AMD: Southern Islands (7*** series) Speculation/ Rumour Thread

[trinibwoy]

I don't see anything that wrong with either the paper launch or high MSRP. AMD has no obligation to give us things on the cheap, their goal should be quite the opposite. All it means is that they will sell all the cards they can make at $550. They can also claim to have "launched" the card in 2011, much like Fermi launched in 2009 At least we got reviews and a hard availability date not too far in the future.

The only thing that could throw a damper on it is a hard launch of a competitive GK104 in Q1 and that doesn't seem likely at this point.

 

[Alexko]

I will give you just two reasons:
1. A HD7970 is more expensive to produce than a HD5870 (larger die, higher cost of 28nm wafer, 50% more memory chips and so on).
2. The price situation on the market is different. While the HD5870 was roughly the same amount faster than a GTX285 as the 7970 tops now a GTX580, the GTX285 was way cheaper back then than a GTX580 is today. In fact, AMD priced the HD5870 about 20% higher than the GTX285 (at least that was the price situation here in September 2009). That was (relatively speaking) higher than they are pricing the HD7970 now. And maybe one should consider that the MSRP of a 3GB HD7970 is comparable to or even lower than the current street price of a 3GB GTX580.

That's right. Besides, Cypress wasn't the first 40nm GPU, RV740 was. And NVIDIA was making DX10.1 40nm GPUs at the same time, so the 40nm process really wasn't as exclusive to Cypress as 28nm is to Tahiti.

That "very high end" is obviously the joke of the day. Call it very high end marketing and I'll agree immediately.

How so? Apart from dual-GPU cards—which have annoying drawbacks—it's the fastest card available by a comfortable margin.

And if it shouldn't trigger a price war? As much as AMD isn't a charity organisation, NVIDIA is times less one.

If prices remain high, I'll happily concede that AMD has a new strategy, and one that doesn't favor consumers. NVIDIA is no charity, but they need competitive pricing to make money, as does AMD. This is how price wars get started, and price wars do happen, even though no company is a charity.

 

[Alexko]

I don't see anything that wrong with either the paper launch or high MSRP. AMD has no obligation to give us things on the cheap, their goal should be quite the opposite. All it means is that they will sell all the cards they can make at $550. They can also claim to have "launched" the card in 2011, much like Fermi launched in 2009 At least we got reviews and a hard availability date not too far in the future.

The only thing that could throw a damper on it is a hard launch of a competitive GK104 in Q1 and that doesn't seem likely at this point.

Paper launches are annoying when a company claims availability for a product that's nowhere to be found. Here, AMD is basically just allowing reviews ahead of the actual launch date. If anything, that's good for the curious people we are.

 

[TKK]

To further add to Gipsel post:

Additionally, when 5800 launched, there were still rumors floating around that GTX 480 would come in Nov.09, with full 512 SP, 64 TMU and clocks of 750/1500/4200 for chip/shader/memory. Maybe AMD was afraid of that and kept prices lower than necessary as a sort of 'preemptive strike', to prevent people from waiting for a potentially far superior Fermi. Which didn't arrive until a year later in the form of GF110, obviously.

 

[rpg.314]

Paper launches are annoying when a company claims availability for a product that's nowhere to be found. Here, AMD is basically just allowing reviews ahead of the actual launch date. If anything, that's good for the curious people we are.

I don't like paper launches unless backed by a credible availability date.

 

[Alexko]

I don't like paper launches unless backed by a credible availability date.

We have a date, January 9. As to whether it's credible, well…

 

[Ernestds]

The problem is AMD sent wrong samples to press. We discovered, that our card doesn't have original thermal grease, but some crappy one, which was applied in a really wrong way (uneven, very thick layer etc). We exchanged it and temperatures, noise and esp. power consumption went down. Chech the table at the bottom of this page.

Could you do some further testing? Mainly power consumption at different fan's level. I mean, that is a huge difference in energy used. Could that indicate that the process doesn't like high temperatures?

