Printable View
Doesn't TMSC have multiple lines for their 28nm process?Quote:
Originally Posted by DarthShader
Given the history of his anti-NV ramblings without proof, I take this with a saltshaker. I think he is speculating more often than not without real tangible info.
Indeed.... :(.Quote:
Originally Posted by spicypixel
That's short-term thinking though. If that trend were to continue and AMD got out of the high end segment, in the long run it would be bad for consumers because we'd be stuck with high-priced cards with mediocre improvements each generation.Quote:
Originally Posted by tajoh111
For most games that might be true but some of the newer games can easily use more than 1.5GB. I saw 2.2-2.4GB of vram usage playing Skyrim with the HD texture pack and 8xAA.Quote:
Originally Posted by dartaz
Too bad really if thats the case.Quote:
Originally Posted by Macadamia
I'm starting to think that Huang's "disappointment" for AMD's new cards might not be much about performance.
It could easily be about stocks/volumes knowing AMD came first with their next gens., but who knows for sure.
And that thing about being patience w/ Kepler, that could mean nvidia has difficulties too(stocks too?) - maybe... 28nm and whatnot.
nVidia being quite about Kepler is one thing, but low prices might not be one of them...
Yup. Check out Newegg. ONE type of HD 7970 in stock. The situation is no better @ NCIX, Amazon or Tigerdirect.Quote:
Originally Posted by DarthShader
Something is obviously going on there.
Perhaps this is part of the reason why GK110 is pushed back so far?
I seriously doubt this is only because of NVidia though. TSMC may love NVidia, but no way they'd completely freeze their 28nm process ONLY because of them if they had plenty of business from AMD... Although, they may still be mad that AMD tried to only do cayman on 32nm (which is why TSMC cancelled 32nm altogether).
If it was having major yield problems though, maybe that's part of the reason AMD were pricing so high.
Big boy chip @28nm now would be masochism I guess.
http://www.geforce.com/News/articles...dia-kepler-gpu
Quote:
Created in a dark, futuristic setting, the demo utilized a host of advanced rendering techniques that smoothly tessellated and morphed facial features, created realistic street scenes using point light reflections, and replicated the work of the best movie directors through the use of fine-tuned out of focus bokeh filters. The only downside was that it took three GeForce GTX 580s to run the demo in real-time.
Quote:
Today GDC 2012 is upon us, and once again Epic has shown the Samaritan demo, but this time with a twist - instead of three GeForce GTX 580s, the demo was shown running on a single next generation NVIDIA graphics card.
Looks like the March 12th date just got some more substance. It claims MSAA on the 580's vs FXAA on Kepler.Quote:
If you’d like to learn more about Samaritan, Unreal Engine 3, or any other aspect of the demo, check out our in-depth Samaritan deep dive from May 2011.
For more info on our next-generation Kepler graphics card, stay tuned to GeForce.com.
Speaking of the "Samaritan" demo, Rein noted that when they showed it off last year, it took three Nvidia cards and a massive power supply to run. However, they showed it off the demo again that ran using a new, not yet released Nvidia card and one 200 watt power supply.
http://www.gamesindustry.biz/article...tan-into-flash
I see Epic still touch money from Nvidia for marketing purpose... i hope for them the samaritan demo have been optimised since last year.
since we dont know the details of a TSMC factory stall, I look at the bright side and assume things will be better once they start going again. you could imagine something went wrong and needed fixing, but I'm imagining them just improving things. we may never know
GeForce GTX 680 Features Speed Boost, Arrives This Month, etc., etc. - TPU
Quote:
Third, the many-fold increase in CUDA cores doesn't necessarily amount to a linear increase in performance, when compared to the previous generation. The GeForce GTX 680 is about 10% faster than Radeon HD 7970, in Battlefield 3. In the same comparison, the GTX 680 is slower than HD 7970 at 3DMark 11.
Fourth, the NVIDIA GeForce GTX 680 will very much launch in this month. It won't exactly be a paper-launch, small quantities will be available for purchase, and only through select AIC partners. Quantities will pick up in later months.
