NVIDIA reveals Fermi's successor: Kepler at 28nm in 2011, Maxwell in 2013

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Sylvanas

Diamond Member
Jan 20, 2004
3,752
0
0
Lets please not muddy up the waters around Nvidia gpus by giving names to cards which have never been mentioned by any reputable source.

There is no 455, no 460x2, no 475 and definetly no 485. The 480 was ment as the 512 part. All the other parts are cutdowns from this piece.

A fully unlocked 460 will not equal the current 480. That would be dreaming. It could however put up a fight against the 5870 if factory overclocked to what most good factory oc cards come as now, minimum 800mhz on the core.

If Nvidia does come with a refresh sometimes next year, id imagine a late spring release, much like the gtx480 or a summer realease like gtx460. ANd yes, they will probably then call the 512 part a gtx485. But nothing like that exists anywhere yet.

There was this in addition to random PCB shots and specs arriving every now and then. It depends how Nvidia wants to play it with a 512 Fermi. It's relatively easy to just harvest good 512 yielding cores and sell them at a high price with perhaps slightly higher clocks, it's much more of an investment to do a 'revB' core much like Intels Q6600 G0 stepping CPU that lowered temps/power consumption with higher clocks with a minor redesign.

I'm sure we will know more come launch day for AMD's 6xxx series.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Is it just me, or does being able to accurately plot the expected performance of a chip two full generations away from production on a node that doesn't yet exist seem a bit... optimistic?

Key word being "expected". And the answer would yes, its always optimistic in that it generally assumes some best-case scenarios regarding risk-mitigation in terms of the error-bars on the timeline as well as the error-bars on the performance.

The performance and timeline expectations are a required component to securing authorization for the project to begin with in the upper executive circles and BoD.

If you can't project a performance ROI to come from the proposed R&D budget then you can't evaluate metrics of expenditure efficiency, corporate benchmarking, etc. In other words corporate America has been operating this way since the days of the railroads.

What is atypical for corporate America in general is to make such internal projections and expectations public. But in this specific industry both Intel and AMD do this in the CPU world. So it is only natural for Nvidia to gravitate towards adopting a similar investor relations model of publicizing their roadmappery.
 
Last edited:

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
Is it just me, or does being able to accurately plot the expected performance of a chip two full generations away from production on a node that doesn't yet exist seem a bit... optimistic?

Well considering that they have been making graphics chips for 15 years, and they've released what a dozen or so different chips, no it's not optimistic.
 

WelshBloke

Lifer
Jan 12, 2005
31,408
9,300
136
I think some of us also have a clear understanding of how the AMD/ATI PR machine works. It seems they have some PR people posting on tech forums to influence opinion.

Theres certainly a lot of newly registered posters suddenly being very vocal on this board from both teams, so you may be half right.

This post is a thread-crap and thread-derail. Please refrain from posting this manner of commentary in threads for which the topic is not about the perception of shadowy inner-workings of marketing departments.

Moderator Idontcare
 
Last edited by a moderator:

ShadowOfMyself

Diamond Member
Jun 22, 2006
4,227
2
0
So if Kepler is the "Tock" due out in 2011, then Fermi II must be the "Tick". The reason I think folks don't expect to see anything "so soon" from Nvidia is because Fermi was very late, making it seem like too short a period of time for the next gen/refresh release. For all intents and purposes, Fermi "should" have been here just about the time when AMD's 5870 launched in 2009. That was a year ago. Fermi has actually been here for half that length, so another architecture being released would "seem" like it's not possible or too soon. And of course, according to the logic of some, if Fermi was late, then every succeeding GPU from Nvidia must be late also. I don't think that's how things are going to work out.

Finally something I agree with it

Best example here would be the X1800... In fact, Fermi looks A LOT like X1800XT - big, hot, late, but still outperforms the competiton... and then just a couple months later, we got a surprise in the form of X1900XT, which actually came earlier than the 7900 series, so whos to say Nvidia wont do the same?
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
When I talked to my source at AMD he said " Were gonna kick nvidia to the curb just like Apple did " so does that make you feel better about the 6xxx series hypothetical performance now :)

What I meant is that if it was a press-release as others claim then why isn't it all over the web....We all know how the Big Green Machines PR department works.

I think some of us also have a clear understanding of how the AMD/ATI PR machine works. It seems they have some PR people posting on tech forums to influence opinion.

Yeah but they don't get free hardware to do it I suppose,if they did....I'd be volunteering:p
Cause we all know that makes no difference in opinions here:whiste:

Why should AMD give out free hardware for folks to pimp their products and turn most of these threads into anti-NV fests? You and 4-5 others seem happy to do it for free ;)

Why should AMD give out free hardware for folks to pimp their products and turn most of these threads into anti-NV fests? Nvidias failures and sleazy marketing seems to result in the same, for free ;)



Fixed that for you :)

Theres certainly a lot of newly registered posters suddenly being very vocal on this board from both teams, so you may be half right.

Knock it off. All of you. This is completely out of bounds for the thread.

Each of you are getting a thread-crap warning. Whether you started the thread-crap or perpetuated it by responding to it, in both cases it is derailing the thread all the same.

The proper response to a post that is a thread-crap is to report the post. Leave the moderating to the moderators, don't make a public rebuttal and risk incurring an infraction for yourself.

Further thread-craps in this thread will be met with infractions. Keep it on topic, stop the thread-crapping and flame-baiting.

Moderator Idontcare
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
Edit: sorry mod, I posted the above right as you were posting.. I will delete my response to the thread derailer. :)
 
Last edited:

Leyawiin

Diamond Member
Nov 11, 2008
3,204
52
91
I'm really excited about this. Looks like we might have something akin to the old 8800 GTX jump when Maxwell is released in 2012. If AMD has a counter we'll have a very good time to look forward to (and two years goes by quickly). Start saving my pennies for a new rig to complement it.
 

GaiaHunter

Diamond Member
Jul 13, 2008
3,650
218
106
Finally something I agree with it

Best example here would be the X1800... In fact, Fermi looks A LOT like X1800XT - big, hot, late, but still outperforms the competiton... and then just a couple months later, we got a surprise in the form of X1900XT, which actually came earlier than the 7900 series, so whos to say Nvidia wont do the same?

The X1900 was bigger than the X1800.

The G70 at 110nm was 334mm^2. The G71 at 90nm was 196mm^2. The R580 at 90nm was 353mm^2 and the R520 was 288mm².

With 7900 series NVIDIA used 90nm and went for tiny chips.

Maybe NVIDIA can do a refresh, but it seems more likely going from GF104 than from GF100.
 
Last edited:

Kuzi

Senior member
Sep 16, 2007
572
0
0
The reason I think folks don't expect to see anything "so soon" from Nvidia is because Fermi was very late, making it seem like too short a period of time for the next gen/refresh release. For all intents and purposes, Fermi "should" have been here just about the time when AMD's 5870 launched in 2009. That was a year ago.

The reason why NV won't/can't release any higher end refresh soon is because they already maxed out the die size @ 40nm. I mean they could still produce something larger, but do you think they would risk doing it? Especially after Fermi faced serious yield/heat issues for a long time before release. The best I see NV doing in the next 12 months is release tweaked/higher clocked versions of GF100/GF104, and maybe a dual GPU card based on the GTX460, which I think would be pretty good.

Cypress on the other hand was small compared to Fermi, so ATI had room to add more transistors at 40nm and produce a more powerful but still efficient chip. The next 12 months should be interesting and I hope NV builds a more balanced GPU with Kepler instead of a huge power hungry die.
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
The reason why NV won't/can't release any higher end refresh soon is because they already maxed out the die size @ 40nm. I mean they could still produce something larger, but do you think they would risk doing it? Especially after Fermi faced serious yield/heat issues for a long time before release. The best I see NV doing in the next 12 months is release tweaked/higher clocked versions of GF100/GF104, and maybe a dual GPU card based on the GTX460, which I think would be pretty good.

Cypress on the other hand was small compared to Fermi, so ATI had room to add more transistors at 40nm and produce a more powerful but still efficient chip. The next 12 months should be interesting and I hope NV builds a more balanced GPU with Kepler instead of a huge power hungry die.

The challenges are really the same for both... They're both working on getting the best performance out of the chip, just going about it in a different way.

AMD went with a less complex chip with Cypress that allowed them to go smaller and more efficient, so their area for opportunity is in increasing size/complexity while keeping efficiency. NVIDIA went with a more complex, larger chip, so their area of opportunity is in boosting performance through increasing efficiency. I don't see any reason why either strategy isn't viable.

Granted, time is also a factor here, and I think it is questionable as to whether or not NVIDIA could pull off a more efficient "Fermi II" in a reasonable time frame to make it worthwhile. In Q1 of 2010 I honestly believed that GTX 470/480 were going to be pretty short lived, and that a replacement had already been in development. This would have been consistent with what we've seen in similar cases, such as FX 5800 getting replaced by 5900 and X1800 getting replaced by X1900 very quickly. This hasn't happened yet, and I don't see any indication from NV that it will.

edit: I don't see any reason why they would release a "GTX 460 GX2" type card. Wouldn't a 5970 already out perform it? Perhaps an OEM (like Asus) may put out a card on their own, but I don't see the sense in NVIDIA launching something targeting the enthusiast market that isn't really an enthusiast part.
 
