[-The_Mask-]

Originally Posted by psolord

Nice ideas mate, but I almost had a stroke when I saw 450$ for the Cayman and 275$ for Barts. lol

You haven't seen the prices from Informatique?

And I have my reasons for the 1120 SP's.

[jaredpace]

Originally Posted by DavidGraham

Juniper had a rectangular look too
that suggests Cayman could be rectangular too .

Yeah it could be, I don;t know.

Originally Posted by Mianca

Yeah, it would seem reasonable that clocking a ~230mm² chip high enough to reach the performance segment they indicated actually came at a considerable Perf/W penalty.

I'm really interested in how they'll pull off a 2xCayman Antilles card with Cayman XT supposedly drawing somewhere between 225W and 300W ...

This is explained a lot already in this topic. CaymanXT will be volted higher as a single chip sku, it will use a lot of power: ~225w. They will bin the cayman xt chips, cherry pick lower power cores and hit them with a lower stock vid. It's the same process they did for Cypress & Hemlock. Two low power undervolted binned Cypress cores that are underclocked. When they have a 188w TDP on the 5870 boards with ~1.168v and can actually pull ~235w when doubled up at 1.056v on a Hemlock board, you can extrapolate that info and get ~225w & ~290w for Cayman & Antilles. And there is room up or down for Antilles as far as power consumption goes. Then they taunt the 6k chips as even more power efficient, the boards with CHil PWMs. Anyways that question will get asked and answered every few pages or so for the next 2-3 months. Then someone will come along saying, "you know what i believe antilles is 2xbarts based.." They have all kinds of technologies like ULPS, Powerplay, and clockgating. Hardware advancements like 2nd gen 40nm binning, decap rings, modified clock states, reworked board components, thicker pcbs, lower leakage designs, I would expect 6990 to have a 300w TDP and 260-295w usage in 3D gaming compared to 300w TDP and 225-240w load with the 5970. And it'll still be equal to or less power hungry than a 4870x2.
5970 pulls less than a 480, there is PLENTY of room for Antilles, even with just 6pin+8pin on the board. People doubted 2900xt, 3870x2 spartan, hemlock, etc...

[Mianca]

Originally Posted by -The_Mask-

And I have my reasons for the 1120 SP's.

I knew I had already read that number somewhere somewhen ... what a coincidence:

Originally Posted by -The_Mask-

HD5700 will be with 1120 stream processors and 56 texture units TMU, the memory is GDDR5. Far as we know there will be HD5770 (199 dollars) and HD5750 (149 U.S. Dollar), HD5770 will replace the current HD4890, the HD5750 is to replace the HD4870. HD5700 opponent is to the NVIDIA GTX275, GTX260, GTS250 and the future D10P GT200b series. Others like DX11, ATI Steam, UVD2, Eyefinity, CrossFire technologies are also readily available.

[-The_Mask-]

I've written that?

[Mianca]

Originally Posted by -The_Mask-

I've written that?

A bit more than a year ago in the R8xx speculation thread, yes

[-The_Mask-]

Originally Posted by Mianca

A bit more than a year ago in the R8xx speculation thread, yes

Can't find it in my history on B3D and I can't remember...

[GZ007]

Iam curious what will hapen to 5850 and 5870 prices with 6850 running for around 190 € (a bit slower than 5850 but new features and maybe better dx11). They are still high.

[ShaidarHaran]

Originally Posted by neliz

4? more like a week or so.

My comment was in response to Rangers' comment that the launch/NDA expiry was on the 18th, with yesterday (the 14th) being the press day. That would mean only 4 days for reviewers to play with both cards. I haven't signed an NDA but I'm not going to ruin AMD's good will by admitting when it does expire, though I'll just say the answer has definitely been posted.

Edit: just saw the CC transcript in which Dirk Meyer mentions the launch next week. no sense playing dumb now

[mczak]

Originally Posted by Duck Dodgers

Well as AMD proclaims it's their 2nd generation DX11 launch, you'd want to come close to or match the competition on the tessellation feature with your second coming, which mind you was one of the big sells of DX11 to begin with. I'd say they will since they were behind on this front.

Actually, looking at some tesselation numbers again (apart from the unigine/stone giant scores where the performance difference between GTX460/HD5850 isn't actually THAT big) (http://www.hardware.fr/articles/795-...e-gtx-460.html) I think eclipsing GF104 numbers shouldn't be too hard. With tesselation, the GF104 seems just barely able to output more than 1 tri/clock. That would mean Barts could (thanks to the higher clock) actually exceed that even with 1 tri/clock limitation. Despite AMD claiming tesselation shouldn't be limited to 1/3 tri rate of non-tesselated case, all tests showed this limitation - so if AMD can change that (whatever was causing it - thread generation?) that would be enough to beat GF104. Actually, just a factor of 2 increase would probably be enough for Barts not to be slower in Unigine Extreme or Stone Giant. So maybe it'll still look a bit weak compared to GF104 on paper but turn out to be at least as fast in practice.

