about this for a while, but I suspect I'll get spanked for misuse of the rep system. blah..blah..
You are taking this far too seriously guys. I mean this IS the "The NEXT LAST R600 Rumours & Speculation Thread". Now if you want to take rumours and speculation out of rumours and speculation threads, what do we have left? :lol:

Bouncing Zabaglione Bros.

21-Mar-2007, 16:45

You are taking this far too seriously guys. I mean this IS the "The NEXT LAST R600 Rumours & Speculation Thread". Now if you want to take rumours and speculation out of rumours and speculation threads, what do we have left? :lol:

I do expect rumours and speculation to be at least based on some kind of industry grounding or have a semblance of a logical train of thoughts.

Saying "R650 is being released early at 65 nm because it has better yields, performance, power requirements and is ready to go while R600 is late, hot, and difficult to make at 80 nm" is a speculation or a rumour.

Saying "R600 is made of green cheese and is powered by the souls of the damned" is not.

jamis

21-Mar-2007, 16:49

I do expect rumours and speculation to be at least based on some kind of industry grounding or have a semblance of a logical train of thoughts.

Saying "R650 is being released early at 65 nm because it has better yields, performance, power requirements and is ready to go while R600 is late, hot, and difficult to make at 80 nm" is a speculation or a rumour.

Saying "R600 is made of green cheese and is powered by the souls of the damned" is not.
Well if you say so. I'll take mine with ketchup.

vertex_shader

21-Mar-2007, 17:03

I think AIB's right when all they are dissapointed for what AMD doing with the R600, delays, and now not give out anything (except some watermarked card, without performance driver, for short time), from AMD aspect i understand the paranoia, leaks always coming from AIB's, and not all AIB partner ATi only, so the chance one of they partners leak something to NV is very big, and from AMD aspect the time is too early, so this can't happening now.

From user aspect the silence is not good, but time will tell everything, when someone bored has the chance to buy a DX10 VGA :smile:

pelly

21-Mar-2007, 17:43

IMHO, the whole R600 topic and Fudo's "speculation" illustrates exactly what makes Beyond3D's forums unique. Whereas other forums and websites might take his speculation as gospel, the B3D forums use their collective knowledge of graphics to disspel the rumors and depcit why something is/isn't true. At the end of the day, we're all here because we love graphics and the associated technology. If anything, Fudo gives us all the catalyst we need for some lively discussion. ;)

vertex_shader

21-Mar-2007, 18:17

trinibwoy

21-Mar-2007, 18:19

Saying "R600 is made of green cheese and is powered by the souls of the damned" is not.

Heh, I'm glad somebody finally recognized my contribution to R600 :twisted:

overclocked_enthusiasm

21-Mar-2007, 18:45

Well, they have some stuff going for them tho. They appear to have 80nm vs 90nm.

I would argue that a smaller process does not mean better margins. If you use Nvidia vs ATI with R520/R580 vs G70 for instance you could easily say that Nvidia realized much higher margins. Appareantly, by using a more mature fab process and not pushing clocks to the bleeding edge Nvidia was able to bin more chips and ATI was able to bin fewer. The other obvious advantage of a more mature fab process is time to market if R420, R520 and R600 are any indication.

I agree that 80 nm vs 90 nm SHOULD be an advantage but in the case of ATI they have yet to demonstrate that 130 nm low k, 90 nm or 80 nm were good fab decisions as none translated into good margins or time to market for the products.

caffeinated

21-Mar-2007, 19:00

Well, at this point, I think that rosy optimism that the drivers will be wonderful and that the delay is because they wanted R600 to be extra special at launch is a bit misguided. They likely have AIB partners down their throats wanting it yesterday, so they aren't delaying for perfection.

I agree with the idea that the lack of anything official is hurting them at this point; the reason why we don't have anything official is because there probably isn't anything official to confirm, yet.

Realistically speaking, I am of the opinion that if there was anything good to say, AMD would be shouting it from the rooftops at this point, and if they knew when it will be released in all of it's glorious splendor, we would know, too - an AIB partner would spill it, if for no other reason than to generate hype for their upcoming R600-based products.

vertex_shader

21-Mar-2007, 19:04

New clearer picture from the OEM version:
http://img442.imageshack.us/img442/8057/r600oemtr7.jpg

http://img239.imageshack.us/img239/1214/r60080nmpo8.jpg
:smile:

Geeforcer

21-Mar-2007, 19:09

I think it's pretty clear the outlined card is not R600, but the 80nm should be noted by "I want to believe" corwd.

BTW, to slightly perhrase myslef:

"Right before a new DirectX revision, a company released an surpiursing, out of nowhere GPU that almost no one saw coming. They stuck with a tried-and-true process, but still managed to put out a huge chip running at very competetive clockspeeds. Their competitor, on the other hand, decided that they needed a new process for their chip, that had around 20 million more transistors. Yet, their design was months late, (ran hot and consumed a lot of power) while running at previously unheard of speeds."

overclocked_enthusiasm

21-Mar-2007, 19:12

Ailuros

21-Mar-2007, 19:16

I think it's within human nature when we don't like the message, to also dislike the messenger and sometimes exactly the opposite if the message is to our liking. Heck my ancestors used to decaptivate the messenger if the message was bad (whereby the script of the "300" is a tad modest there with a simple kick into the pit), so I guess we should be glad we're a tad more civilized these days (or not LOL)...

As for the specific fore mentioned gentleman....errr I can't "dislike" or "hate" someone (lemme be a tad gentle here...) with a limited horizon. The best way to treat stupidity after all should be simply to ignore it.

ChrisK

21-Mar-2007, 19:16

Hmm, I admit to not reading all of the messages. Is it consensus now that R600 being 65nm is bogus? This message from someone I remember as knowledgeable and quite objective (weren't his forums once hosted by Beyond3d?) I found interesting:

Job Seesink is representing my site at Cebit, and he talked directly to folks at AMD/ATI who are in the know, and they confirmed that (R600) was going to be 65 nm.

Link here (http://techreport.com/ja.zz?id=173043). So was AMD flat out lying?

caffeinated

21-Mar-2007, 19:32

Hmm, I admit to not reading all of the messages. Is it consensus now that R600 being 65nm is bogus? This message from someone I remember as knowledgeable and quite objective (weren't his forums once hosted by Beyond3d?) I found interesting:

Link here (http://techreport.com/ja.zz?id=173043). So was AMD flat out lying?

I believe that at this point the consensus is that AMD was misunderstood and misquoted. Or deliberately misunderstood and misquoted.

Rys

21-Mar-2007, 19:37

R600 is 80nm. That's pretty much all there is to it. As far as the "huh, could it be 65nm instead?" goes, that's just people not taking what Bergman said at the press conference in the right way. I was in the room, I didn't sit there thinking, "hang on a second..........". If there was chatter after that with Bergman et al, that clearly stated R600 was 65nm, I think those people that heard it should talk to their AMD PR contacts for some clarification.

Further, all the R600 samples at CeBIT (I saw boards with my own eyes) were using 80nm silicon.

Rys

21-Mar-2007, 19:43

Arty

21-Mar-2007, 19:47

Hell, we're no better ourselves here in this forum, we just don't have a website with "-zilla" ending.
ughhh :nope: speak for yourself :roll:

Jawed

21-Mar-2007, 19:59

Rys

21-Mar-2007, 20:03

R580 is in two versions: one for GDDR3 and lower and another which is solely dedicated to GDDR4. That's correct, isn't it?

If so, then wouldn't R600 have to have two versions, one for GDDR3 and one for GDDR4?

Or, is it just the circuit board that has to be different?
The spin was just to optimise running with GDDR4, not to add it in or replace a GDDR3-only MC. You can use 'R580+' with GDDR3 just fine.

The differences are board-level, not chip-level as you guess :)

Kaotik

21-Mar-2007, 20:04

For all I know AMD representatives talking about R6xx/RV6xx chips, is that they've been talking a lot about how great 65nm is, but they've been only talking about RV6xx's and 65nm, not a word about R600 being 65nm, so it's easily possible that they've been misunderstood

Geo

21-Mar-2007, 20:04

R580 is in two versions: one for GDDR3 and lower and another which is solely dedicated to GDDR4. That's correct, isn't it?

If so, then wouldn't R600 have to have two versions, one for GDDR3 and one for GDDR4?

Or, is it just the circuit board that has to be different?

Jawed

Hmm. I don't think so. R580+ was certainly tweaked to improve the GDDR4 peformance. But they advertised the original R580 rev as supporting GDDR4 as well, and I've never heard anyone say the R580+ version won't support GDDR3. So as far as I know that was just a case of "with a little more experience we decided to tweak it a bit".

The PCB apparently does have to be quite a bit different.

rosman

21-Mar-2007, 20:05

I dont know if it has been posted yet but here is the ruby demo :

http://www.hexus.tv/show.php?show=66

Looks really nice better than the adrianne thing. But still no R600 available or any true info about the performance. :( Sad

Tim Murray

21-Mar-2007, 20:09

rosman

21-Mar-2007, 20:44

Apparently, you are not reading this thread at all. :razz:

Its just better quality than the youtube thing, i ve seen posted here :p. But i may have forgotten another link. But i really want to see the benches, i need to feed my 24inches !!!

icecold1983

21-Mar-2007, 20:50

Doubtful, the OC cards already hit those clocks. If the Ultra is going to be anything like the 7800GTX-512 (hopefully with better availability), core clocks would have to be a lot closer to 700. Or else they could just let XFX and BFG take on R600 :)

In the case of G70 it was a big step up to 90nm. The transition is a little easier now so I wouldn't be surprised in the least if Nvidia was already working on an 80nm shrink of G80 before the 90nm version even launched. They need to shrink that chip if they have any intentions of doing a GX2 anyway.

i dont expect another 7800 gtx 512, because availability would be terrible.

Natoma

21-Mar-2007, 20:51

Secondly, the anti-Fudo chatter is just plain boring now. If you don't like his work, go over to his forums and tell him that since it's just a waste of database space here to repeat it over and over again.

This is a thread for R600 speculation, not web politics and drama.

There wouldn't be anti-Fudo chatter if there weren't any Fudo links. Yell at the posters, not the bitchers. ;-)

Natoma

21-Mar-2007, 22:26

I've actually been thinking about this for a while, but I suspect I'll get spanked for misuse of the rep system. I mean TheInq is a well known gossip site and all, but most of the criticism it got from this board was down to Faud's inane and incorrect rambling. Now Faud's gone, people have just followed him and keep linking to the same old rubbish.

It's not like this is insider info, this is just total rumours that he keep splurging out. They are just unfounded guesses pretending to be based on some kind of secret knowledge. Heck, Faud's new site makes TheInq look like a paragon of objective and accurate reporting.

Take a look back at Faud's old R600 news. He swings from one extreme to another, completely contradicting his previous stories. If he knew anything, he wouldn't have to completely backtrack over what he said last week.

Actually I was mostly joking. I've been guilty on occasion of referencing L'Inq as well.

http://forum.beyond3d.com/showthread.php?p=937364#post937364 :wink:

But yea, it does get old. The guy knows nothing. Get used to it. :lol2:

_xxx_

21-Mar-2007, 23:21

ughhh :nope: speak for yourself :roll:

Err, you have been reading this forum, right? :???:

I'm lazy to look for quotes, but if I got a penny for every "insider info" crap post here, I'd be a very rich man by now.

