First: GTX 680 review !!!

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

Screech

Golden Member
Oct 20, 2004
1,203
7
81
AMD and some of its vendors' OCd versions of the 7970 remind me a bit of nvidia's 460 and its overclocked variants. nvidia touted 460 OC type cards as competition for some of the 68xx cards iirc, maybe amd will be in a similar boat here.
 

mak360

Member
Jan 23, 2012
130
0
0
4jKjF.jpg


first non-fake benchmark?


Is that the temp?
 

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0

Sure it could be, but then again consider the source...an unknown source NOT affiliated with a website? Come on. There's no need to stand up for an obviously shady source. I suspect its fake, but in any case we'll see true numbers in less than a week.

I'll be first to praise the gk104 and eat crow if the rumors are true, but these fake reviews aren't cutting it. I'll wait for the AT benchmarks.
 
Last edited:

mak360

Member
Jan 23, 2012
130
0
0
A site showing benches/leaks and painting nvidia in the best light but failing to show overclocking says it all for me.
 

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
Just read up on adaptive Vsync and FXAA in the driver. Finally. Big props to nvidia for this, even if we're not certain about the performance.

About damn time they put FXAA in the driver, happy they did this. I was never a fan of modifying .exe files to enable FXAA. :thumbsup:
 

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,250
136
Take the fake review with a pound of salt :)

Will be interesting once real reviews pop up tho.
 

epidemis

Senior member
Jun 6, 2007
794
0
0
If this image on the Dutch website is correct:
NVIDIA-GeForce-GTX-680-GK104-Kepler-specs-575x419.jpg


Then it has a transistor density of 0,012 gigatransistor/mm^2 opposed to GTX 580's 0,0057 gigatransistor/mm^2
The growth is transistor density is then (0,012-0,0057/0,0057)= 110%
The physical scaling of the transistor from 40nm to 28 is (40^2-28^2)/28^2 = 104%

So it's above perfect scaling! They have crammed the transistors tighter together than their 40 nm process. That might explain why some sites have reported some problems at producing them at TMSC!

The real feat is though (if the review is correct) it accomplishes more than 7970 despite having a footprint that's 20% percent smaller!
 
Last edited:

blackened23

Diamond Member
Jul 26, 2011
8,548
2
0
If this image on the Dutch website is correct:
NVIDIA-GeForce-GTX-680-GK104-Kepler-specs-575x419.jpg


Then it has a transistor density of 0,012 gigatransistor/mm^2 opposed to GTX 580's 0,0057 gigatransistor/mm^2
The growth is transistor density is then (0,012-0,0057/0,0057)= 110%
The physical scaling of the transistor from 40nm to 28 is (40^2-28^2)/28^2 = 104%

So it's above perfect scaling! They have crammed the transistors tighter together than their 40 nm process. That might explain why some sites have reported some problems at producing them at TMSC!

That chart is riddled with errors. Like, too many to even bother naming. Seriously, WTF.
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Sure it could be, but then again consider the source...an unknown source NOT affiliated with a website? Come on. There's no need to stand up for an obviously shady source. I suspect its fake, but in any case we'll see true numbers in less than a week.

I'll be first to praise the gk104 and eat crow if the rumors are true, but these fake reviews aren't cutting it. I'll wait for the AT benchmarks.

Sure, more investigations and reviews from third parties are more than welcomed.
 

wand3r3r

Diamond Member
May 16, 2008
3,180
0
0
Supposing there is a spec of truth in any of the charts, when it says 1006 MHz, but if it automatically overclocks up to 1100 MHz I wonder what the overclocking headroom could be. It seems it is already overclocking itself to TDP limits.

The charts should probably be displaying the overclocked turbo clocks in my opinion.
 

DarkKnightDude

Senior member
Mar 10, 2011
981
44
91
Benchmarks look fake. 74 fps on a 7970 and 680 for MW3? We're talking about MW3 here. You would think the framerates are way higher.
 

sontin

Diamond Member
Sep 12, 2011
3,273
149
106
I think they don't have access to the real Review board, drivers and bios. Look at the "press deck" or better the fotos...

There is another picture of the board in the wild which has all 5 power phase.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
I sure would like Keys to chim in . As these NV gpus look most impressive. These are the 104s correct . Keys is this the highend or is something better coming. If I need to wait for the high end thats fine . I can get one of these . But if this is the highend I need 2. As for the high end being canceled I not real sure as the 580 coming out when it did took everyone by surprize . It also caused me to move to the green team as I really liked how NV handled the 580 release. I suspect this is the midrange card and NV is sand bagging. Just like they did with the 580 release . GO NV. Great drivers buy the way . NV does do drivers pretty well. I am hooked on green. Than the die size very impressive . Even the AMD guys would agree as we have thousands of past post . praising AMDs smaller die .
 
Last edited:

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
Benchmarks look fake. 74 fps on a 7970 and 680 for MW3? We're talking about MW3 here. You would think the framerates are way higher.

I get higher FPS than the 580 in that bench. I don't know how anyone can take these benches seriously.

16185034341892959696.png


Vs

43883.png


8xAA or not, how did the 7970 and 580 lose almost half their performance?

1702470727834307124.png


Vs

43890.png


I didn't know the 7970 was this much faster than the 580 in BF3. And apparently their magical 8xAA boosts performance.
 
Status
Not open for further replies.