More news on Kepler and Southern Islands

opethdisciple · 21 Jul 2011 at 01:12

Article here

Sounds like the 7000 series from ATI will be power/heat efficient which was a strong point of the 5000 series.

mejobloggs · 21 Jul 2011 at 02:23

Hmm interesting. A pity TSMC often seem to have issues

I'm quite a fan of low power stuff though. I just wonder how much HPL will affect performance

I wonder if they'll do a refresh using HKMG so you can choose between power/performance. Would be neat.

Doubt it though, just me dreaming

Duesix · 21 Jul 2011 at 02:40

a refresh would be a good excuse to sell more though so it might happen??? rejuvenate the market

gurusan · 21 Jul 2011 at 04:14

So looks like we will expect Fermi fail all over again.

Marine-RX179 · 21 Jul 2011 at 07:34

gurusan said:
So looks like we will expect Fermi fail all over again.

Well, the question is though...how much potential performance of the HD7000 would be lost, with them switching from the 28nm HKMG high-performance processes to the 28nm HPL, which is basically a cut-down version of the 28nm HKMG process...

From the sound of it, the Southern Islands would no doubt beat the Kepler in terms of low power consumption, but how much slower would it be comparing to the Kepler which gonna use the full fudge 28nm HKMG high-performance processes...I guess that's what most of us would want to know

Mr. May · 21 Jul 2011 at 08:03

where is the compitition for rawwwww poweerrrr ?

sunny007 · 21 Jul 2011 at 08:28

When are the next gen cards due ?

WingZero30 · 21 Jul 2011 at 08:43

AMD NewZealand anyone? !! :eek:

It will be interesting to note the length of these cards.

An Exception · 21 Jul 2011 at 09:00

Marine-RX179 said:
Well, the question is though...how much potential performance of the HD7000 would be lost, with them switching from the 28nm HKMG high-performance processes to the 28nm HPL, which is basically a cut-down version of the 28nm HKMG process...

From the sound of it, the Southern Islands would no doubt beat the Kepler in terms of low power consumption, but how much slower would it be comparing to the Kepler which gonna use the full fudge 28nm HKMG high-performance processes...I guess that's what most of us would want to know

Tbh, AMD will probably release a refresh as soon as HKMG HP is actually ready, which will probably be before Nv also ironically.

Marine-RX179 · 21 Jul 2011 at 09:04

Ejizz said:
Tbh, AMD will probably release a refresh as soon as HKMG HP is actually ready, which will probably be before Nv also ironically.

Refresh...as in like how GTX260 192SP-> GTX260 216SP back then you mean? But it's not gonna help anyone that buy the AMD card on the first launch though, as it is not like AMD gonna recall them and replace with with V2 of the cards lol.

WingZero30 · 21 Jul 2011 at 09:08

Whatever the case guys, I think these cards will get seriously bottlenecked by my Q6600 even at 1920x1080p

Marine-RX179 · 21 Jul 2011 at 09:11

WingZero30 said:
Whatever the case guys, I think these cards will get seriously bottlenecked by my Q6600 even at 1920x1080p

To be honest it goes in this order:
Games not optimised for using Quad-core fully (using only 2-3 cores) bottleneck overclocked Q6600-> bottleneck 28nm mid-high range cards.

An Exception · 21 Jul 2011 at 09:11

theone91 said:
When are the next gen cards due ?

Going roughly by the rumours so far...

AMD = 2-3 Months.

Nvidia = 7-8 Months best case, as long as they don't need any revisions which is probably unlikely considering it's a new process + new arch + Nvidia, so best guess would be roughly 12 months time for Nvidia to get new 28nm architecture to market.
However, If the new arch is delayed due to Yields etc. Nv may release die shrinks of existing chips like GF104 etc. if the arch scales down well to 28nm.

Or the other sensible thing to do would be to release the smaller mainstream chip first to compete with the AMD mainstream, and release the GTX780's etc. when yields are fixed.

WingZero30 · 21 Jul 2011 at 09:18

How far can GPUs get in die shrinkage?