It would be nice if you tested power level at one more game, like Crysis 2 or Metro 2033.

Thanks.

 

[denev2004]

That seems unpossible to me.

~520mm2 to 250mm2 going from 40nm to 28nm? That would be some ...serious... optimizations, quite the feat of engineering. I would expect a pure die shrink to be around 350mm2, and then shrink more if there were layout optimizations to be had, or features to drop.

Anyway, back to Tahiti.

Speeed! Speeeeeeeeeeeeeeeed!

At 2560 and higher resolutions.

I think it sounds reasonable a 520mm2 move to 400mm2 going form 40nm to 28nm...Still,not enough for them to think of using this way and have a higher frequency

 

[Blastman]

So is there any news on ATI/AMD's other 7000 series cards like the 7870/7850 -- code name Pitcairn XT, Pitcairn Pro? Is Pitcairn a die shrink of the Cayman or GCN?

 

[caveman-jim]

I will give you just two reasons:
1. A HD7970 is more expensive to produce than a HD5870 (larger die, higher cost of 28nm wafer, 50% more memory chips and so on).
2. The price situation on the market is different. While the HD5870 was roughly the same amount faster than a GTX285 as the 7970 tops now a GTX580, the GTX285 was way cheaper back then than a GTX580 is today. In fact, AMD priced the HD5870 about 20% higher than the GTX285 (at least that was the price situation here in September 2009). That was (relatively speaking) higher than they are pricing the HD7970 now. And maybe one should consider that the MSRP of a 3GB HD7970 is comparable to or even lower than the current street price of a 3GB GTX580.

And the price of the 285 was lower because... of RV770. GTX 280 hit at 650, 4870 dropped at 299 and NV was issuing triple digit rebates. 285 refresh came in at 400, and boom, there's 5870 at 379. NV hits back with 480, then 580 at 500, the only AMD card at 500 was the 5870 E6 2GB, the 6970 hits at 369... it's a pattern based on competitive performance analysis. So, the strategy here is either GK100 isn't going to be faster, or at least only 10% faster, than Tahiti, so NV will aim for 499 again, AMD can drop down to 479 having made boatloads of cash for 6mo while it was the best solution in the price range. the question becomes, what does the 580 price drop to (and we'll see if it needs to drop if the 6950 puts pressure on it by being 449 and the same performance or something).

sometimes I wonder what the price would be if the tag was based on cost + margin, amortized over 2 years projected sales, instead of what the market will bear...

 

[Dave Baumann]

If it's only a title or two then they're doing a whole damn lot of damage to the average performance increase compared to its predecessor.

It is already 40%-50% fast than the predecessor, 35% in the title in question, that no small gap by any means and again, no reason to push things out. And the example given is not even a "driver" issue as this is something that is implemented in the game.

 

[Dave Baumann]

Here, AMD is basically just allowing reviews ahead of the actual launch date.

And this is largely driven by the timing relative to the holiday period. With both product shipping now and samples having to go to press pre-Christmas leaving that amount of time was bound to create leaks and some reviews going early; this thread over the last week or so is a testiment to the lack of respect over embargo material given to certain parties, leaving it another 2 weeks would have been a lot worse in that respect.

 

[Arty]

And this is largely driven by the timing relative to the holiday period. With both product shipping now and samples having to go to press pre-Christmas leaving that amount of time was bound to create leaks and some reviews going early; this thread over the last week or so is a testiment to the lack of respect over embargo material given to certain parties, leaving it another 2 weeks would have been a lot worse in that respect.

Kudos then, good move.

 

[Acert93]

And this is largely driven by the timing relative to the holiday period. With both product shipping now and samples having to go to press pre-Christmas leaving that amount of time was bound to create leaks some reviews going early; this thread over the last week or so is a testiment to the lack of respect over embargo material given to certain parties, leaving it another 2 weeks would have been a lot worse in that respect.

