AMD: R9xx Speculation

[keritto]

In which case it should have just been called 67xx if it's coming in at a lower price. The ONLY reason to call it a 68xx is to deceive and mislead the consumer. Hence, it is complete fail, IMO.

Not to mention making a complete mockery of the consistent and practical naming scheme since 3xxx. We're basically going back to the dark ages of naming schemes where performance is just randomly assigned numbers with virtually no consistency. Again, even more fail...

Dont believe it would be a fail they could easily price it 250/200USD even at intro and have huge return. Look at hd 5850/5870 they never stumble down below ~300/450USD (thanks nVidia braniax and their scheming)

btw, those "misleading naming scheme" don't believe it happened first time looking at x1950pro/gt/xtx or x2900pro/xt parts, but your absolutely right after they skipp 2900->3870 in same generation they really didnt approach to naming scheming. It's not something they really need. But hey they didnt ever reasonably explain why they change name to HD6000 series! After all it's only HD5000 with "turbo-boosted" logic to oversaturate SP that they change. It would not even bring to us considerable TDP reduction, if we assume that this is "fully mature 40nm (V.3.0) TSMC node" and that chip is now 30% smaller than RV870/Cypress.

Means, that 5870 (Cypress) should have been priced under 200 USD since 4870 had already fallen below 150 USD by the time it launched. And I believe was closer to 100 than 150 USD at many places already when 5870 launched.

4870 had intermediary in hd4890 incarnation and price falls were mostly due ATi not having anything to compete in mainstream market after ridiculosly low number of 4770 came to market (TSMC hilarious 40nm bug V.1.0 .. yada-yada)

btw. where did you get those bunch of 100USD hd4870 (even stock 512MB) i'd rather bought half-a-dozen at those prices at that time. You know what's most ridiculous ... that now when 4870 is one year obsolete you can still bought 1GB "only" part at some insane 250US+VAT prices (in selected countries )

 

[Rangers]

I thought the NDA was up today??

 

[AlphaWolf]

I thought the NDA was up today??

Think today was just press day, so probably a few more days for NDA.

 

[jimbo75]

So I'm thinking there must be a reason why Barts XT is 900mhz instead of the usual 850.

900mhz being just enough to take it past the 5870 in performance, without badly affecting yields at the expense of slightly higher than anticipated TDP. Good, bad or plain ugly theory?

 

[AlphaWolf]

So I'm thinking there must be a reason why Barts XT is 900mhz instead of the usual 850.

900mhz being just enough to take it past the 5870 in performance, without badly affecting yields at the expense of slightly higher than anticipated TDP. Good, bad or plain ugly theory?

Most likely to hit a certain TDP for oems, while still getting the most performance to be under that mark.

 

[RedVi]

So I'm thinking there must be a reason why Barts XT is 900mhz instead of the usual 850.

900mhz being just enough to take it past the 5870 in performance, without badly affecting yields at the expense of slightly higher than anticipated TDP. Good, bad or plain ugly theory?

To fool people into thinking it's better than 5870?

 

[Duck Dodgers]

I think the odds of tesselation performance being better than what GF104 offers are pretty slim. The other question though is why would you want that, just so that it's faster on unigine extreme?

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.

And to the AMD apologists here I would retort with caveat venditor, let the seller beware because the consumer isn't presented with choice here. The weather has changed and is sliding in a discrete graphics duopoly. Notwithstanding superficial tribal allegiances you'd really have to stand back and ask yourselves why be complicit in a "win" that might burn you someday lest you say "not good enough" now.

 

[ShaidarHaran]

Think today was just press day, so probably a few more days for NDA.

Today was the press day.

 

[Rangers]

Must be the original 18th then.

Though that's supposedly the on sale date I think too...ridiculous.

 

[ShaidarHaran]

Must be the original 18th then.

Though that's supposedly the on sale date I think too...ridiculous.

4 days to review 2 new products would be cutting it pretty close, in my book.

 

[neliz]

4 days to review 2 new products would be cutting it pretty close, in my book.

4? more like a week or so.

 

[Rangers]

NDA apparently ends Friday 22nd. Suppose most of you knew that.

 

[onethreehill]

AMD Quietly Reveals Next Radeon Series, Launches Next Week
The cat is out of the bag, so to speak, after today's earnings conference call for AMD. One of the things that ended up being discussed by AMD CEO Dirk Meyer in today's call was the future of AMD's graphics division, where AMD's "second-generation DX11" GPUs were mentioned.

We will be launching our second-generation DX11 graphics offerings next week.