An Exception · 21 Jul 2011 at 09:20

Marine-RX179 said:
Refresh...as in like how GTX260 192SP-> GTX260 216SP back then you mean? But it's not gonna help anyone that buy the AMD card on the first launch though, as it is not like AMD gonna recall them and replace with with V2 of the cards lol.

Huh, seriously?

Think GTX280 to GTX285, did Nv recall all the GTX280's and replace them with V2's when they moved to a better process??? NO they just updated their product line and increased the numbers on their products a little, this is how the game works...

An Exception · 21 Jul 2011 at 09:22

WingZero30 said:
How far can GPUs get in die shrinkage?

Not sure what the question is exactly?

Marine-RX179 · 21 Jul 2011 at 10:19

Ejizz said:
Huh, seriously?

Think GTX280 to GTX285, did Nv recall all the GTX280's and replace them with V2's when they moved to a better process??? NO they just updated their product line and increased the numbers on their products a little, this is how the game works...

You are missing my point. All I'm saying is the AMD cards might hit the shelf earlier, but it is still not a full-fudge 28nm process card...as far as that is concerned, it is like selling a product with corners cut (comparing to the original plan and design) and rushing the product to shelf, unlike refresh in the pass which were actualy "improve on the process". I'm in no way trying to drag this down to become a Nvidia vs AMD discussion...I was using GTX260 as example for refresh because I couldn't think of an example from AMD/ATI side, so GTX260 was the first example came to my mind that's all.

Granted lots of us wouldn't care less for the AMD card being on the 28nm HPL process on first launch as long as the performance is a big enough jump, but I was just saying there would also be people that prefer waiting for cards on the HKMG high-performance processes instead, may it be AMD or Nvidia card.

An Exception · 21 Jul 2011 at 10:36

^^^
There's always people who get caught up playing the waiting game, look at all the folks who didn't jump on the 58XX's and decided to wait for the competition?
Did that even have any effect on product uptake? No, the GPU's were supply constrained anyway due to manufacturing capacity and also very popular, hence why retailers were able to obscenely price gouge.

It will be a similar story this time around, some people will want to 'wait it out' for 6 months or more, but most will get far to excited over the new tech and won't be able to help themselves...

Marine-RX179 · 21 Jul 2011 at 11:32

Ejizz said:
^^^
There's always people who get caught up playing the waiting game, look at all the folks who didn't jump on the 58XX's and decided to wait for the competition?
Did that even have any effect on product uptake? No, the GPU's were supply constrained anyway due to manufacturing capacity and also very popular, hence why retailers were able to obscenely price gouge.

It will be a similar story this time around, some people will want to 'wait it out' for 6 months or more, but most will get far to excited over the new tech and won't be able to help themselves...

Yea, I'm aware that this is a simliar to the last time, but to be honest, the biggest fail part on the late launch of the Fermi wasn't so much on the late arrival, but more on it being too expensive comparing to the AMD/ATI alternative when it eventually arrived.

Based on all the rumours so far, the launch of next gen Nvidia would most likely arrive later than the next AMD, but the question is...what would the price be like on their cards. In the pass, Nvidia has always charge a "premium" because of its cards, but in this gen they have pretty much bring their pricing down to along side with AMD's. So it would be interesting to know if they can actually deliver better bang for bucks, like back in the GeForce 8 vs HD2000 series days. Also let's not forget that we only got rumors on the launch days at the moment...for all we know the AMD next gen launch date can get pushed back as well...I mean just look AMD's next gen CPU that's supposed to have arrived around the same time as SandyBridge, and up to now, we still don't have a confirmed launch date due to delay after delay...

RobotGirl · 21 Jul 2011 at 11:36

WingZero30 said:
How far can GPUs get in die shrinkage?

Who knows? as long as the atoms have enough space to operate in the way they’re supposed to ,some people around here ought to know though . Them we have quantum computing on the horizon too ,when we get to that stage size will not really matter any more ,its all down to quantum superpositions . :rolleyes: