Originally Posted by bojinglebells
Originally Posted by 2010rig
LOL at AMD fans who refused to accept that GK104 was a mid-range part, and claimed GK104 was NVIDIA's "highend part".
and unless this card can make it out before mid march it will have been a full year with GK104 as nVidia's "highend part". From nVidia's past standards, yes, GK104 is "mid-range" in size, however that doesn't mean it wasn't their high-end part
Not from the PoV of the consumer, no, but it was always their mid-range part and always will be.
Originally Posted by juano
Originally Posted by Brutuz
but AMD has the performance/mm² advantage
I don't beleive that is correct when considering how close the GK104 and tahiti perform relative to each other and that the GK104 is smaller.
Compare them at the same clock speed, nVidia's cards tend to be at much higher clock speeds than AMDs (At least until fairly recently) thanks to GPU Boost. (AMDs fastest version leaves them with a max clock speed of 1050Mhz, where it beats a 1200Mhz GTX 680)Look at this
, it's only conjecture, but with a Core Clock of 1210Mhz and a boost clock of ~1300Mhz, it isn't that much faster than a stock 925Mhz HD7970 running Catalyst 12.3/12.4, compare it to a 1200Mhz HD7970 running 13.1 and you'll see what I mean...There's also the additional fact that the GTX 680 is more ROP limited (If not that, something else in that area) than Core/Shader limited too, AMD would gain more from running an OC than nVidia.
Originally Posted by TheBlademaster01
He's correct though, GeForce and Quadro have better support in the majority of software in GNU/Linux.
As someone who goes back and forth from nVidia and AMD fairly often, their drivers are pretty equal...You rarely get show-stopping bugs, but still get a few here and there (eg. nVidia drivers crashing, AMDs stuttering and MMU issues) unless you're talking Linux, even then though AMD drivers are often said to be bad when in reality, they're just acceptable with an annoying level of support to deal with too. (eg. My laptop being limited to Catalyst 12.6 and thus xorg 11.1)
Originally Posted by Rayce185
Originally Posted by Arni90
The source says 732MHz, interestingly the clocks and functional units are completely identical to the K20X
If you look at former Tesla cards, they are generally clocked lower than their Geforce counterparts. I'm guessing the GK110 Geforce will be in the mid 900's, as the GK104's are.
This big of a GPU clocked that high at stock? I expect we'll see it in the high 800s as a stock speed at best.
Originally Posted by Rubers
Originally Posted by Rayleyne
So let me get this straight, It's going to consume a ton of power (Don't care) Still lack compute capabilities and... cost 899 usd? Great 1500 australian for a single gpu, That price could get me 4 of whatever AMD is selling at the time.WAY TO LOSE CUSTOMERS NVIDIA
Originally Posted by xoleras
Actually, it is in the first linked article at sweclockers? Did you check it?
The "soure" suggests. That's as good as a SemiAccurate article with their "sources" that "suggest".
Just like when SA's sources were suggesting Fermi was going to be leaky, late and not as good as everyone was expecting? That was proven false, right?
Generally, Taiwan is a massive rumour mill for the computer industry, if nVidia is beginning to start making these now (As they would to start stockpiling a few cards, ensure there's no bugs anywhere, etc) then other companies and people in Taiwan will know which is where we get the information...Some of it is pure conjecture, but there's always a gain of truth; that price is probably correct or close to the mark at this point in time, if AMD launches a HD8970 that matches it for $500 or if yields are worse than expected among the many other things that can happen pre-launch that affect prices then you can bet the price will change though.
Funny thing is, that first quote in your post...I can get a HD7950 here for $309 with a decent cooler (PCS+), same price is a 660Ti...Yet my card can leave a stock 680 in the dust with a bit of an OC.
Originally Posted by Cloudfire777
FYI: Sweclockers is 100x more reliable than SemiAccurate.
They had launch of Kepler 680 + price right before it was announced by Nvidia and the media. Their sources are trustworthy
No where near the clown Charlie from SemiAccurate
Charlie had that 1) the Fermi we first saw was fake, 2) Fermi was delayed, 3) Fermi was leaky and 4) That Fermi was going to be a hot GPU correct...He's biased as all hell, but if you ignore the biased parts he has some good information a lot of the time.