Arty

22-Mar-2007, 00:46

Err, you have been reading this forum, right? :???:

I'm lazy to look for quotes, but if I got a penny for every "insider info" crap post here, I'd be a very rich man by now.
Yes, I have been reading this forum. I objected to you using the world "we" in "we're no better ourselves here in this forum".

Jawed

22-Mar-2007, 00:55

http://www.xbitlabs.com/articles/editorial/display/cebit2007-3_3.html

What's particularly interesting about the second pic is that it appears to reveal the layout of the memory chips around the core - look carefully for the "raised" copper.

Jawed

Geo

22-Mar-2007, 01:00

INKster

22-Mar-2007, 01:05

Nice. Validates the catty-corner core shots too, it would seem.

Humm, memory chips on both sides of the PCB, perhaps ?
The three rows (like the G80) seem a little too "tight" for 16 pieces.

Geo

22-Mar-2007, 01:08

Humm, memory chips on both sides of the PCB, perhaps ?
The three rows (like the G80) seem a little too "tight" for 16 pieces.

That would seem likely. It's what they do with FireGL after all.

Jawed

22-Mar-2007, 01:09

I'm gonna guess the memory is organised as 3-3-2, for what it's worth (not very much, really).

Jawed

Jawed

22-Mar-2007, 01:10

Humm, memory chips on both sides of the PCB, perhaps ?
Yeah. GDDR4 shouldn't need active cooling I think, and GDDR4 is presumably the only intended 1GB configuration.

Jawed

Jawed

22-Mar-2007, 01:17

I'm gonna guess the memory is organised as 3-3-2, for what it's worth (not very much, really).
Ugh, hang on, that doesn't make a 512-bit bus :oops: :roll:

ARGH brain fade.

Jawed

Jawed

22-Mar-2007, 01:58

Ah, I've realised that the GDDR3 version prolly has 16x 32MB (256Mbit) chips, which I guess means the memory is on both sides of the board for both GDDR3 and GDDR4 variants.

But the pix of the "OEM", long, board didn't seem to have memory on the back. The black metal has cut-outs for a few small chips, otherwise it's solid (and not very thick). It doesn't seem likely that there's memory under that metal.

ARGH - stuff isn't adding up. Bed.

Jawed

Russell

22-Mar-2007, 02:00

INKster

22-Mar-2007, 02:00

Yeah. GDDR4 shouldn't need active cooling I think, and GDDR4 is presumably the only intended 1GB configuration.

Jawed

This would also help explain the metal plates on the back (the shape is too elaborate to be just a simple anti-flex safeguard for the PCB).

http://resources.vr-zone.com//newspics/Mar07/19/R600-vrzone3.jpg

http://resources.vr-zone.com//newspics/Mar07/19/R600-vrzone6.jpg

The OEM version also has similar metal plates.

Megadrive1988

22-Mar-2007, 02:04

the Ruby demo would've been more impressive if the framerate wasn't so low. it seems to be range between 20 and 30 fps.

Kaotik

22-Mar-2007, 02:09

the Ruby demo would've been more impressive if the framerate wasn't so low. it seems to be range between 20 and 30 fps.

Were you there seeing it 1st hand, or are you just basing that on flash video?

Graham

22-Mar-2007, 02:26

the Ruby demo would've been more impressive if the framerate wasn't so low. it seems to be range between 20 and 30 fps.

The real question is how well Ruby will run on an 8800. :razz:

caffeinated

22-Mar-2007, 03:35

This would also help explain the metal plates on the back (the shape is too elaborate to be just a simple anti-flex safeguard for the PCB).

http://resources.vr-zone.com//newspics/Mar07/19/R600-vrzone3.jpg

http://resources.vr-zone.com//newspics/Mar07/19/R600-vrzone6.jpg

The OEM version also has similar metal plates.

Are you sure? It really doesn't look as though memory would fit under those...they appear fairly flush with the PCB.

Bob

22-Mar-2007, 05:01

8 64MB chips? Is such a thing widely available to make it doable?
It would have to be 8 64-bit chips to make a 512 bit interface. Those aren't really mass manufactured. GPUs typically use 32-bit DRAMs, which would imply that you need 16 chips to make 512 bits.

This is all independent from the actual capacity of those DRAMs, which can be (nearly) arbitrary.

INKster

22-Mar-2007, 05:11

Are you sure? It really doesn't look as though memory would fit under those...they appear fairly flush with the PCB.

I don't think it is that flushed tight.

Look carefully a that middle foreground screw on the black metal plate (first link i gave above).
Immediately to the right of it there is a small depression downwards in the metal, but the edge closest to the auxiliary power connectors stays (or so it appears to me) on the same plane as the place where the supposed back memory chips lie underneath.

Granted, contrast in that photo is not very telling, so i may have been misled by it, but it's a startling point to discuss nevertheless.

Russell

22-Mar-2007, 05:30

It would have to be 8 64-bit chips to make a 512 bit interface. Those aren't really mass manufactured. GPUs typically use 32-bit DRAMs, which would imply that you need 16 chips to make 512 bits.

I thought as much but was seeking confirmation of this. Thanks.

So those cards we're seeing have got to be 512MB models then, because there's no way there's room behind that metal on the back for memory. Unless the metal is indented on the board side to give more room behind it than it appears, but I don't see that as being very likely at all.

Shtal

22-Mar-2007, 07:07

Next, we got hold of some preliminary benchmarks figures of the R600 XTX card with core clock at 800MHz vs a GeForce 8800 GTX card. Using a Core 2 Extreme 2.93GHz processor on an Intel 975X board, the 3DMark06 score at 1600x1200 resolution is 97xx on the R600XTX compared to 95xx on the 8800GTX. Seems like R600XTX is running slightly faster than 8800GTX on the early release of drivers for R600. AMD is still working hard on the drivers and there are some more performance left to unlock. http://www.vr-zone.com/?i=4830

How much in percentage % "2000 points - 3DMark06 points" would that be....
Can some one help me out please :)

[Edit: I wonder if you over-clock (G80) GF8800GTX to "650MHz GPU" how much it will close the gap with R600-XTX] "GF8800Ultra might do it"

"800MHz GPU of R600" reminding me of 4 pipeline-NV30; it had to be clock @ 500MHz in order to compete with 8 pipeline-R300 @ 325MHz GPU.

IbaneZ

22-Mar-2007, 08:10

Secondly, the anti-Fudo chatter is just plain boring now. If you don't like his work, go over to his forums and tell him that since it's just a waste of database space here to repeat it over and over again.

This is a thread for R600 speculation, not web politics and drama.

Some day now Fudo will brake the news that R600 scores 18000 in 3dmark-06, and everybody here will love him. Thats how it works right? ;)

dizietsma

22-Mar-2007, 08:58

Only 200 points up on G80 in 3dmark06 ? Would that be down to drivers at this stage or is the resolution just not stressing it enough ( assuming it is true of coursde ).

Not mind blowing to say the least.

SugarCoat

22-Mar-2007, 09:10

How much in percentage % "2000 points - 3DMark06 points" would that be....
Can some one help me out please :)

[Edit: I wonder if you over-clock (G80) GF8800GTX to "650MHz GPU" how much it will close the gap with R600-XTX] "GF8800Ultra might do it"

"800MHz GPU of R600" reminding me of 4 pipeline-NV30; it had to be clock @ 500MHz in order to compete with 8 pipeline-R300 @ 325MHz GPU.

~2%, but i dont see what it matters. And to be honest i hope nVidia isnt done with their drivers either. Their last official release for XP and the 8800 was back in the first week of January which is a mere 2 months after the card launched. BOTH cards have driver tweaking to be done, the difference is ATiMD doesnt have a product out.

Shtal

22-Mar-2007, 09:13

I actually have few issues with AMD about that.
why is R600 so secretive?
they don't really say or explain anything about the product. wtf?
nobody is sure whether it is 65nm or 80nm, nobody knows the release date, nobody knows nothing.
everything is just a big pile of rumour! http://www.rage3d.com/board/showthread.php?threadid=33886175

Nvidia NV30 was exact same way; All Nvidia did is run the program call "Are you Ready" which gave squat about the product; and at the end "NV30" it turn out to be trash.

The only thing I could say about R600-XTX; that it could be apposite of NV30, it may turn out to be a miracle.

SugarCoat

22-Mar-2007, 09:21

Nvidia NV30 was exact same way; All Nvidia did is run the program call "Are you Ready" which gave squat about the product; and at the end "NV30" it turn out to be trash.

The only thing I could say about R600-XTX; that it could be apposite of NV30, it may turn out to be a miracle.

miracle or not they're already screwed just because the 8800s have had that crucial time to get into everyones brains. As far as gamers and the internet is concerned the 8800 is the best thing on the block and its going to take weeks if not months for R6xx based cards to begin break into that. If they arent anything special then they'll even lose the small percent of sales from people who know what they're talking about and know what they're buying. I really have enjoyed ATi's products so it does personally bug me to say that, and im pretty irked at how poorly their planning is.

As i already outlined in a pretty lengthy post in the last thread, the important thing is brand recognition, especially of the flagship cards, if you have the performance people associate the brand with it all the way down the line and without researching, so if the 8800 is praised, the majority of people will buy the 8600s and the 8300s over the X2600s and X2300s and so on. Plus it pleases OEMs and AIBs alike because their sales get a boost. Even go to Dell, right now, and look at many of their gaming computer configurations, they dont offer nearly as many ATi options as nVidia ones (mainly from the geforce 7 series but i think its a 4:1 or 5:1 ratio in favor of nVidia). The R600 may be great, but they're already screwed in my humble opinion and its going to get harder and harder on their sales the longer their product launches are spaced apart like this.

Shtal

22-Mar-2007, 09:28

miracle or not they're already screwed just because the 8800s have had that crucial time to get into everyones brains. As far as gamers and the internet is concerned the 8800 is the best thing on the block and its going to take weeks if not months for R6xx based cards to begin break into that. If they arent anything special then they'll even lose the small percent of sales from people who know what they're talking about and know what they're buying. I really have enjoyed ATi's products so it does personally bug me to say that, and im pretty irked at how poorly their planning is.

As i already outlined in a pretty lengthy post in the last thread, the important thing is brand recognition, especially of the flagship cards, if you have the performance people associate the brand with it all the way down the line and without researching, so if the 8800 is praised, the majority of people will buy the 8600s and the 8300s over the X2600s and X2300s and so on. Plus it pleases OEMs and AIBs alike because their sales get a boost. Even go to Dell, right now, and look at many of their gaming computer configurations, they dont offer nearly as many ATi options as nVidia ones (mainly from the geforce 7 series but i think its a 4:1 or 5:1 ratio in favor of nVidia). The R600 may be great, but they're already screwed in my humble opinion and its going to get harder and harder on their sales the longer their product launches are spaced apart like this.

Agreed!....