Later on, he also had the following to say about what's launching and what the expected volume is:

We'll be introducing our second-generation of DX11 technology into the market with some launch activities actually next week. We'll be shipping all the family members of that product line I'll call it, by the end of this quarter, and total volume think in terms of several hundred thousand, or hundreds of thousands of units....

http://www.anandtech.com/show/3976/amd-quietly-reveals-next-radeon-series-launches-next-week

 

[Kaotik]

Slightly side of the topic, but Viewsonic has launched it's first AMD-certified 3D display
http://www.maximumpc.com/article/news/viewsonic_launches_amd-certified_3d_led_display
Could this also be the first incarnation of Bit Cauldron's RF-using 3D glasses, which were designed with tight co-operation with AMD? After all, they said they won't sell the glasses themselves, but rather they'll be sold by 3rd partiesl. (Curious about those glasses is that Bit Cauldrons site gives the impression that they only work with AMD GPUs ("Transmitters available for 3D-Ready TV’s and PC’s with AMD GPU’s"))

 

[Mianca]

Viewing those two slides side-by-side, I can just hope that the >150W Max Board Power Barts XT card will actually trump GTX 460 by a fairly huge margin ... anything less than ~ HD 5850 performance (i.e. ~ GTX460 + 20%) would practically correspond to a decrease in Perf/W ...

I really couldn't care less about the controversial branding etc. - but Perf/W is something I (and a lot of reviewers) keep a very close eye on ...

 

[AnarchX]

At Chiphell was posted an interesting 3DMark Vantage screen which is supposed to be from a HD 6850. The interesting stuff are the feature test.


Feature Test 1 (texture) & Feature Test 6 (perlin noise ALUs) ~ HD 5750 level
Feature Test 2 (Pixel ROPs) & Feature Test 5 (GPU particles, with high vertex shader load) ~ HD 5850 level

 

[SimBy]

Viewing those two slides side-by-side, I can just hope that the >150W Max Board Power Barts XT card will actually trump GTX 460 by a fairly huge margin ... anything less than ~ HD 5850 performance (i.e. ~ GTX460 + 20%) would practically correspond to a decrease in Perf/W ...

I really couldn't care less about the controversial branding etc. - but Perf/W is something I (and a lot of reviewers) keep a very close eye on ...

Interesting point. I don't think there's much if any improvement as far as performance/W goes. In contrast to substantial performance/mm2 gain.

Unless Barts XT is faster than Cypress XT but I kinda doubt that.

 

[Mianca]

I don't think there's much if any improvement as far as performance/W goes.

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 ...

 

[Kaotik]

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 ...

Handpicked Cayman Pros?

 

[Arnold Beckenbauer]

http://www.microsofttranslator.com/...pu.org/viewthread.php?tid=3405&extra=page%3D1


i.e. 1/2 for POW and 1/2.5 for SIN.

For HD5870, GPU Shader Analyzer shows 129 ALU instructions for the POW shader, i.e. half MUL rate. Here's how 4 POW instructions compile:
...

SIN is a bit of a tangle to compile in GPUSA, but with a bit of fiddling it comes out as 162 instructions, i.e. 1/2.5.

...
As far as I can tell the throughput for XYZT would be the same as XYZWT in all three of these tests.

But I think it rules out XYZW with emulated transcendentals.

HD4850 (700 MHz):

--------------------------------------------------------------------------
                             Instruction Issue
--------------------------------------------------------------------------
512      109.5269       ADD          4         64
512      109.7506       SUB          4         64
512      109.7168       MUL          4         64
512      109.6555       MAD          4         64
512      109.3467       EX2          4         64
512      109.3772       LG2          4         64
[B]512      55.4172       POW          4         64[/B]
512      136.9928       FLR          4         64
512      136.7206       FRC          4         64
512      109.6678       RSQ          4         64
512      109.7598       RCP          4         64
[B]512      44.0939       SIN          4         64[/B]
512      44.0870       COS          4         64
512      104.8287       SCS          4         64
512      111.4375       DP3          4         64
512      111.4533       DP4          4         64
512      87.8590       XPD          4         64
512      109.5514       CMP          4         64

--------------------------------------------------------------------------
                     Scalar vs Vector Instruction Issue
--------------------------------------------------------------------------
512      110.9219       ADD          1         40
512      108.3673       ADD          4         40
512      110.6266       SUB          1         40
512      108.3721       SUB          4         40
512      110.9219       MUL          1         40
512      108.3673       MUL          4         40
512      110.7065       MAD          1         40
512      108.4488       MAD          4         40