Last edited:

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
edit: I don't see any reason why they would release a "GTX 460 GX2" type card. Wouldn't a 5970 already out perform it? Perhaps an OEM (like Asus) may put out a card on their own, but I don't see the sense in NVIDIA launching something targeting the enthusiast market that isn't really an enthusiast part.

Not necessarily, and not in every instance -- due to better scaling and performance of NV's drivers. A full bodied 460 with all shader clusters enabled in dual GPU config could handily outperform the 5970 on nv-friendly titles and give it a good run for the money on the others.

And by now NV should have harvested plenty of GPUs capable of full-bodied operation. It makes no sense to release a 106-based '460' part because it'd eat the 470 market completely and even take a bite out of the 480 market. However, for a dual-gpu halo part the game changes.
 

Kuzi

Senior member
Sep 16, 2007
572
0
0
edit: I don't see any reason why they would release a "GTX 460 GX2" type card. Wouldn't a 5970 already out perform it? Perhaps an OEM (like Asus) may put out a card on their own, but I don't see the sense in NVIDIA launching something targeting the enthusiast market that isn't really an enthusiast part.

The xbitlabs review shows two GTX460 in SLI being even or outperforming the 5970 in most situations:

http://www.xbitlabs.com/articles/video/display/gf-gtx460-sli_6.html#sect1

A Dual GPU card based on the GTX460 might be slightly down-clocked, but it should still hold it's own against the 5970. You also have to keep in mind the price factor too, the 5970 would probably cost much more.
 

WelshBloke

Lifer
Jan 12, 2005
31,408
9,300
136
While it would certainly be fantastic if NV release a new chip soon, I'm not too amazed by that chart.

Its performance/watt and in DP which is not too relevant to gaming. A big bit of the improvement there could be had from keeping the same overall performace and going to the next manufacturing node (not that I dont think there will be performance increases).
 

Flipped Gazelle

Diamond Member
Sep 5, 2004
6,666
3
81
The xbitlabs review shows two GTX460 in SLI being even or outperforming the 5970 in most situations:

http://www.xbitlabs.com/articles/video/display/gf-gtx460-sli_6.html#sect1

A Dual GPU card based on the GTX460 might be slightly down-clocked, but it should still hold it's own against the 5970. You also have to keep in mind the price factor too, the 5970 would probably cost much more.

I'm sure that if Nvidia released a dual-GPU, AMD would cut the price of the 5970 to be competitive.
 

Stoneburner

Diamond Member
May 29, 2003
3,491
0
76
according to Ryan Smith of Anandtech, we are looking at late 2011 for kepler.

Anandtech? Bah! Who cares what that haven for rumormonging scoundrels has to say! We need a credible source like Charlie!

IN all seriousness, it seems to be Nvidia needs to be worried about ATI's tactic which is cost effectiveness. Nvidia needs to learn having the biggest gun around doesn't mean crap if you cant' draw it fast enough. We are getting Radeon 6000 details in a month with no information on nvidia's short term response.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
So if Kepler is the "Tock" due out in 2011, then Fermi II must be the "Tick". The reason I think folks don't expect to see anything "so soon" from Nvidia is because Fermi was very late, making it seem like too short a period of time for the next gen/refresh release. For all intents and purposes, Fermi "should" have been here just about the time when AMD's 5870 launched in 2009. That was a year ago. Fermi has actually been here for half that length, so another architecture being released would "seem" like it's not possible or too soon. And of course, according to the logic of some, if Fermi was late, then every succeeding GPU from Nvidia must be late also. I don't think that's how things are going to work out.

didn't they end up pulling some engineers out of kepler to push fermi through more quickly? depending on whom/how many they pulled that could have some impact on kepler's timing if nothing else, though I agree that as a separate arch with separate engineering team there's typically no reason for the timing of them to be related at all. heck, theoretically you could have both teams finish at or around the same time if one is late and the other early. I'd hate to be on the "late" team if that happened, however.
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
I'm sure that if Nvidia released a dual-GPU, AMD would cut the price of the 5970 to be competitive.

Yeah. Plus with the impending Cayman/Antilles launch, it just seems a little late. Maybe had NVIDIA launched a dual GF104 gpu card simultaneously with the GTX 460, but I think it's just too late now. Of course we'll see, but I personally don't think it will happen.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
I agree with the Kepler card not being destined to be late as it is their next new 'series' card and there is no reason for Fermi's issues to infect their next architecture.