[ShaidarHaran]

Originally Posted by mczak

Actually, looking at some tesselation numbers again (apart from the unigine/stone giant scores where the performance difference between GTX460/HD5850 isn't actually THAT big) (http://www.hardware.fr/articles/795-...e-gtx-460.html) I think eclipsing GF104 numbers shouldn't be too hard. With tesselation, the GF104 seems just barely able to output more than 1 tri/clock. That would mean Barts could (thanks to the higher clock) actually exceed that even with 1 tri/clock limitation. Despite AMD claiming tesselation shouldn't be limited to 1/3 tri rate of non-tesselated case, all tests showed this limitation - so if AMD can change that (whatever was causing it - thread generation?) that would be enough to beat GF104. Actually, just a factor of 2 increase would probably be enough for Barts not to be slower in Unigine Extreme or Stone Giant. So maybe it'll still look a bit weak compared to GF104 on paper but turn out to be at least as fast in practice.

Barts won't be outputting even 1 tri/clk though even when tessellating. It's more like 4 triangles every 5 clocks.

[fehu]

Originally Posted by jaredpace

This is explained a lot already in this topic. CaymanXT will be volted higher as a single chip sku, it will use a lot of power: ~225w. They will bin the cayman xt chips, cherry pick lower power cores and hit them with a lower stock vid.

the new x850 Press Edition?

[fbuffer]

Originally Posted by fehu

the new x850 Press Edition?

more like Fermi 512 Edition

[mao5]

txssxxxxxx AVPLP2

[Gipsel]

Originally Posted by mao5

TxxxSxxx AVP

Do we have to understand that?

[chavvdarrr]

Originally Posted by mao5

TxxxSxxx AVP

[Babel-17]

Huh, Dell is offering $150 off all Alienware videocard upgrades (ATI and nVidia), are they moving inventory because they are in the know?

[Topman]

Originally Posted by mao5

txssxxxxxx AVPLP2

hmm... improved tesselation? Alies vs Predator and Lost Planet 2?

[mao5]

Originally Posted by Topman

hmm... improved tesselation? Alies vs Predator and Lost Planet 2?

I have no idea, just wondering if it will be that?

[Kynes]

Originally Posted by mao5

txssxxxxxx AVPLP2

Sometimes I can't follow you, and I'm almost sure that I'm not the only one.

[ZerazaX]

Originally Posted by Babel-17

Huh, Dell is offering $150 off all Alienware videocard upgrades (ATI and nVidia), are they moving inventory because they are in the know?

Of course they are in the know. Why they do it though, is another question

[trinibwoy]

Originally Posted by ShaidarHaran

Barts won't be outputting even 1 tri/clk though even when tessellating. It's more like 4 triangles every 5 clocks.

What do you mean by "even when tessellating"? Right now triangle throughput on Evergreen is faster with tessellation off than on. Is that going to change?

[no-X]

Originally Posted by AnarchX

Could it be possible that only Cayman has 4D VLIW ALUs?

Yes, it is possible.

[mczak]

Originally Posted by ShaidarHaran

Barts won't be outputting even 1 tri/clk though even when tessellating. It's more like 4 triangles every 5 clocks.

Ah you seem to know more than I do .
In any case that wouldn't be too bad neither. If you look at Tridam's numbers the maximum he got on the GTX460 was 750MTris/s with tesselation - and that only with "high culling" (radeons did not show any effect on culling with triangle rate). So 4 tris every 5 clocks at 900Mhz would be 720MTris/s - which is just slightly less and if Barts continues to not show dependency on culling it would be quite a bit faster without culling compared to GTX460 - of course, that's not a full GF104 chip so against a 8 SM GF104, clocked a bit higher (I'm still waiting for this...) it would be a bit worse, but still in the same ballpark. If it actually can do 4 tris every 5 clocks in practice that is still more than twice over what we saw with Redwood/Juniper/Cypress. Might not be class-leading but "good enough". Of course, with Barts Pro the numbers would be worse, but not terrible neither.

[Alexko]

Originally Posted by no-X

Yes, it is possible.

But very unlikely. Why would AMD not implement those in Barts? Just to launch it a month earlier?

[DarthShader]

There's an article at SA, written by LGN, summarising the rumours about the 6800 series. Not much new stuff, but there is following piece of interesting info:

Early 6870 boards will use eight layer PCBs, while later on we'll see cheaper six layer partner boards. The 6850 will use six layer boards as standard, but there's an option here for "premium" boards which will use eight layers.

Could that be the reason we see the big difference in PRO and XT price? That means the 6870 should soon drop, unless supply won't meet with demand.