http://www.abload.de/img/map_c_1120x941d2o94.png
http://ascii.jp/elem/000/000/658/658986/index-3.html
Printable View
for the ASCII graphs: Asynchrone shaders clock ? 1200/3000? look a bit too fast for hot clock, GK110: 20SM x 32SP - 4TMU/SM ( GK104 is the GTX 560TI 448SP with faster clock, 14 SM or the GTX470 ).
Well at least there's new numbers to put in the rumors mill.
yeahQuote:
Well at least there's new numbers to put in the rumors mill.
I have never seen information that was so contradictory. Hotclock, no hotclock...GK110= single chip, GK110=2xGK104...384bit, 512bit
Maybe Nvidia will surprise us all?
Edit:
The memory bus data of Fermi is messed up bad...
So no 512 bit memory bus? If NVIDIA decides to keep using the 384 bit memory bus, I really hope they up the VRAM to 3GB.
I have read this time the traduction on AscII, this is even worse we can imagine. I think they mostly make a compilation of rumors, with some we have allready heard for GCN ( XDR2 )....
http://translate.google.com/translat...2Findex-3.html
I think there's something we dont seen when rumors come from Chinese, Taiwanese or Japanese site. There's a lot of "company", "well informed source (lol ) ", who play to launch a lot of rumors. There's a big numbers of company involved in the production on electronic components, a lot of little brand who sell offtly one type of gpu, a lot of brand we will never seen outside the Asia market.
Jensen on Kepler
In the last year, if not even longer, Nvidia has talked more about Tegra than any of their other products. This is despite the fact that anyone who has seen at their balance sheets knows that they make money mostly on Geforce and Quadro products.
After the keynote, we managed to ask Jen Hsun Huang, Nvidia CEO, a single question. We asked, where is Kepler? The answer we got was that we have to be "patient" about it.
http://www.abload.de/img/jensen_press_ces2012kmwco.jpg
Furthermore, we discovered that there was an internal discussion to showcase Kepler at this keynote but Nvidia decided not to. As far as we know, it was a business decision to wait, and Kepler is production scheduled to launch at a later date. Current estimates are that it will launch in first half of 2012, possibly sometime late in Q1.
Written by Fuad Abazovic :D
I like how "Fatal1ty" is in this pic with Jen. In fact before reading your post I thought it was about him. Barely noticed Huang..
oh lol, didn't know Jonathan Wendel? was still around. hahah
Q3 ex-master on top :)
adrenaline Site Talking about the performance of GTX 660 (GK104) card It will come faster than GTX 580 (GF110) 15% , and this card GTX 660 will cost only 399$ ! If we consider that this is true This means that this card GTX 660 (400$) will come with the same performance of HD 7970 (550$) card Or slightly lower( between the Radeons HD 7970 and 7950). Because HD 7970 Faster than GTX 580 15-20% and GTX 660 faster than GTX 580 15% This gives us the result that GTX 660 will be the same performance of HD 7970 or slightly lower.
Of course, sudden major if this card is priced 399$. Mathematically this is mean HD 7970 cost you 550$ and GTX 660 cost you 399$ Percentage GTX 660 (399$) Cheaper than HD 7970 (550$): 399$-550$= 150$ (37.5%) and The difference between the performance of the cards is : Either the card will be the same performance HD 7970 or slightly less. This means that GTX 660 card is the best solution.
Of course, before anyone attacks me :D This is the analysis of the rumor that we talk about it... Who knows maybe this rumor be true may be incorrect
Its happened before!
Why not? If 660 is around 7950, and 670 slightly above 7970 and 680 above that it wouldn't be to different from earlier series. nVidia might be aiming higher than AMD, the later release suggests that.
Besides, I don't think 15% higher performance would put it that close to 7970. 7970 is around 35% faster than GTX 580 at high resolutions. It would still be closer to GTX 580 than 7970. So it would mirror the situation in the last series pretty good.
These rumors actually come from OBR (here) so I doubt that they are true
15% - are you sure?
e.g. http://www.computerbase.de/artikel/g...stung_mit_aaaf (14 games/2600k@4.5ghz)
1920x1080 4xAA/16xAF +20%
2560x1600 4xAA/16xAF +30%
1920x1080 8xAA/16xAF +29%
2560x1600 8xAA/16xAF +33%
1920x1080 no AA/AF +18%
2560x1600 no AA/AF +25%
I’m talking about performance, that's best measured with no bottlenecks at higher resolutions. In those cases 7970 is at least 35% faster. Of course you won’t see such results with bottlenecks in the mix.
http://www.sweclockers.com/image/dia...1caa4640d34453
http://www.sweclockers.com/recension...deon-hd-7970/1
Review is in Swedish but I hope you can read graphs, the first image is an Index over the average performance, in individual games it’s sometimes faster and sometimes slower.
NVIDIA would have to be out of their minds to price a card that outperformed the GTX 580 at a level that UNDERCUTS their own product. The entire point of price structuring through old / new lineups is to effectively clear out stock of their existing products while still selling a new architecture.
Very simply and as illustrated well by AMD: if a graphics card outperforms a previous generation part that's still being sold, expect it to be priced at a higher level.
Seems like people who have no clue about the GPU market are talking out of their rear ends again in an effort to generate traffic.
I make a recapitulation, will be more easy.
5th jan: http://www.fudzilla.com/home/item/25...amds-28nm-lead
11th : http://www.fudzilla.com/component/k2...nsen-on-keplerQuote:
Meanwhile AMD’s 28nm lineup looks very good and despite silent promises made by Nvidia to its partners that “Kepler is much better than that“, they are expressing serious concerns about AMD’s lead, or rather they how to go compete against AMD until Kepler launches. Of course, at this point nobody knows what to expect from Kepler in terms of performance and all we have to go on are Nvidia’s own rosy predictions.
A few partners also confirm that they have heard talk of a Q1 launch for Kepler, but there is no better data than that. A few other independent sources expect to see the cards in Q1 2012 but then to actually start selling the parts in volume in Q2 2012. In any case, AMD will launch much of its 28nm lineup by then and it will open a significant lead with the HD 7970.
We will try to learn more next week at CES, probably some huge announcements on Monday, January 9th 2012 so stay tuned.
11th after confirmation by AMD of the 7950 launch the 31th Jan. http://www.fudzilla.com/graphics/ite...n-january-31stQuote:
After the keynote, we managed to ask Jen Hsun Huang, Nvidia CEO, a single question. We asked, where is Kepler? The answer we got was that we have to be "patient" about it.
Quote:
We managed to overhear that AMD's Radeon HD 7950 will officially launch on January 31st. The card should be available as of that week, while the rest of the lineup should follow in February.
In case you somehow missed it, the Radeon HD 7950 features 1792 stream processors, or 28 GCN compute units, and a 384-bit memory interface paired up with 3GB of GDDR5 memory. The reason behind the January 31st launch date might have something to do with the Nvidia's plan to launch its 28nm Kepler based GK104 GPU sometime around that date and AMD simply wants something to coincide with Nvidia's launch, or ruin its day.