Yeah, I am not "getting" all the "Paper Launch" mantra from some. I am only loosely following the launch compared to those in this thread so maybe it is because I am not as invested in the exact launch moment of a piece of hardware I will never own (alas I am here reading about it though) and I thought it was painfully obvious based on the history of leaks that if reviewers were to have hardware another couple weeks it would be a big old fiasco. Moving the publication dates forward was a sensible solution. And lets not parse words: based on past leaks it would have created angst for no reason. And the cherry on the top is for those getting some cash for the holidays it gives them something fresh in mind to spend it on!

As for the actual release date I don't get how allowing reviews 2 weeks before physical launch is a "paper" launch. I think reviewers get pissy when they have tight deadlines and can score some clicks with the ol' scabs. Just change the product to software and the pre-release review becomes a "service" to consumers. In fact in many industries being able to release review material BEFORE availability is a good thing. When a company allows media to review products before release BUT give a firm release date down the road, that isn't a paper launch. That is access to pre-release material. It actually does consumer little good to wait until launch day to have reviews. I am sure the media want more time with the hardware but them the breaks as they say.

As for price, yeah the 7970 is expensive, but to deny it is top end single GPU... sheeesh. It has a pretty similar price to the 580 but is faster and more feature rich. It is called product placement. Regardless of how AMD has positioned products in the past, if NV is happy to have their single GPU product at $550 then AMD isn't hurting anyone by hitting with a faster, better product at the same price.

Dave, AMD should consider adopting more community outreach programs to help community members explain why a faster card at the same price as a competitors product is a good, not bad, deal?

On a more serious note it seems MSAA is a big reason for BF3 performance. Without MSAA and there is a pretty big performance gap. Dave, do you know why BF3 in particular is getting hit here so much more than other titles?

 

[neliz]

We have a date, January 9. As to whether it's credible, well…

I really hope it's not empty boxes we're shipping.

 

[shiznit]

Finally a BF3 benchmark w/o MSAA. Well done AMD, that is really fast on Ultra.

Eagerly waiting for the 7950 1.5 GB.

 

[jaredpace]

For the overclockers out there, this was stock volts to the core. Looking forward to the next afterburner controlling that chil regulator. At 1.2ghz it's just under a 590/6990.

 

[CarstenS]

I don't see anything that wrong with either the paper launch or high MSRP. AMD has no obligation to give us things on the cheap, their goal should be quite the opposite. All it means is that they will sell all the cards they can make at $550. They can also claim to have "launched" the card in 2011, much like Fermi launched in 2009 At least we got reviews and a hard availability date not too far in the future.

That's exactly my take on this as well.

And this is largely driven by the timing relative to the holiday period. With both product shipping now and samples having to go to press pre-Christmas leaving that amount of time was bound to create leaks and some reviews going early; this thread over the last week or so is a testiment to the lack of respect over embargo material given to certain parties, leaving it another 2 weeks would have been a lot worse in that respect.

Quite frankly spoken: You should know by now who of your partners and which press is prone to leakage (c'mon... the VR-zone pic or did they just hate one specific IHV and wanted to damage it's reputation - in any case things like these have happened at every launch I remember). I just didn't see any other company than Apple to have the balls to exclude those people from pre-launch information or launch events.

 

[Love_In_Rio]

I think it sounds reasonable a 520mm2 move to 400mm2 going form 40nm to 28nm...Still,not enough for them to think of using this way and have a higher frequency

Looking at the prior generation I see Ati packed the HD6970 2,6 billion transistors in 378mm2 and Nvidia 3 billions in 529mm2. Is this difference in density due to Nvidia hot clock?.

 

[Arnold Beckenbauer]

So AMD, now there is no way back!
http://www.amd.com/us/products/desktop/graphics/7000/7970/Pages/radeon-7970.aspx#3

Image quality enhancement technology

• Up to 24x multi-sample and super-sample anti-aliasing modes
• Adaptive anti-aliasing
• Morphological Anti-Aliasing (MLAA)
• DirectX® 10/11 Super-Sample Anti-Aliasing (SSAA)
• 16x angle independent anisotropic texture filtering
• 128-bit floating point HDR rendering