At least here in Germany we don't have this situation. 14 Different 7970 in stock and half of them in stock nearly everywhere. The availability is practically perfect here and prices are beginning at 510$ (without VAT), which is ok. 7950 and the 77xx cards are also in stock in most places.Quote:
Originally Posted by SKYMTL
But still i can't believe that TSMC stopped their 28nm process since 3 weeks. No asian page notices that, not even Digitimes which are normally pretty good informed in case of TSMC. TSMC makes no press release on something that everyone will see in the Quarterly results, but which could make shareholders pretty angry if they'll only hear about it at the Earnings call.
Pretty much what I expected. GK110 is nowhere necessary so they are keeping that one under wraps until AMD steps up with something else, and hey presto, nVidia is ahead again. Dual GK104 coming in May is surprising as that's pretty much around the corner. That performance should last them for some time.
Also, it seems that there is massive potential if Epic's Samaritan demo required 3 580s, and now only one Kepler.
But 2,304 CUDA cores for GK110?! Wow, that's a monster.
Isnt 680 gk110 anyway? Either way it'll be a hot big beasty card! Same as last time really the faster card but over priced ah well.Quote:
Originally Posted by Tim
That demo seems off, think, a massive system with 3x 580's thats what 1000w plus when you add in cpu etc...
Now a one card system an the whole thing is on a 200w psu? Even sandy oc'd is gonna pull half that so 100w is low/mid territory and i dont think thats beating 3 580's?
Just fishy really .
GTX680=GK104
The demo ran on a 2-way SLI setup last year, not 3-way. The third card was used for GPU-PhysX. This time, FXAA was used (instead of MSAA?). Also no info on fps. And a year of time to do optimizations. And GK104 has double the computing power (not performance!) of a GTX580. So it's plausible, you just have to think critical.
10% faster than Radeon HD 7970 in Battlefield 3 sounds bad. Is one of the games where 7970s performs worse compared to last generations, so the GTX 680 would be very close or below in other games.
It performs as expected or better for a smaller 256-bit 2GB card. At the moment it looks like the card will end up between 7950 and 7970, and given it has only 2 GB VRAM they should price it closer to 7950. Not that I think they will, most likely if it can beat 7970 in a few games it'll be $499 minimum simply because Nvidia likes their premium in the high end. And GK110 won't matter as long as it's not ready.
For be honest, i will not worry too much about the 7950... a 7870 pre oc to 1100mhz should close the 7950 ( the 7870 is what.. 10-15% slower of the 7950.. )... i suspect, the 7950 will not really be a card who will stay in memory of best sellers... ( i speak about prices, not performance wise against the gk104. )Quote:
Originally Posted by Pantsu
How can you believe thisQuote:
Originally Posted by Handrox
@Damien I totally agree with you
http://weibo.com/fengke117Quote:
Originally Posted by Googlish
GTX670Ti
http://www.abload.de/img/gtx680_mfa_keplerx7uud.jpg
http://www.redquasar.com/forum.php?m...7234&fromuid=1
It seems tpu have not completely translate in their news the post given on Heise.de .. http://translate.google.com/translat...z-1465891.htmlQuote:
Originally Posted by Olivon
That's a fake photo from 570. It's been photoshoped and the job was done quite badly.Quote:
Originally Posted by kaktus1907
Is it only me that have the feeling that Nvidia delayed GK110 mostly because they saw GK104 would be enough to combat this series of AMD GPUs and will bring out GK110 to combat AMDs next refresh which would be a big win from Nvidias side (saves Nvidia plenty of time to better be able to focus on that next generation after Kepler) as there's no need for a "Kepler" refresh chip as GK110 will handle the next AMD update.
IMO if this is true would be that BIG WINNER what Charlie wrote about earlier haha.
Any source on that? Because I just tried Google and no luck. Everybody mentions Tri-SLI with PhysX, but no word on one card being used purely for PhysX.Quote:
Originally Posted by boxleitnerb