22-Mar-2007, 11:23

vertex_shader

22-Mar-2007, 11:46

Some more leaked pics. Now if I could only see the outline of this card and not it's colors, I would've thought it was a GF8800GTX. :lol:

http://www.tweaktown.com/articles/1069/amd_r600_retail_radeon_x2900xtx_better_photos_imme rge/index.html

Looks good with this flames, i hope its fast too not only looks nice :smile:
First picture bad watermarked, tweakown.com :lol:

Razor1

22-Mar-2007, 11:53

as long as those flames stay fake :grin:

vertex_shader

22-Mar-2007, 12:01

as long as those flames stay fake :grin:

Maybe flame is just the watermark, or this is the need for speed carbon version :smile:

trinibwoy

22-Mar-2007, 12:04

Hmmm they're claiming that it has 1GB RAM but I can't see how they could fit memory below that back-plate. It seems pretty tight.

Jawed

22-Mar-2007, 14:09

I have to admit I find myself drawn to the conclusion that AMD has fitted all 16 memory chips on to just one side of the board, for both 512MB-GDDR3 and 1GB-GDDR4 configurations.

The pictures of RV610 showed memory chips placed really close to each other - I'm thinking that R600 needs similarly tight packing of the memory. Something like 6-6-4. Can't say I feel like trying to "measure" the spaces to see if that's possible. At least, not right now.

Anyway, I seem to remember several comments along the lines of "squeezing all the components on to the board", which might have been a reaction to seeing 16 closely spaced memory chips.

Jawed

Jawed

22-Mar-2007, 14:09

Some more leaked pics. Now if I could only see the outline of this card and not it's colors, I would've thought it was a GF8800GTX. :lol:

http://www.tweaktown.com/articles/1069/amd_r600_retail_radeon_x2900xtx_better_photos_imme rge/index.html
So, the question has to be asked, did our suggestions that Level505 couldn't show board pix - because they were liberally covered in "identifying marks" get turned into an actual procedure for sample boards, or was AMD doing that already :?:

Oh, and are those flames ironic?

Jawed

Rys

22-Mar-2007, 14:22

So, the question has to be asked, did our suggestions that Level505 couldn't show board pix - because they were liberally covered in "identifying marks" get turned into an actual procedure for sample boards, or was AMD doing that already :?:
All R600 the samples at CeBIT were liberally scribbled on by AMD to identify the AIB partner, should pictures get out. I'm not 100% convinced that board is unidentifiable, but they've done a good job obscuring things.

Oh, and are those flames ironic?
:lol2:

neliz

22-Mar-2007, 14:47

Geo

22-Mar-2007, 14:53

Some day now Fudo will brake the news that R600 scores 18000 in 3dmark-06, and everybody here will love him. Thats how it works right? ;)

No, that's not how it works, and that was pretty offensive for you to say so, protective winky or not.

Rys

22-Mar-2007, 14:55

One of the people over at vr-zone noticed that the back of the card is marked with TUL :)

now we know which partner was unhappy about not being able to show R600! :)

http://resources.vr-zone.com/newzhunter/r600inthewild_01l-tul.jpg
It's there in the other pics too.

http://www.beyond3d.com/images/rys/is-it-tul.jpg

:!: (http://www.beyond3d.com/images/rys/i-think-so!.jpg)

Sound_Card

22-Mar-2007, 15:06

So the 8800 Ultra is comming out in mid April. I hope that ATi brings out R600 very shortly of that release.:sad:

Jawed

22-Mar-2007, 15:08

Well, it seems like the markings served their purpose.

A year ago -ish there was talk about ATI-exclusive AIB partners becoming NVidia partners too. It didn't seem to come to anything. I have to admit I sorta concluded that the grass didn't look any greener on the other side for these AIBs so they stayed put.

Hmm, I'm now wondering if there are any truly exclusive partners. Someone needs to draw a family tree...

It seems MSI decided to reveal NVidia products at CeBIT, with L'Inq talking about partners being unhappy at NVidia's delays.

Jawed

Razor1

22-Mar-2007, 15:19

That was Sapphire but Sapphire's parent company already sells nV cards, which PC partner picked up about a year ago.

Mark

22-Mar-2007, 15:25

Flames? That's like putting a picture of Michael Jackson on the side of a Day Care Center.

Geo

22-Mar-2007, 15:27

So the 8800 Ultra is comming out in mid April.

Tell us over in the Beyond G80 thread if you have any source for that other than NGOHQ that isn't citing NGOHQ, and on what basis NGOHQ is a credible source for that kind of exclusive.

Rys

22-Mar-2007, 15:27

That was Sapphire but Sapphire's parent company already sells nV cards, which PC partner picked up about a year ago.
There was no picking up. PC Partner, Sapphire's parent, created Zotac with the specific intent to sell NV boards to retail. And bear in mind that they were already manufacturing them for other companies anyway, so it wasn't a massively new endeavour or something they bought.

Razor1

22-Mar-2007, 15:29

There was no picking up. PC Partner, Sapphire's parent, created Zotac with the specific intent to sell NV boards to retail. And bear in mind that they were already manufacturing them for other companies anyway, so it wasn't a massively new endeavour or something they bought.

ah ok, thx for the info!

Sobek

22-Mar-2007, 17:11

It's there in the other pics too.

*img-snip*

:!: (http://www.beyond3d.com/images/rys/i-think-so!.jpg)

Forgive me, but where exactly in that image is the TUL?

*edit* Never mind, I see now that TUL = Powercolour holder. Still, where abouts is the TUL logo on that second image?

neliz

22-Mar-2007, 17:16

Forgive me, but where exactly in that image is the TUL?

*edit* Never mind, I see now that TUL = Powercolour holder. Still, where abouts is the TUL logo on that second image?

Near the fan.

If you take the ati logo on the fan, go left and you'll see that TUL is written on top of a grey flame.

You can see appendices sticking out of the flame, the T and the L are sticking out.
The rest of the TUL's seem to be blurred out, but you can clearly see some lighter silver written over the flames, where it said TUL

Rys

22-Mar-2007, 17:21

Forgive me, but where exactly in that image is the TUL?

*edit* Never mind, I see now that TUL = Powercolour holder. Still, where abouts is the TUL logo on that second image?
There's actually more than one blurred out written occurrence of TUL in that image. This is the most noticeable one (http://www.beyond3d.com/images/rys/i-think-so!.jpg) IMHO, although I spot at least two more.

neliz

22-Mar-2007, 17:26

some more:

http://home.wanadoo.nl/deachterberg/neliztul.JPG

After al, as said before, level505 was partially right, ATI did cover all cards with markings everywhere, but nothing that a good photoshop healing brush can't fix.

vertex_shader

22-Mar-2007, 17:30

There's actually more than one blurred out written occurrence of TUL in that image. This is the most noticeable one (http://www.beyond3d.com/images/rys/i-think-so!.jpg) IMHO, although I spot at least two more.

What they penalty will be? can't sell any r600 for a month after launch?:wink:

neliz

22-Mar-2007, 17:31

What they penalty will be? can't sell any r600 for a month after launch?:wink:

Their bioses will be clocked 3MHz slower.

Mark

22-Mar-2007, 17:45

Pretty safe to say it's TUL, then. So what are the consequences?

Creig

22-Mar-2007, 17:55

I can't see any other reason for the black backplate other than as a heatspreader. It's placed along the same areas where you would likely find memory chips and VRM. It's certainly not needed as a stiffener for the PCB. The HSF it's bolted to is 100X stiffer than that backplate ever could be. GDDR4 might not need a heatspreader, but what about fast GDDR3?

caffeinated

22-Mar-2007, 17:56

Pretty safe to say it's TUL, then. So what are the consequences?

Their product boxes will have to be horrible.

Oh wait, that's PowerColour, so they will be anyway, my bad. :grin:

Don't make me unearth the box shot for Power Colour that had the picture of the ahem, extremely homosexual looking male on it. Several posters commented on that particular box as being something they would hide in a brown paper bag on the way home, and that it would get disposed of immediately. It did look rather, um, flaming, but not like the R600 shots we have just seen.

Too late, I found it anyway

http://img62.imageshack.us/img62/77/r9700box1cx0.jpg (https://addons.mozilla.org/firefox/1174)

I particularly like the phallic-looking industrial towers in the background.

Just imagine something like that wrapped around an R600. I'd say that is punishment enough!

{Sniping}Waste

22-Mar-2007, 20:12

Ah, not the Evil Colon Commander Power Color Box. I still can't burn that image out of my mind even thou its been like 5 years.

Silent_Buddha

22-Mar-2007, 21:00

I'd imagine that any company caught leaking information would...

1. Get less information and get it later in the future. Meaning less time to plan for launch meaning a disadvantage compared to other companies.

2. Be lower on the chip supply chain, meaning if there is higher demand than available supply they will have to wait longer or get a smaller number of chips.

I'd imagine there's probably other things, like less favorable pricing structures, etc.

Of course, that's all predicated on there being enough partners willing to invest in large quantities at launch. If not, then AMD/ATI would have to bite the bullet and just say, "Please, don't do that again or we'll remember this when the product proves to be a hot seller."

Regards,
Croaker

Creig

22-Mar-2007, 21:16

I just can't fathom how leaking pictures of the R600 is actually hurting AMD in any way, shape or form. The seemingly endless string of delays has already done about as much damage as AMD is going to get at this point in time. Their only saving grace now will depend on its price/performance ratio.

Bouncing Zabaglione Bros.

22-Mar-2007, 21:34

AMD AIB partners must be desperate to start publicising and selling R6xx boards. Can you imagine what it's like for AMD-only AIB seeing all the sales made by G80, and the halo effect that is having on the rest of the Nvidia products?

I'm sure AIBs will be turning round and telling AMD that if AMD were executing or publicising R600, the AIBs wouldn't be in such a state of desperation that they end up leaking photos.

It was exactly in this situation (when R300 was early and impressive and NV30 MIA) that led to a lot of Nvidia-only AIBs starting to sell ATI boards. Customers got fed up of waiting for the late arrival, and so did AIBs who couldn't compete without a comparable product.

AMD missed the autumn back-to-school, they missed Christmas, missed CeBit, missed their own editor's day, they are going to miss Easter... how much more will their partners and customers stand before going to the competition?

Jawed

22-Mar-2007, 21:46

Serious question now: are there any AIB partner companies that are exclusively NVidia or AMD? That is, excluding "sister companies" within the same group or under the same parent company.

Jawed

Razor1

22-Mar-2007, 22:13

evga, bfg, pny, xfx for nV

visiontek and connect 3d for ATi

Could be wrong though just thinking off the top of my head.

Rys

22-Mar-2007, 22:14

PNY, Pine (XFX) and BFG are exclusively NVIDIA in that respect. There might be more on that side of the fence, too (NVIDIA have 30+ worldwide AIB partners, not all operate in the same regions, etc etc).

Foxconn don't sell ATI boards under their own brand, for example, nor do they own a company that does (nor does Foxconn's parent), so that's another NVIDIA-only partner.

As for AMD, you have TUL (Powercolor), Info-tek (GeCube), HIS and Yuan (off the top of my head).

So there are still plenty of exclusive-to-the-IHV deals and partnerships around.

ERK

22-Mar-2007, 22:22

Actually, I have to disagree with some here as it looks to me as if there is memory on the back under the black plate. I can even imagine I see the edges of the chips (brownish) on the "2" side (of 3-3-2 arrangement) in this pic. (http://www.tweaktown.com/popImg.php?img=r600inthewild_01l.jpg) It is near where they edited with the red circles. The plate even angles away where there is no third chip. The openings in the plate could be only for dip packages or tall mounted elements.