But for the here and now they have Fermi to work with. With it coming late due to whatever issues they had, amongst them clearly being heat and power consumption, this will affect whatever refresh plans they had for Fermi. I doubt we will ever see anything more of GTX 480 with a 512 shader part.

They may release a fully unlocked GF104, but that won't even be as fast as the 480 is now, they may release a dual gpu card, but that won't be as fast as whatever AMD is going to release as the 6990/6970 or whatever they end up calling their new dual-gpu card.

This announcement confirmed that there is no new architecture or series coming from nvidia until September or later next year. So that leaves us with refreshes of Fermi. I'm sure the refreshes will offer something more than the performance NV has now, I would be surprised to see a single GPU that is faster than a GTX 480 though, probably going to be a multi-gpu card.

I've never seen a refresh of a current series compete with a brand new series. It would be akin to the GTX 285 being faster than a 5870, it's not going to happen, although the GTX 295 was faster than a 5870 at launch. If they drop a multi-gpu card, it will likely be faster than a 6870, and will probably be a good buy because they'll have to price it fairly low to stay under wherever AMD is pricing their new dual-gpu flagship and not too far from wherever AMD's single-gpu flagship is selling at.

I think this will be their problem in the present, they can't offer an exciting card using refreshes against AMD who is going to be selling a card on a new architecture. They'll probably gain traction from selling cards on their cheap price points, not dominant performance.

In short, yeah they'll definitely release new cards between now and 2H 2011, but they're not going to be the sort of cards you can get excited about and look forward to seeing benches of. I've seen a ton of refresh parts, I know what to expect, 5 more fps.

bold prediction: nvidia is going to copy amd's small ball strategy asap, at least for us gamers. amd was simply better prepared with 5xxx, but with 6xxx coming out now the only reason nvidia can't counter is that their gpu is already too big and 460 was almost an afterthought. the "performance per watt" comments were very astute imho. nvidia has a lot more resources to throw at this than amd, I wonder how long it will take them to catch up and surpass amd using amd's own game? kepler might do it, but watch out for maxwell. I think that we'll find that nvidia will be able to get more performance/mm^2 than they thought and will end up with a small AND faster gpu.


The reason why NV won't/can't release any higher end refresh soon is because they already maxed out the die size @ 40nm. I mean they could still produce something larger, but do you think they would risk doing it? Especially after Fermi faced serious yield/heat issues for a long time before release. The best I see NV doing in the next 12 months is release tweaked/higher clocked versions of GF100/GF104, and maybe a dual GPU card based on the GTX460, which I think would be pretty good.

Cypress on the other hand was small compared to Fermi, so ATI had room to add more transistors at 40nm and produce a more powerful but still efficient chip. The next 12 months should be interesting and I hope NV builds a more balanced GPU with Kepler instead of a huge power hungry die.

they have plenty of time to roll out a 50% larger gtx 460. right now they're using 7/8 of a "full fledged gf 104", maybe the could go to ~ 500 mm^2 and use 11/12 of it instead. they have 12-15 months to kill, plus they've already had 3 or so months to start on it. all 6xxx cards will be on the table, too. how fast would a 528sp gtx 460+++ be? I'll bet it could at least go toe to toe with cayman xt.
 
Last edited:

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
I'm sure that if Nvidia released a dual-GPU, AMD would cut the price of the 5970 to be competitive.

I thought the problem with a dual GF104 part is that they would have to slow it down to keep it's power use down so it might not be faster than the 5970. Of course they could promote it how AMD promotes the 5970, as if it's built to be overclocked.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
I thought the problem with a dual GF104 part is that they would have to slow it down to keep it's power use down so it might not be faster than the 5970. Of course they could promote it how AMD promotes the 5970, as if it's built to be overclocked.

Single GTX460 1GB max at 160W. Two of these on one PCB with 2GB does not double the consumption. (320W).

GTX275's (219W) wattage did not double for GTX295. That would have been 438W. It could easily be made under 300W spec. At least the 336sp versions of the 460's. And that would have competed well with a 5970 according to Xbit's article linked above.

I think we've (not you and I necessarily) had this discussion before.

Must be some other reasons for no dual 460 card.
 
Last edited:

Creig

Diamond Member
Oct 9, 1999
5,170
13
81
What is atypical for corporate America in general is to make such internal projections and expectations public. But in this specific industry both Intel and AMD do this in the CPU world. So it is only natural for Nvidia to gravitate towards adopting a similar investor relations model of publicizing their roadmappery.
That was the general gist of my comment. There are just too many variables involved in attempting to make predictions that far out. The amount of salt required to accept that graph at face value would kill an elephant.