NVIDIA Kepler speculation thread

@Ailuros - I went back to the original article, and it does indeed talk about "power plans" (whatever that is), not planes - so much for my reading comprehension skills.
AFAIK, the original source of information was Theo Valich's piece over at VR-Zone, all others are referencing him. And he was writing power planes. With an "e". If that makes it more believable is another question as the earlier news on heise.de he linked indeed mentioned the dynamic clock adjustment, but first on a far smaller scale (+7% max) and secondly heise.de didn't talk about power plan(e)s at all.
 
If it is power "plan" then this might be referring to closed loop voltage + freq control of some kind (mobile SOC's do this a lot).

If it is power planes, then I guess that word is loosely used. Partitions are routinely power gated, but number of power rails/planes itself is going to be a very small number.
 
unless google tricks me, PHK has the card with press kit and return from US.. also says it's 195W and faster than 7970..

145551v2zl5zctmcchttom20w6.jpg


 
Only a summary of recent information/rumors.

btw.
According to PHK GTX 680(?) could be ~500 points faster than HD 7970 in 3DM11 X-score.
 
btw.
According to PHK GTX 680(?) could be ~500 points faster than HD 7970 in 3DM11 X-score.

also PHK expects a performance driver by AMD in reference to Pitcairn..

new bits
PHK said:
compared to 7970 for the advantages of the new stuff is as follows: low-voltage, high frequency (to breaking GHz), low power, low noise, high-performance (DX11 ), AA, and get the support of the business of the game.

 
Only a summary of recent information/rumors.

btw.
According to PHK GTX 680(?) could be ~500 points faster than HD 7970 in 3DM11 X-score.

It seems previous performance rumours talked about 670ti then...

If this is with 195W... this would be the most efficient chip ever...
 
So ... they basically set one overall power budget for the "core" and the shaders - and can seperately send either one of them into "turbo-mode" when encountering asymmetrical shader/core loads?

I guess those Bitcoin miners just placed some preorders ;)
 
Seems to me that those who didn't buy 7970 because it's "not faster enough" won't be buying this either.
 
Seems to me that those who didn't buy 7970 because it's "not faster enough" won't be buying this either.

15% faster -by 3d mark-, smaller, cooler and cheaper -at least in performance per buck will be surely-?. Enough reassons to buy above a 7970 imo.
 
Seems to me that those who didn't buy 7970 because it's "not faster enough" won't be buying this either.
Pricing is the big question:
$499+ - sub-performing upgrade offer to GTX 580
~$299 - a bit over-priced upgrade offer to GTX 560 Ti
 
There is no way this thing will be priced at anything below HD7970. It's just common sense.

So we have two disappointed groups here: those who think it will be $299 and those who think this thing is performance monster.
 
GTX 560 Ti was $249 and also a bit higher priced than previous G...4 GPUs.
50-80% more performance would be a nice upgrade offer to GF104/114 users, even at 30% higher price than they used to pay for this cards.

There is no way this thing will be priced at anything below HD7970. It's just common sense.
Its a 256-Bit 2GB performance card with only ~300mm² chip.
 
There is no way this thing will be priced at anything below HD7970. It's just common sense.
That assumes it's actually at least as fast, if not faster. I have my doubts about that personally. It takes a miracle to get that huge performance increase with significantly smaller die having much less bandwidth and using less power. It is possible they have pulled a rabbit out of their hat but it's rather unlikely.
 
Back
Top