ERK

Topman

22-Mar-2007, 23:31

http://www.hardware.info/images/articles/r600-dragon-schema.jpg
B007 = Dragonshead2?

bye

Deusp

22-Mar-2007, 23:40

http://www.hardware.info/images/articles/r600-dragon-schema.jpg
B007 = Dragonshead2?

bye

I see 8 memory chips. 256-bit version?

INKster

23-Mar-2007, 00:03

Where's the 45º angled die ?

turtle

23-Mar-2007, 00:04

Yes, B007 is Dragonshead2 (2900XTX).

I guess that answers if they are all on the front (no). Jawed's 3-3-2/6-6-4 looks to be correct.

Arty

23-Mar-2007, 00:05

Where's the 45º angled die ?
AFAIK, the pin shot was not rotated.

Kaotik

23-Mar-2007, 00:14

AFAIK, the pin shot was not rotated.

That's correct, die is rotated, pins are not.

caffeinated

23-Mar-2007, 00:46

If you go there, the article is from 2/27/07.

Kind of old.

According to the article, only two PSUs have been validated for R600
_______________________________________
Quote:

Naturally, the Radeon X2900 range is suitable for Crossfire, and therefore the cards have two internal connector as we've seen them on the Radeon X1950 Pro, amongst others. The power consumption of a system with two Radeon X2900 XTX cards in a Crossfire setup will probably reach absurd levels, considering the videocards alone already consume 560 Watts. One or two additional processors and some harddisks added to that means that a total power consumption figure would exceed 800 Watts. Considering the fact that power supplies work most efficiently at half of their total power, we should cross our fingers and hope that the PSU manufacturers will turn to the production of 1500 Watt models very shortly.

ATI has incidentally started validating power supplies for systems with Radeon X2900 cards. Till now, just two power supplies have reached the bar with a single R600 card: a 1000 Watt Enermax Galaxy and a 1100 Watt PC Power & Cooling power supply.
__________________________________________

This information is supposedly from last month. If true, though, that certainly is interesting.

EDIT: Oops, thanks for the heads up, INKster. Sorry about that. That will teach me to be a triggerhappy poster. I'll just have to tame that overly quick clicking index finger ;)

http://www.hardware.info/en-US/articles/amdnY2pxZGGa/ATI_R600_the_rumours_so_far/3

INKster

23-Mar-2007, 00:49

http://www.hardware.info/en-US/articles/amdnY2pxZGGa/ATI_R600_the_rumours_so_far/3

If you go there, the article is from 2/27/07.

Kind of old.

Dead link.

Arnold Beckenbauer

23-Mar-2007, 01:18

I see 8 memory chips. 256-bit version?

512 MB version?

MulciberXP

23-Mar-2007, 01:22

Dead link.

i got there just fine

INKster

23-Mar-2007, 01:24

i got there just fine

Well, it doesn't work here.

If those power numbers are accurate, i may have to dust off my old Coolermaster adapter to plug two PSU's into the same system, just to power a single R600... :(

Kaotik

23-Mar-2007, 01:45

Now that thing is just pure and simple bullcrap.
Let's presume the 240W figure is correct for the R600 retail cards (which it can't be if it indeed works with 2x6pin, when the max is 225W) - the card would use 20A from 12V lines, pretty much any decent PSU can give you that especially when it can be split among several 12V rails, and you don't need to go the 1kW PSUs either.
Now, of course other components need power from the 12V lines too, but when most new PSUs have up to 3 12V rails with 18A peak each, with usually around 35-40A combined at once, there's nothing to worry as far as I can see

Jawed

23-Mar-2007, 01:51

Yes, B007 is Dragonshead2 (2900XTX).

I guess that answers if they are all on the front (no). Jawed's 3-3-2/6-6-4 looks to be correct.
LOL, actually I was thinking that 6-6-4 would get all the memory onto one side, tightly packed, I'd decided 3-3-2 was a no-go. LOL.

So, erm it seems like there is memory under that black metal plate on the back, so it's 3-3-2 on each side.

As for the "only two power supplies have been validated" rumour, I suspect what they mean is that at the time of the article only two power supplies had PEG2 power connectors, i.e. the 8-pin plugs, rather than 6-pin. The new Enermax Galaxy range is known to have PEG2 power connectors. PCP&C seems like a good candidate for the other, especially as it's an OEM for a lot of other companies as well as selling branded units.

Jawed

Russell

23-Mar-2007, 02:01

So, erm it seems like there is memory under that black metal plate on the back, so it's 3-3-2 on each side.

Which I admit makes a ton of sense, but I still can't see them fitting under that unless the back of the heatplate thingy were indented to give the chips more room. It's not thick enough for that though.

Jawed

23-Mar-2007, 02:13

Which I admit makes a ton of sense, but I still can't see them fitting under that unless the back of the heatplate thingy were indented to give the chips more room. It's not thick enough for that though.
Inspired by ERK I looked at some R600 board pix (blimey I've got quite a collection now) and in a few I see two things: 1. the metal is not very thick; 2. the metal seems to be raised clear of the circuit board.

So in my random, drINQly, to-ing and fro-ing, on this question I have to admit the metal doesn't seem like quite the barrier to the double-side memory config that it has in the past.

Anyway, that board schematic is worth way more than all this :lol:

Jawed

Russell

23-Mar-2007, 02:28

Inspired by ERK I looked at some R600 board pix (blimey I've got quite a collection now) and in a few I see two things: 1. the metal is not very thick; 2. the metal seems to be raised clear of the circuit board.

So in my random, drINQly, to-ing and fro-ing, on this question I have to admit the metal doesn't seem like quite the barrier to the double-side memory config that it has in the past.

Anyway, that board schematic is worth way more than all this :lol:

Jawed

Re, Board schematic: Very true. With all of this it's clear to me that it MUST be that way, which is how I felt all along but didn't want to accept it as fact. Regardless, it's a silly thing to debate lol. Look at how bored we are, waiting. Come on AMD, release your damned card or at least leak some more info!

caffeinated

23-Mar-2007, 03:43

INKster

23-Mar-2007, 03:46

Yes, but how did that site obtain the board schematic? How do we know what they are reporting is accurate? What does the designation of "B007" mean, as opposed to "A13" which was a known number?

That article was posted last month, as well, which I mentioned. Those schematics just now made the rounds with no one noticing them for over 3 weeks?

A13 -> GPU revision
B007 -> PCB revision

Makes sense.

Russell

23-Mar-2007, 03:47

Yes, but how did that site obtain the board schematic? How do we know what they are reporting is accurate? What does the designation of "B007" mean, as opposed to "A13" which was a known number?

That article was posted last month, as well, which I mentioned. Those schematics just now made the rounds with no one noticing them for over 3 weeks?

I have seen the R600 referred to as B007 elsewhere, though I forget where exactly.

But it is odd that it would sit unnoticed for almost a month. Good point there.

Unknown Soldier

23-Mar-2007, 03:58

And for the last time, just as the knowledgable ppl here(Rys, Geo etc.) have been saying. R600 is 80nm (http://www.fudzilla.com/index.php?option=com_content&task=view&id=221&Itemid=1)

US

R300King!

23-Mar-2007, 04:53

That's correct, die is rotated, pins are not.

Can you count the pins on it? :-D

Farhan

23-Mar-2007, 05:21

Can you count the pins on it? :-D

I think we counted something like 2140 pins back then from the picture.

_xxx_

23-Mar-2007, 08:28

I can't see any other reason for the black backplate other than as a heatspreader.

Black surface spreads heat about 1.5x better than a silver/white surface made of the same material.

vertex_shader

23-Mar-2007, 08:34

http://www.hardware.info/images/articles/r600-dragon-schema.jpg
B007 = Dragonshead2?

bye

When the die not 45º angle rotated than someone forget to tell thermaltake about this Link (http://www.xbitlabs.com/misc/picture/?src=/images/editorial/cebit2007-3/r600_cooler_bg.jpg&1=1) :smile:

_xxx_

23-Mar-2007, 08:36

When the die not 45º angle rotated than someone forget to tell thermaltake about this Link (http://www.xbitlabs.com/misc/picture/?src=/images/editorial/cebit2007-3/r600_cooler_bg.jpg&1=1) :smile:

The die is, but the pins aren't.

vertex_shader

23-Mar-2007, 08:52

The die is, but the pins aren't.

The picture dark, I not see the die 45º angle rotated, and i see no way to rotate the die without the pins.

Kaotik

23-Mar-2007, 08:58

The picture dark, I not see the die 45º angle rotated, and i see no way to rotate the die without the pins.

The picture doesn't have the chip on it, so no wonder you can't see it.
We've had several pictures of the core, it is rotated 45 degrees, and we've also seen pics of the backside of the chip, the pins are not. Even if you don't see a way to do it, AMD/ATI engineers did.

vertex_shader

23-Mar-2007, 09:14

The picture doesn't have the chip on it, so no wonder you can't see it.
We've had several pictures of the core, it is rotated 45 degrees, and we've also seen pics of the backside of the chip, the pins are not. Even if you don't see a way to do it, AMD/ATI engineers did.

Yes sorry :oops:, some more brightness helped, now i see good :smile:

neliz

23-Mar-2007, 09:54

I used photozoom Pro 2 to see if I could enhance the text.

The schematics are for the production board, it says "R600 GDDR4"
pn is "209"-609231-B0
Dat is Jan 15 2007, Jasmine "Lim"/Kim Figure 11 of "21"/23

board length is 24.1 cm

no-X

23-Mar-2007, 10:08

The die is, but the pins aren't.
Pins are, too. Each of them is rotated 45 degrees :wink:

Kaotik

23-Mar-2007, 10:46

Pins are, too. Each of them is rotated 45 degrees :wink:

You sure? They just might be rotated 135, 225 or 315 degrees too :shock:

neliz

23-Mar-2007, 11:04

You sure? They just might be rotated 135, 225 or 315 degrees too :shock:

Since they're round, I'm sure AMD came up with something like Variable Pin Rotating. or a Continuously Rotating Grid.

vertex_shader

23-Mar-2007, 11:33

Looks like Catalyst makers working hard with r600, 7.3 version still not released, TDU need some extra frame, and Atari can't supply any more yet, i hope AMD can :smile:

Hanners

23-Mar-2007, 12:17

As for AMD, you have TUL (Powercolor), Info-tek (GeCube), HIS and Yuan (off the top of my head).

So there are still plenty of exclusive-to-the-IHV deals and partnerships around.

TUL/Powercolor have Zogis as their NVIDIA arm. You're right on the others though as far as I know.

vertex_shader

23-Mar-2007, 12:40

TUL/Powercolor have Zogis as their NVIDIA arm. You're right on the others though as far as I know.

Zogis not part of sapphire?

23-Mar-2007, 12:50

I know some people get the shivers when they get linked here, but they've got something new (http://uk.theinquirer.net/?article=38442) online about the launch date.

Razor1

23-Mar-2007, 12:56

Zogis not part of sapphire?

Zotac is probably what you are talking about, thats a sister company of PC Partners just as Sapphire is as well.

too many ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ Z to remember :)

Rys

23-Mar-2007, 12:59

I know some people get the shivers when they get linked here, but they've got something new (http://uk.theinquirer.net/?article=38442) online about the launch date.
Theo should probably reevaluate what the word launch means.

Arty

23-Mar-2007, 13:14

Theo should probably reevaluate what the word launch means.
Techday? And he still insists on the 65nm part. :huh:

vertex_shader

23-Mar-2007, 13:15

Zotac is probably what you are talking about, thats a sister company of PC Partners just as Sapphire is as well.

too many ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ Z to remember :)

right :smile:

When i good remember Club3d part of TUL corporation too as Powercolor (Club3d was the only company sell ATi,NV,Volari,S3 cards.)
Info-Tek (Gecube) part of Gigabyte.
Looks like HIS the only real worldwide ATi VGA only company.

ATi AIB list (http://support.ati.com/ics/support/default.asp?deptID=894&task=knowledge&folderID=588)

neliz

23-Mar-2007, 13:57

http://uk.theinquirer.net/?article=38442
:wink:

Welcome to one hour ago! :wink:
I know some people get the shivers when they get linked here, but they've got something new (http://uk.theinquirer.net/?article=38442) online about the launch date.

Geo

23-Mar-2007, 13:58

Theo should probably reevaluate what the word launch means.

It seems to be a recent phenomenon, rather than an isolated instance with them. For instance: http://uk.theinquirer.net/?article=37407

fellix

23-Mar-2007, 14:03

Africa!
Well, I guess this will be some hell of a trial for the air cooling capabilities of the top board. :lol:

nicolasb

23-Mar-2007, 15:02

So, is it time for "the LAST LAST R600 Rumours & Speculation Thread" yet?

Bouncing Zabaglione Bros.

23-Mar-2007, 15:09

So, is it time for "the LAST LAST R600 Rumours & Speculation Thread" yet?

Definitely not! It's another month to launch, and then there might be another month to ship. That puts us almost into June even before we see any slippage. It would merely be tempting the graphic card gods to punish us if we have a LAST LAST thread.

And still they are still reporting "production chips" as being 65nm...?! If this is true I expect it more likely for further delay, not less.

Oh AMD! The humanity!

_xxx_

23-Mar-2007, 15:15

Definitely not! It another month to launch, and then there might be another month to ship

You surely mean "maybe" or "probably"? We thought the same for dec, then feb, then march and right now I believe in nothing I can't verify.

caffeinated

23-Mar-2007, 15:37

For those of you that use Firefox, a quick way to zoom in on an image is the imagezoom extension.

http://img89.imageshack.us/img89/128/imagezoominactiontx6.jpg (https://addons.mozilla.org/firefox/1174)

That way, you don't even have to open up a different application to better see the details on a screen shot. I know this is a bit OT, but it seemed appropriate to share.

_xxx_

23-Mar-2007, 16:16

"Advanced dork", wtf is that? :lol:

caffeinated

23-Mar-2007, 17:13

"Advanced dork", wtf is that? :lol:

It uses the Advanced and Boolean search strings on user-highlighted text through Google automatically. Silly name, but very useful for those that remember the older ways of web searching and wish to employ them in an updated fashion ;).

It can be a good way to ferret out those R600 tidbits (I included that to keep this from getting OT - it wasn't my intention to cause topic drift, so feel free to PM me with further questions, or perhaps if there is a desire I could create a thread elsewhere for my "Caffeinated's Interesting Browser Addons)

Deusp

23-Mar-2007, 23:34

Definitely not! It's another month to launch, and then there might be another month to ship. That puts us almost into June even before we see any slippage. It would merely be tempting the graphic card gods to punish us if we have a LAST LAST thread.

And still they are still reporting "production chips" as being 65nm...?! If this is true I expect it more likely for further delay, not less.

Oh AMD! The humanity!

If the 65nm claims are true, and they slipped there again, then don't call me shocked if we see a LAST LAST LAST thread. :p

Bouncing Zabaglione Bros.

23-Mar-2007, 23:43

If the 65nm claims are true, and they slipped there again, then don't call me shocked if we see a LAST LAST LAST thread. :p

Maybe we need a MAYBE OR MAYBE NOT LAST thread? Or a DEFINITELY NOT LAST thread?

For the love of god AMD, spill some beans, let us know something! <shakes fist at sky>

Power_man

24-Mar-2007, 00:05

Hm. . . I don't understand ,what is ThiS (http://img70.imageshack.us/my.php?image=r600dragonschemasi4.jpg) ? :roll:

turtle

24-Mar-2007, 01:56

Rage Theater chip, most likely, or perhaps just a voltage regulator.

edit: Theatre chip looks to be off the bottom-left...I'm not quite sure what it is.

Glad this isn't R3D though, someone would be suggesting EDRAM, disconnected ROPs or summat. :P

INKster

24-Mar-2007, 02:04

Rage Theater chip, most likely, or perhaps just a voltage regulator.

No, that would be the square to the right of one of the DVI ports.

turtle

24-Mar-2007, 02:10

Yep, you're right. tried to edit that before someone made a fool of me. Too late. :razz:

I had forgotten how the 200 was more of a square, as opposed to the earlier RT's that were more of a rectangle shape similar to that.

Anarchist4000

24-Mar-2007, 02:19

Would a separate UVD chip be that much different from what was on the Crossfire cards before? Their functions seem to be similar so that chip could be for both UVD and Crossfire.

turtle

24-Mar-2007, 02:43

AFAIK crossfire capabilities are built into the die...but the UVD (Xilleon chip) part...That makes sense.

We keep hearing about how they will be separate and there is a "independent video processor" (http://www.bit-tech.net/news/2007/03/16/r600_family_has_independent_video_processor/) (I assume meaning a separate chip alongside the RT200 and R600), without UVD functions effecting graphics performance. It would make sense if that was a Xilleon-like chip off-die for those purposes. Had the RV550 seen the light of day, it might be easier to state with certainty if that's the case with what that chip is or not.

Sounds like a good assumption though.

edit: Just out of curiosity, could the audio subsystem mentioned on the Xilleon 260 be also what we're hearing about pertaining to the R600/RV6xx series? Would the audio system used in a TV/Home theater work similarly to that used in PC applications via a sound card?

caffeinated

24-Mar-2007, 03:31

So has everyone decided that these schematics sat on that site for 3 1/2 weeks without being noticed, but that doesn't matter because they are "accurate looking"? I'm just playing :twisted: 's advocate a bit here. The BS level has been awfully high; it just seems prudent to question everything.

I don't know anything about that particular site; that doesn't mean that they didn't get hold of some new information, it just seems strange that the article was dated 2/27/07, and no one noticed these R600 schematics until someone posted the picture in the forum. No offense to the poster of the pic is intended - it's merely a sincere desire for more information (such as the source of these schematics).

Anarchist4000

24-Mar-2007, 04:09

Someone probably noticed them but just didn't know what to make of them. Granted they could be fake but I haven't noticed any glaring mistakes yet. I wondered about the validity as well when I first saw the article and date.

Graham

24-Mar-2007, 05:08

Ok this is just a random thought....
But...

If the consumer release is going to be 65nm, ie R650 or whatever, the chip will be significantly smaller.
Now also can't we assume the old leaked shots of the chip were definitely 80nm. This chip was rotated...

Would the new chip be small enough to not have to be rotated, and even if it still is rotated, would it be measurable on that third party water cooling kit? I wouldn't imagine 3rd parties are making water coolers for OEMs if only oems are getting the 80nm part..

Sorry if this has been mentioned already...

Shtal

24-Mar-2007, 08:36

For the love of god AMD, spill some beans, let us know something! <shakes fist at sky>

Are you saying that because you want to buy R600-XTX and making your decision right now; thinking either it is worth the wait for R600-XTX or just to go ahead and buy GF8800GTX anyway.

Does people care about R600 performance or afraid that it will out-perform G80. Or neither of the options, but just out of curiosity.

IbaneZ

24-Mar-2007, 09:13

Arty

24-Mar-2007, 12:17

For those of you that use Firefox, a quick way to zoom in on an image is the imagezoom extension.

http://img89.imageshack.us/img89/128/imagezoominactiontx6.jpg

That way, you don't even have to open up a different application to better see the details on a screen shot. I know this is a bit OT, but it seemed appropriate to share.
Built-in feature in Opera, works faster too. :razz:

Arty

24-Mar-2007, 12:18

VR-Zone (http://www.vr-zone.com/?i=4854) got the picture of the water-cooled XTX:

http://img364.imageshack.us/img364/7632/r600wc1vs8.jpg

We seen enough of retail X2900 XTX card photos over the past few days and this time round we will bring you the water-cooled version of the R600 card to enjoy over the weekend. This cooling solution is probably provided by Aavid Thermalloy and coincidentally Aavid is touting their Turbo Tube cold plates (http://www.aavidthermalloy.com/products/liquid/turboflow.pdf) with mass production in May, same time as R600 launch. ATi's AIB partners can choose to use it, design their own or use a 3rd party water-cooler from Thermaltake (http://www.vr-zone.com/?i=4821).

R300King!

24-Mar-2007, 12:33

Built-in feature in Opera, works faster too. :razz:

QFT! Opera here too m8. ;)

BRiT

24-Mar-2007, 20:30

Built-in feature in Opera, works faster too. :razz:

For those using Mouse Gestures in FireFox, a simple right-click diagnol motion will zoom in on an image [upper-left to lower-right, ie: \ ]. It works well and is quick.

Are you saying that because you want to buy R600-XTX and making your decision right now; thinking either it is worth the wait for R600-XTX or just to go ahead and buy GF8800GTX anyway.

Does people care about R600 performance or afraid that it will out-perform G80. Or neither of the options, but just out of curiosity.

The reason I'm curious to have at least some info is I want to build a new system. I can use my current card and upgrade it later. The troublesome spot is the PSU. I don't know if I need a new PSU, and/or how the 8pin PCI-Express plug plays into it.

Bouncing Zabaglione Bros.

24-Mar-2007, 21:17

BZB will buy an 8800 GTX the day hell freezes over. :lol:

That's not completely fair. I'd be happy to give Nvidia my money (again) if they offer products and a business model that is what I want to pay for. I'd explain that a bit more, but it would be off-topic for this thread.

I'm pretty sure he's just curious. ;)

Isn't everyone? I am also intending to make a buying decision for "the big upgrade" as I jump to PCIe.

Shtal

24-Mar-2007, 21:47

The reason I'm curious to have at least some info is I want to build a new system. I can use my current card and upgrade it later. The troublesome spot is the PSU. I don't know if I need a new PSU, and/or how the 8pin PCI-Express plug plays into it.

But is it also true that you could plug 6pin into 8pin connector for ATI R600.... "Technically you don't have to have 8pin connector"

caffeinated

24-Mar-2007, 22:19

Built-in feature in Opera, works faster too. :razz:

But then you have to ...

... :shudder: use Opera :twisted:

R300King!

24-Mar-2007, 23:20

But then you have to ...

... :shudder: use Opera :twisted:

Only shudders of joy! ;)

Opera's great. It has everything I could want. The new version 9 is great, check it our here (http://www.opera.com/products/desktop/). Sorry for the OT.

Russell

25-Mar-2007, 00:49

Only shudders of joy! ;)

Opera's great. It has everything I could want. The new version 9 is great.

Now, let's hope that R600 is great as well :wink:

CMAN

25-Mar-2007, 02:57

Are you saying that because you want to buy R600-XTX and making your decision right now; thinking either it is worth the wait for R600-XTX or just to go ahead and buy GF8800GTX anyway.

Does people care about R600 performance or afraid that it will out-perform G80. Or neither of the options, but just out of curiosity.

I'm more curious about the R600's equivalents performance to the 8800 GTS. I'm looking to spend $350 soon on a graphics card, and I want the best deal. Not only am I curious if the ATI card will be quicker and/or better image quality, but I'm hoping Nvidia's prices will fall too. It's getting so close to launch, I just want to know. If it would be 4 months, then I'd just go and purchase an Nvidia card.

Deusp

25-Mar-2007, 05:51

I'm more curious about the R600's equivalents performance to the 8800 GTS. I'm looking to spend $350 soon on a graphics card, and I want the best deal. Not only am I curious if the ATI card will be quicker and/or better image quality, but I'm hoping Nvidia's prices will fall too. It's getting so close to launch, I just want to know. If it would be 4 months, then I'd just go and purchase an Nvidia card.

I find hard to believe that ATI can produce a price competitive solution (unless they don't want to make much money). The 512-bit memory bus probably can't scale down that well, nor does the huge power consumption bode well for it either. Cutting the power consumption down to reasonably levels probably means a huge hit in performance. Same with going with a 256-bit memory bus.

Silent_Buddha

25-Mar-2007, 08:07

Wow, when will these huge power consumption rumors die. It's like G80 rumors all over again. As has been reported numerous times R600 will aparantly run just fine on 2x6pin connectors, at least on the retail boards.

It may end up being more than G80 but not that much more.

As far as I can see that 8 pin connector could be for...

1. Being able to run with only 1 connector. So you could do a 1x8 or 2x6 for the same amount of power. I'm sure people with modular power supplies would love to only have to use one connector.

2. Rumor has it you can only use the build in ATI overclocker with a 1x8 and 1x6. However, I find this a bit odd and doubtful.

3. Who knows maybe they've been blowing air up our behinds and it actually does use a gajillion watts of power. :roll:

Regards,
SB

icecold1983

25-Mar-2007, 08:27

id pay a high price if it was a big jump over g80. id be the first person in line to buy 2 of em.

trinibwoy

25-Mar-2007, 12:03

Heh, thought you hated PC games ice? You know R600 will just run them faster - won't make them look any better. That's a lot of money to spend on something you don't enjoy :)

Ike Turner

25-Mar-2007, 13:13

Heh, thought you hated PC games ice? You know R600 will just run them faster - won't make them look any better. That's a lot of money to spend on something you don't enjoy :)

Lol good one...

nicolasb

25-Mar-2007, 16:21

I find hard to believe that ATI can produce a price competitive solution (unless they don't want to make much money). The 512-bit memory bus probably can't scale down that well, nor does the huge power consumption bode well for it either. Cutting the power consumption down to reasonably levels probably means a huge hit in performance. Same with going with a 256-bit memory bus.There are strong rumours that the R600 mid-range parts will use a 128-bit bus, not 256. Given that they are also made on a 65nm process, they may well be significantly cheaper than the Nvidia equivalents (which are also 128-bit but made on an 80nm process).

As has been reported numerous times R600 will aparantly run just fine on 2x6pin connectors, at least on the retail boards.It has also been reported numerous times that it will not. One quite persistent rumour is that the card will just function with 2x6-pin connectors, but only at stock speeds; overclocking will be disabled in the driver unless you're using an 8-pin and a 6-pin.

The "240W" power consumption figure for the retail version also refuses to go away, and, even if that's a peak power draw, it's still more than you can pull through the slot and two 6-pin connectors.

We don't know, of course; it's still too early to be definite about anything. But that's my point. :)

Arun

25-Mar-2007, 16:44

There are strong rumours that the R600 mid-range parts will use a 128-bit bus, not 256. Given that they are also made on a 65nm process, they may well be significantly cheaper than the Nvidia equivalents (which are also 128-bit but made on an 80nm process).I've said it times and again, but the idea tends to be that 65nm is more expensive per-mm2 than 80nm. A 65nm wafer might cost you $6K, while a 80 or 90nm wafer will cost perhaps $4K to $5K. Whether it's cheaper per-transistor depends on yields.

Another aspect, of course, is that a smaller process will tend to allow for higher clocks and/or lower power dissipation, so your performance-per-transistor will also be higher. Anyway, it's definitely not that clear-cut, but you'd expect both companies to know what they're doing.

EDIT: Actually, the price gap between 65nm and 90nm apparently isn't that large. So it's still more expensive per mm2, but not dramatically so I guess. Yields will be lower too, but you'd assume that isn't too drastic right now either for low-end/mid-end chips.

Kaotik

25-Mar-2007, 18:21

There are strong rumours that the R600 mid-range parts will use a 128-bit bus, not 256. Given that they are also made on a 65nm process, they may well be significantly cheaper than the Nvidia equivalents (which are also 128-bit but made on an 80nm process).

It has also been reported numerous times that it will not. One quite persistent rumour is that the card will just function with 2x6-pin connectors, but only at stock speeds; overclocking will be disabled in the driver unless you're using an 8-pin and a 6-pin.

The "240W" power consumption figure for the retail version also refuses to go away, and, even if that's a peak power draw, it's still more than you can pull through the slot and two 6-pin connectors.

We don't know, of course; it's still too early to be definite about anything. But that's my point. :)

I'm fairly sure the whole 240/270W numbers are just plain and simple FUD - I mean, there's even OEM models out there with 2x6pin physical connectors

INKster

25-Mar-2007, 18:28

I'm fairly sure the whole 240/270W numbers are just plain and simple FUD - I mean, there's even OEM models out there with 2x6pin physical connectors

It's not certain yet that those models weren't just developer boards, with lower clocks and/or smaller amount of memory.

turtle

25-Mar-2007, 19:10

I've said it times and again, but the idea tends to be that 65nm is more expensive per-mm2 than 80nm. A 65nm wafer might cost you $6K, while a 80 or 90nm wafer will cost perhaps $4K to $5K. Whether it's cheaper per-transistor depends on yields.

Another aspect, of course, is that a smaller process will tend to allow for higher clocks and/or lower power dissipation, so your performance-per-transistor will also be higher. Anyway, it's definitely not that clear-cut, but you'd expect both companies to know what they're doing.

EDIT: Actually, the price gap between 65nm and 90nm apparently isn't that large. So it's still more expensive per mm2, but not dramatically so I guess. Yields will be lower too, but you'd assume that isn't too drastic right now either for low-end/mid-end chips.

I'm trying to understand your logic.

Bear with me while I pull numbers out my ass that may be completely unrealistic, but irregardless I'm sure you'll catch my drift.

TSMC uses 300mm (12inch) wafers, correct?

Concerning R600, Let's say you can fit 15 80nm chips on a wafer, and 20 on a 65nm wafer....which to me sounds realistic. Obviously not correct, but probably relatively in the ballpark.

If a 65nm wafer costs 10-15% more than an 80nm one (as I imagine it's closer to 10-20% rather than 50%), let's say 3k to 3.5k, and say each working R600 is worth ~$300. the yields would have to be ~20-30% worse on 65nm complete dice (counting those with shaders etc disabled for 'GT'-like SKUs) to not justify the switch, not taking into account the performance/power difference and what it's worth to AMD/ATI, as it would obviously positively effect sales in every part using it versus the competition during it's lifetime. This is even more-so important in the mid-range, where the yields are no doubt much much higher, and market larger, and therefore each complete chip more lucrative. This is not to mention that RV630 seems to be the same size as G84, and 65nm may be needed to stay competitive, even with the extra added cost.

With boatloads of respect, I keep hearing this argument, but don't understand the reasoning behind it. I cannot see how switching to 65nm could be a bad thing, other than the fact that they had to (or would have had to) go back and completely redesign the high-end chip and would have less time for manufacturing. All things considering though, and how bad the 80nm seemed to fair (huge power consumption, seemingly similar performing to G80) I see this as the better decision in both the long and short term. For mid-range, especially in ATi's case concerning their past with performance per transistor (and therefore die size) especially on their mid/low-end parts being less than nvidia, it seems to be a no-brainer.

Deusp

25-Mar-2007, 20:02

I'm trying to understand your logic.

Bear with me while I pull numbers out my ass that may be completely unrealistic, but irregardless I'm sure you'll catch my drift.

TSMC uses 300mm (12inch) wafers, correct?

Concerning R600, Let's say you can fit 15 80nm chips on a wafer, and 20 on a 65nm wafer....which to me sounds realistic. Obviously not correct, but probably relatively in the ballpark.

If a 65nm wafer costs 10-15% more than an 80nm one (as I imagine it's closer to 10-20% rather than 50%), let's say 3k to 3.5k, and say each working R600 is worth ~$300. the yields would have to be ~20-30% worse on 65nm complete dice (counting those with shaders etc disabled for 'GT'-like SKUs) to not justify the switch, not taking into account the performance/power difference and what it's worth to AMD/ATI, as it would obviously positively effect sales in every part using it versus the competition during it's lifetime. This is even more-so important in the mid-range, where the yields are no doubt much much higher, and market larger, and therefore each complete chip more lucrative. This is not to mention that RV630 seems to be the same size as G84, and 65nm may be needed to stay competitive, even with the extra added cost.

With boatloads of respect, I keep hearing this argument, but don't understand the reasoning behind it. I cannot see how switching to 65nm could be a bad thing, other than the fact that they had to (or would have had to) go back and completely redesign the high-end chip and would have less time for manufacturing. All things considering though, and how bad the 80nm seemed to fair (huge power consumption, seemingly similar performing to G80) I see this as the better decision in both the long and short term. For mid-range, especially in ATi's case concerning their past with performance per transistor (and therefore die size) especially on their mid/low-end parts being less than nvidia, it seems to be a no-brainer.

There's a couple things wrong with this analysis. A GPU is not worth $300, but more like $100-150. Also, there's no guarantee that yields won't be dismal. A poorly conceived dieshrink could have terrible yields.

trinibwoy

25-Mar-2007, 20:07

turtle, what makes you think 20-30% lower yields on a part that size is not reasonable? I guess you can make whatever assumptions you like in order to justify the feasibility of either option.

Arty

25-Mar-2007, 20:53

There's a couple things wrong with this analysis. A GPU is not worth $300, but more like $100-150. Also, there's no guarantee that yields won't be dismal. A poorly conceived dieshrink could have terrible yields.
Correct. AFAIK, the costliest GPU of the last generation (R580) was in the neighbourhood of $150.

Geo

25-Mar-2007, 22:07

Correct. AFAIK, the costliest GPU of the last generation (R580) was in the neighbourhood of $150.

Cost (i.e what they paid TSMC)? Or price (as in what they charged AIBs)?

But then R580 is looking kinda petite, innit? Particularly if you add NVIO costs back. . .

INKster

25-Mar-2007, 22:26

Cost (i.e what they paid TSMC)? Or price (as in what they charged AIBs)?

But then R580 is looking kinda petite, innit? Particularly if you add NVIO costs back. . .

Even more so when Crossfire with R520/R580 is still reliant on some external chips, so that's one more cost to write off.
I'm sure integrating Crossfire on-die in R600 will be much more cost-effective than adding a bunch of chips from outside sources.
RV570 was certainly an early proof of that.

RussSchultz

25-Mar-2007, 22:30

20-30% yield is attrocious.

You should be getting upwards of 80-90% yield, or at the very least 60-70%. (Unless the yield falloff to larger die is worse than I guestimate)

A 300mm wafer should get you about 500 parts of 150mm2, with somewhere around 300-400 of which are 'good'.

With wafers costing $6K-ish(using Arun's numbers), that puts die cost at about $15-ish. Packaging & test probably a few more dollars(depending on how expensive flip chip packaging is), providing a cost of goods at about $20-ish.

Arty

25-Mar-2007, 22:42

Cost (i.e what they paid TSMC)? Or price (as in what they charged AIBs)?
The latter.
But then R580 is looking kinda petite, innit? Particularly if you add NVIO costs back. . .
Yeah, looks like an SKU for the performance range.

silent_guy

26-Mar-2007, 04:40

Concerning R600, Let's say you can fit 15 80nm chips on a wafer, and 20 on a 65nm wafer....which to me sounds realistic. Obviously not correct, but probably relatively in the ballpark.
Well, not realistic really, about one order of magnitude too low. At 400mm2 in 80nm, you'd get ~150 dies, not 15. And assuming perfect scaling, that'd give you ~220 dies in 65nm. :wink:

It really all depends about the time of introducing a new chip: the price per wafer (ppw) goes down significantly over the lifetime of a process. The curve starts with a high relatively flat plateau (~early adopter stage), but then starts to come down hard and keeps on going down for quite a while. It was like that for all technologies up to 130nm. It has to be the same for 110nm and below. As more fabs learn the specifics of a particular process, competition drives prices down. That's one.

In addition, the defect density (D0) also goes down steadily over the life of the process, until it bottoms out at some point. So in the early life of a process, your yield won't be great, but give it a year or two and it can be spectacular: so good, in fact, that for small chips it becomes more costly to test them before packaging! That's two.

Those 2 factors have to be multiplied to determine the final outcome. If you're going to introduce a new chip when ppw and D0 is still high and your competitor is riding the downward slopes, you're in for trouble.

So Arun is right: releasing a chip in a smaller process too soon can result in dramatically higher prices than those for chips with similar functionality in larger processes. But once D0 and ppw start coming down at the same time, the price curves intersect quickly and the smaller process wins.

I have seen it happen in the telecom world that chips were deliberately delayed to avoid this effect...

Another point: the price per wafer is essentially determined by the fab. They are using complex models to come up with reasonable prices. Companies with their own fab can and will use different pricing models because owning a fab requires entirely different accounting optimizations (wrt depreciation, tax credits and what not.) So introducing a product too soon at a smaller process can be harmful for a fabless company, but may be good for those that have one (e.g. because the fab would otherwise be running at too low capacity.)

Now the big question: where on the D0 and ppw curves are we at this point for 80nm and 65nm processes and is a 65nm R600 cheaper than the 80nm version?

I don't know. And neither does anyone of us. And those who do know won't reveal those closely guarded trade secrets. So the discussion ends right there. :wink:

Deusp

26-Mar-2007, 04:50

Well, not realistic really, about one order of magnitude too low. At 400mm2 in 80nm, you'd get ~150 dies, not 15. And assuming perfect scaling, that'd give you ~220 dies in 65nm. :wink:

It really all depends about the time of introducing a new chip: the price per wafer (ppw) goes down significantly over the lifetime of a process. The curve starts with a high relatively flat plateau (~early adopter stage), but then starts to come down hard and keeps on going down for quite a while. It was like that for all technologies up to 130nm. It has to be the same for 110nm and below. As more fabs learn the specifics of a particular process, competition drives prices down. That's one.

In addition, the defect density (D0) also goes down steadily over the life of the process, until it bottoms out at some point. So in the early life of a process, your yield won't be great, but give it a year or two and it can be spectacular: so good, in fact, that for small chips it becomes more costly to test them before packaging! That's two.

Those 2 factors have to be multiplied to determine the final outcome. If you're going to introduce a new chip when ppw and D0 is still high and your competitor is riding the downward slopes, you're in for trouble.

So Arun is right: releasing a chip in a smaller process too soon can result in dramatically higher prices than those for chips with similar functionality in larger processes. But once D0 and ppw start coming down at the same time, the price curves intersect quickly and the smaller process wins.

I have seen it happen in the telecom world that chips were deliberately delayed to avoid this effect...

Another point: the price per wafer is essentially determined by the fab. They are using complex models to come up with reasonable prices. Companies with their own fab can and will use different pricing models because owning a fab requires entirely different accounting optimizations (wrt depreciation, tax credits and what not.) So introducing a product too soon at a smaller process can be harmful for a fabless company, but may be good for those that have one (e.g. because the fab would otherwise be running at too low capacity.)

Now the big question: where on the D0 and ppw curves are we at this point for 80nm and 65nm processes and is a 65nm R600 cheaper than the 80nm version?

I don't know. And neither does anyone of us. And those who do know won't reveal those closely guarded trade secrets. So the discussion ends right there. :wink:

There was a report (I think from eetimes) that stated yields for 65nm right now were pretty bad. If someone could find that report, I would be very grateful.

BRiT

26-Mar-2007, 04:55

There was a report (I think from eetimes) that stated yields for 65nm right now were pretty bad. If someone could find that report, I would be very grateful.

Report or Rumor? Big difference there.

Arty

26-Mar-2007, 05:12

Report or Rumor? Big difference there.
Report. You can check Arun's post history for the link. Its buried in there. ;)

(I'm lazy to look it up because I have to hit the sack now, I hate Mondays :mad: )

Russell

26-Mar-2007, 05:13

There was a report (I think from eetimes) that stated yields for 65nm right now were pretty bad. If someone could find that report, I would be very grateful.

I remember that article. That was the one that said how Intel was one of the few that was having good 65nm yields. Wasn't that a few months ago? Who knows what may have changed. It was also very non-specific. It gave neither numbers nor a ballpark estimate suggesting what "bad yields" were. Hardly anything to base assumptions on.

Their idea of bad yields may also not have factored in that you can disable portions of defective chips and sell them as cut down models, as AMD/ATI will surely do. Sure, you may only get 50% yields on your best chip, which sounds terribly wasteful and expensive, but perhaps you can still sell 70% of what you manufacture via disabled pipes et al. Suddenly 50% doesn't sound so bad anymore.

Or, on the other hand, they could have meant that manufacturers are typically getting 30% yields and AMD/ATI can only manage to salvage 42% of their total output for sales. There's nothing to indicate that it's one way or the other.

silent_guy

26-Mar-2007, 05:24

Their idea of bad yields may also not have factored in that you can disable portions of defective chips and sell them as cut down models, as AMD/ATI will surely do.

Yes, that's why using 'yield' as a metric is about as useful to describe the quality of a process as 'mips' is for the quality of a CPU. (Or worse, actually). 'Defect density' is much more precise. Yield is useful to compare specific chips, but that's clearly not what EET was talking about...

Probably too much of a nuance for certain types of journalists...

Russell

26-Mar-2007, 05:37

Yes, that's why using 'yield' as a metric is about as useful to describe the quality of a process as 'mips' is for the quality of a CPU. (Or worse, actually). 'Defect density' is much more precise. Yield is useful to compare specific chips, but that's clearly not what EET was talking about...

Probably too much of a nuance of certain types of journalists...

Nuanced is certain. My concern is that they COULD have been talking about overall chip yields, which does bode poorly for firms using a 65nm process, but they could also not have been. Their statement that yields are bad is about as good of a metric as me saying "65nm yields are probably not so great because it's fairly new" is. It gives absolutely no information as to what or why.

caffeinated

26-Mar-2007, 06:05

I have to admit, I'm getting a bit confused with all of this 65nm discussion. I thought R600 was going to be released at 80nm, and the mainstream/value priced parts at 65nm.

::sigh::

I guess I should just go to sleep. I'm sure when I wake up in the morning the new rumors will be different anyway, so no reason to sort it all out right now. :grin:

Russell

26-Mar-2007, 06:09

Tim Murray

26-Mar-2007, 06:51

I have to admit, I'm getting a bit confused with all of this 65nm discussion. I thought R600 was going to be released at 80nm, and the mainstream/value priced parts at 65nm.
Go back several pages. Read what Rys said. This is the truth. Indisputable truth. He was in the room when all the 65nm nonsense started.

caffeinated

26-Mar-2007, 06:54

Go back several pages. Read what Rys said. This is the truth. Indisputable truth. He was in the room when all the 65nm nonsense started.

I know...:grin: That's why I got confused. I didn't think anything had changed in the last day or so since I had checked, but :eek:

nicolasb

26-Mar-2007, 10:28

I have to admit, I'm getting a bit confused with all of this 65nm discussion. I thought R600 was going to be released at 80nm, and the mainstream/value priced parts at 65nm.We were discussing 65nm because we were talking about the "mainstream/value priced parts". Keep up! :roll:

Arty

26-Mar-2007, 14:52

Russell

26-Mar-2007, 17:47

Inq (http://theinquirer.net/default.aspx?article=38457)

So, in other words, AMD is apparently showing some R600's to some press in Europe and that fact on its own isn't enough to write an entire article with. So Theo filled the white space with some bullshit.

Sobek

26-Mar-2007, 18:05

So, in other words, AMD is apparently showing some R600's to some press in Europe and that fact on its own isn't enough to write an entire article with. So Theo filled the white space with some bullshit.

My god, that is EXACTLY what I was thinking!

It's like he could've just used one sentence, two at most, but for the sake of appearing to have an interesting and detailed article, he bloated it with useless crap. :razz:

suryad

26-Mar-2007, 19:18

neliz

26-Mar-2007, 20:29

Think its Africa not Europe.

Knocking on the walls of fortress europe.

I still hate them for not having the event in Amsterdam

Russell

27-Mar-2007, 05:47

I still hate them for not having the event in Amsterdam

Not that this is 100% reliable or anything.

Sazar

27-Mar-2007, 06:56

Inq (http://theinquirer.net/default.aspx?article=38457)

So 400 R600 boards and approx 200 journalists are going to a city west of Carthage, the place where Hannibal started its campaign against the Romans. The historical background only goes so far. Carthage is not AMD and Intel is not Rome, and although Hannibal could have had Rome, he and his elephants didn't make most people in western Europe non-Romance speakers.

Hannibal unfortunately was shite at siege warfare, hence why he couldn't take Rome regardless of how much he wanted. Plus he was shite at alliances.

If this is about analogies, might not bode well for AMD. Good thing Theo Valich ain't too bright nor is he, apparently, a historian :)

Russell

27-Mar-2007, 08:22

I can't help but point out that Fudzilla is reporting RV610 and RV630 delayed.

I wonder what this does for R600, assuming of course that he's right (which, sadly, he probably is...Fuad or not, ATI does love their delays).

http://www.fudzilla.com/index.php?option=com_content&task=view&id=306&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=298&Itemid=1

*Groan*

hoom

27-Mar-2007, 13:23

a city west of Carthage, the place where Hannibal started its campaign against the Romans.So the launch is in Valencia (http://maps.google.com/maps?q=39.676389,-0.273333&ie=UTF8&z=7&ll=39.774769,-1.362305&spn=7.149954,20.917969&om=1) then.
Wasn't the original location Barcelona? Close enough :wink3:

Hannibal unfortunately was shite at siege warfare, hence why he couldn't take Rome regardless of how much he wanted.After kicking things off by successfully taking Saguntum quickly & by storm (after years of expansion in Iberia which included a bunch of other sieges), he became shite at siege warfare by leaving his heavy & slow moving siege gear behind in order to get to Northern Italy before the Roman army could be put in place to intercept him.
Also the ~40,000 soldiers that Rome was able to whip up in short order put paid to a possible siege of the city by Hannibal's victorious but depleted army of about 20,000 after Cannae.
Besides, his goal was not to defeat Rome itself but to strip it of the vital support provided by the allied Italian states & thus politically contain Rome.

neliz

27-Mar-2007, 13:55

Besides, his goal was not to defeat Rome itself but to strip it of the vital support provided by the allied Italian states & thus politically contain Rome.

You mean, the other green will cut off support for R600 based products?

oh .. wait.. this isn't the R600 rumors & speculation thread... :sad:

hoom

27-Mar-2007, 14:05

Heh they will concentrate on trying to retake leadership in chipsets (http://maps.google.com/maps?f=q&hl=en&q=sicily&layer=&ie=UTF8&z=9&ll=37.50101,14.37561&spn=1.919617,5.229492&om=1) instead :lol:

Ailuros

27-Mar-2007, 21:42

So the launch is in Valencia (http://maps.google.com/maps?q=39.676389,-0.273333&ie=UTF8&z=7&ll=39.774769,-1.362305&spn=7.149954,20.917969&om=1) then.
Wasn't the original location Barcelona? Close enough :wink3:

*looks around if the administrator isn't watching...* to further build on your sarcasm....he sucks with geography just as much as with history. Almost half the mediteranean isn't exactly close heh....

Arty

28-Mar-2007, 16:59

Well Fuad is now saying that the core frequency is 750-800Mhz and memory for the GDDR3 board is 800Mhz. That works out to 102.4GB/sec. That sounds very conservative, like a single slot SKU.

Russell

28-Mar-2007, 17:38

Well Fuad is now saying that the core frequency is 750-800Mhz and memory for the GDDR3 board is 800Mhz. That works out to 102.4GB/sec. That sounds very conservative, like a single slot SKU.

I think both XTX and XT will be dual-slot and that Fuad is definitely talking about XT.

As for this (http://www.fudzilla.com/index.php?option=com_content&task=view&id=334&Itemid=1) article, I'd really like to see a source for this. If indeed some university researcher xrayed the thing and is convinced that G80 has 160 shaders (like with the previous rumor) then things could get interesting.

Anon Lamer

28-Mar-2007, 18:40

trinibwoy

28-Mar-2007, 18:40

Well Fuad is now saying that the core frequency is 750-800Mhz and memory for the GDDR3 board is 800Mhz. That works out to 102.4GB/sec. That sounds very conservative, like a single slot SKU.

A 800Mhz core sounds conservative now? Do we forget that dual-slot boards were running at slightly over 400Mhz a short while ago? :lol:

swaaye

28-Mar-2007, 19:18

trinibwoy

28-Mar-2007, 20:29

It sounds conservative after reading about a 1.35 GHz shader clock for several months.... Let's just hope they aren't hurting on IPC and are cranking that core for all it's worth just to match up. It would be like X1800 all over again...

Well G80's shader clock actually hasn't had any impact on my expectation of core speeds - the 575Mhz core clock kept those expectations well in check :) G80 is very much an exception - not the rule.

Arty

28-Mar-2007, 21:33

A 800Mhz core sounds conservative now? Do we forget that dual-slot boards were running at slightly over 400Mhz a short while ago? :lol:
No, I meant the mem clock is conservative. Conservative enough to be single slot. I guess I should've worded it better.

trinibwoy

28-Mar-2007, 21:45

Yeah understood but since when has memory speed been a major factor in the single/dual slot decision?

Arty

28-Mar-2007, 21:55

Yeah understood but since when has memory speed been a major factor in the single/dual slot decision?
I dont know exactly but I vaguely remember the use of 0.9ns in the GTX512 reminded where it got really hot.

trinibwoy

28-Mar-2007, 22:05

Oh I'm tempted to think that it had a lot more to do with pushing 110nm to 550Mhz :)

IbaneZ

29-Mar-2007, 13:11

Linking to fudzilla.com is fun! :smile:

http://www.fudzilla.com/index.php?option=com_content&task=view&id=345&Itemid=1

We managed to get some early numbers from the R600 and this baby scores 11400+ in 3Dmark06. The test bed included Quad core Intel CPU and 2 GB memory.

neliz

29-Mar-2007, 14:37

oh noes.. I always thought the XT going to do 14k?! AMD is doomed!

trinibwoy

29-Mar-2007, 14:45

Yeah but it's going to score 11400 with 8xAA too :twisted:

_xxx_

29-Mar-2007, 14:57

Yeah but it's going to score 11400 with 8xAA too :twisted:

And it'll also give you a massage after a hard day and make you a great meal :yes:

Razor1

29-Mar-2007, 16:16

And it'll also give you a massage after a hard day and make you a great meal :yes:

rather have a real girl friend do that ;)

INKster

29-Mar-2007, 16:41

"First rule of Fight Club is..." (http://img65.imageshack.us/img65/4214/cebitr600ndajc4.jpg) ;)

Bouncing Zabaglione Bros.

29-Mar-2007, 16:55

"First rule of Fight Club is..." (http://img65.imageshack.us/img65/4214/cebitr600ndajc4.jpg) ;)

At this point, leaks of R600 being quite spectacular is the only thing that will stop people throwing up their hands in the air and buying a G80 out of frustration with AMD not providing anything competitive for half a year.

Deusp

29-Mar-2007, 17:27

Linking to fudzilla.com is fun! :smile:

http://www.fudzilla.com/index.php?option=com_content&task=view&id=345&Itemid=1

They're getting a 10.6k on the 8800GTX. So the rumors sorta correspond, with the R600 being marginally faster than than the G80, but with vastly great power and cooling requirements, and with a better process and more memory bandwidth. Pretty lame all around I would say.

Natoma

29-Mar-2007, 17:30

Linking to fudzilla.com is fun! :smile:

http://www.fudzilla.com/index.php?option=com_content&task=view&id=345&Itemid=1

Useless without AA/AF information.

vertex_shader

29-Mar-2007, 17:38

"First rule of Fight Club is..." (http://img65.imageshack.us/img65/4214/cebitr600ndajc4.jpg) ;)

AMD sometimes looks very paranoid.

pelly

29-Mar-2007, 17:46

Vendors are usually only this strict and paranoid of leaks when there is something either extremely good or extremely bad.....Here's to hoping for extremely good....I'd love to see the GPU industry take yet another big leap forward....

Tim Murray

29-Mar-2007, 17:49

At this point, leaks of R600 being quite spectacular is the only thing that will stop people throwing up their hands in the air and buying a G80 out of frustration with AMD not providing anything competitive for half a year.
Just about anything can be spectacular from certain points of view...

Geo

29-Mar-2007, 18:08

Just about anything can be spectacular from certain points of view...

Now, now, don't go all Obi-wan on us! :lol:

Or are you suggesting niche brilliance somewhere, like say GPGPU, that may or may not be matched with more mainstream brilliance in games?

Sobek

29-Mar-2007, 18:13

Useless without AA/AF information.

And resolution, driver settings, Windows version, driver version, and a few other niggly system specs that can potentially impact 3dmark06 scores. There's absolutely nothing to go on there, he might aswell state that G90 gets 20,000 points.

Tim Murray

29-Mar-2007, 18:17

Or are you suggesting niche brilliance somewhere, like say GPGPU, that may or may not be matched with more mainstream brilliance in games?
:eek: Maaaaaaaaybe... I'm basing this on the teraflop-in-a-box demo, supposed leaked benchmarks, and Mike Houston's comment that "G80 is more like 330GFlops with user available shader resources" on the gpgpu.org forums. Clearly, R600 is a ton faster in GPGPU apps, but nobody's reporting that kind of difference (>50%) in 3D apps.

3dilettante

29-Mar-2007, 18:43

Geo

29-Mar-2007, 18:59

Something would have to be holding R600 back if it can't manifest something close this supposed 50% advantage over G80.

The teraflop in a box demo was closed-box. Any potential bottleneck that would be unavoidable in real 3d applications may have been worked around in a toy software loop.

There's all kinds of bottlenecks that could come into play when you consider "the leaked benchmarks" we have so far are mostly either 1) trash or 2) default (no aa/af) 3DM numbers that even if they aren't trash are useless for high-end gpus. GPGPU is quite a different kettle of fish, as R580 vs G7x proved conclusively.

northfieldz

29-Mar-2007, 19:06

And judging by the specs, I can tell NV30/35 should be at least 50% faster than R9700/9800.

Sigh, I guess fanboys never give up...

To be honest, to compare R600 with NV30 is an insult to NV30, since after all, NV30 is only 4 months later.:lol:

3dilettante

29-Mar-2007, 19:08

I can't rule out the very likely possibility those benches are garbage.

If it's not that, I don't know how much AA or AF would affect the math throughput in a GPGPU demo like the teraflop one.
Unless R600 is able to utilize those resources as additional FLOPS, the baseline should be 50% higher.

How could a 50% math advantage fail to produce significant performance gains just because AA and AF are off?

northfieldz

29-Mar-2007, 19:10

northfieldz

29-Mar-2007, 19:14

By the way, in theory, NV30 has a 25.6GFLOPS when it comes FP32, and a >50GFLOPS when it comes to FP16, while R300 has a 20.8GFLOPS math power.

3dilettante

29-Mar-2007, 19:23

Talk about comaring apple to orange: Vector-based flops is not as useful as scalar based flops.

I'm comparing theoretical FLOP counts.
If there is a mathematical advantage to R600, some aspect of current apps would have to show it. It has to materialize as a real gain somewhere.

If the reason R600 can't take advantage of its theoretical advantage is that it is suffering from vector utilization issues (we don't know that is true, and some rumors seem to indicate it isn't that), it goes back to my argument that something is holding R600 back, something that might be easier to hide in a closed-box demonstration.

Creig

29-Mar-2007, 19:40

And resolution, driver settings, Windows version, driver version, and a few other niggly system specs that can potentially impact 3dmark06 scores. There's absolutely nothing to go on there, he might aswell state that G90 gets 20,000 points.

Congratulations. You've just given Fuad his news headline for tomorrow. It'll read something like...

"G90 to hit 20,000 in 3DMark06!

An anonymous source has leaked secret benchmarks that the G90 can hit an AMAZING 20,000 points in 3DMark06! Obviously we can't reveal the source of our information with such an amazing scoop, but we'll keep you updated with additional benchmarks/info as soon as our source reveals them!"

Except with more typos.