GeForce GTX 680 review [Guru3D.com] - Page 11

Noisiv · **Memory**: 4GB OCZ 5-4-4-4-12

Quote:

Originally Posted by mameira

why don't you guys go buy a 680 or a 7970 and shut up? its very annoying, none of you won an argument, you are just speaking BS and making this forum look like 4chan.

wow you're a real credit to the team

Agent-A01 · 03-23-2012, 20:37 | posts: 5,645 | Location: USA

Quote:

Originally Posted by Firestorm

what is a 4chan?

Lol... cmon i thought everybody knew what 4chan was

---TK--- · 03-23-2012, 20:39 | posts: 14,744 | Location: New Jersey, USA

not me either

PhazeDelta1 · 03-23-2012, 20:39 | posts: 9,357 | Location: 90° N

Quote:

Originally Posted by Agent-A01

Lol... cmon i thought everybody knew what 4chan was

i know what 4chan is. you'd have to of lived under a rock for the last 9 year to not know.

Quote:

Originally Posted by ---TK---

its the channel between 3 and 5

LOL.

---TK--- · 03-23-2012, 20:43 | posts: 14,744 | Location: New Jersey, USA

must be older than you guys then

PhazeDelta1 · 03-23-2012, 20:47 | posts: 9,357 | Location: 90° N

im not that young. ill be 32 in a few months.

Redemption80 · 03-23-2012, 21:15 | posts: 13,447 | Location: Glasgow

lol, hello fellow child of 1980

Nice to see this has calmed down now, am i just getting bored of it all, or do these GPU launches get more hostile nowadays.

PhazeDelta1 · 03-23-2012, 21:17 | posts: 9,357 | Location: 90° N

yea i have noticed that as well.

this is why we can NEVER have nice things lol.

k1net1cs · 03-23-2012, 21:33 | posts: 3,328

Quote:

Originally Posted by Redemption80

lol, hello fellow child of 1980

1983 here; I'm still young. =b

Redemption80 · 03-23-2012, 21:39 | posts: 13,447 | Location: Glasgow

Last year of your 20s though, all downhill now

So has anyone on here got a hold of one of these yet?

NAMEk · *Last edited by NAMEk; 03-23-2012 at 22:02.*

Quote:

Originally Posted by MonstroMart

I don't think there's any marketing strategy or conspiracy here.

Dunno maybe i don't see one cause i'm not a nVidia fan (i'm not an Amd fan either i buy the best bang for the buck I curently own a 6950 but previously owned a 6800gts other nVidia cards). Why would nVidia not release the best card they can right now ? I mean yes the 680 is a better card (less expensive, better perf overall, better architecture) but the difference is marginal in most games and the 7970 even manage to beat the 680 occasionally. Imo the perm difference is not enough big to steal the Amd fanboys market if both cards are sold at the same price. And it's definately not good enough to convince 7970 owners to trade their card for a 680 unless they are nVidia fans or have money and time to waste. Why would nVidia do that ? Why not release that awesome sauce unbeatable card right now and beat Amd by 30-40% and never get beaten in any game not even Metro ? I just don't understand the strategy at all.

Imo the 680 is the best nVidia could do now, at this power usage, at this temp, at this noise level, at this profit margin and price. I think that saying otherwise is being a little bit delusional. I don't say nVidia will not release a better card later this year (Amd can release one too 3-4 months later) what i'm saying is they simply can't right now and meet their own expectation for the power usage, temp, noise level, profit margin and price.

Now a lot of people will buy gk104, when they release gk110, a lot of people will buy that, instead of buying just gk110. I'm Fan of NVIDIA, but it's facts. First sell 104, later 110, don't you see anything strange with that? heh ;D unless 110 is not ready yet, i mean currently in developing state. I wouldn't be the one, who would change 580 to 680 or 680 to gk110, but a lot of top hardware guys will. I'm not saying that NVIDIA is cheating or anything like that, no. Just strategy. And if i'm wrong - i can't always be right eh? ;D

k1net1cs · 03-23-2012, 21:54 | posts: 3,328

Quote:

Originally Posted by Redemption80

Last year of your 20s though, all downhill now

Some people are like fine wine; they just get better with age.

Me, probably a Bacardi or something, heh.

Quote:

Originally Posted by Redemption80

So has anyone on here got a hold of one of these yet?

Would be as rare as a chicken tooth, I guess.
Pretty much of the stock had dried up across major e-tailers.
If anything it's probably from people selling on eBay already.

IIRC a new batch would be around the 1st week of April, but CMIIW.
Besides, all of the current batch of cards being sold by various manufacturers are actually coming straight from Nvidia, with differences only in sticker and warranty.
So, if there's any manufacturer priced it higher than MSRP (like $529; hello, PNY), it's a sign of which manufacturer you should avoid in the future.

Personally, I'd get it once IB shows up its sexy arse and then I can sell my 2500K - 560 Ti system before the price drops too much.
Maybe wait until MSI gets a Hawk version out; dunno.

Xtreme1979 · *Last edited by Xtreme1979; 03-23-2012 at 22:08.*

Okay first bench ran was 3DMark 11 and here's my results:
http://3dmark.com/3dm11/3023193

Stock clocks on the GPU, only CPU overclock.

Sever · 03-23-2012, 21:57 | posts: 4,826 | Location: Land of the Great Downunder

Quote:

Originally Posted by Noisiv

because need for gk110 does not stem from gaming market - at all,

but computing/workstation.

certain people with certain supercomputers were promised power efficient GPU,
for which GK104 with its graphic workload focus is simply not suitable.

last time i checked, supercomputers generally dont use geforces, they use teslas. im sure nvidia will fix the GPGPU processing power in the tesla range whenever that launches.

Quote:

Originally Posted by MonstroMart

I don't think there's any marketing strategy or conspiracy here.

Dunno maybe i don't see one cause i'm not a nVidia fan (i'm not an Amd fan either i buy the best bang for the buck I curently own a 6950 but previously owned a 6800gts other nVidia cards). Why would nVidia not release the best card they can right now ? I mean yes the 680 is a better card (less expensive, better perf overall, better architecture) but the difference is marginal in most games and the 7970 even manage to beat the 680 occasionally. Imo the perm difference is not enough big to steal the Amd fanboys market if both cards are sold at the same price. And it's definately not good enough to convince 7970 owners to trade their card for a 680 unless they are nVidia fans or have money and time to waste. Why would nVidia do that ? Why not release that awesome sauce unbeatable card right now and beat Amd by 30-40% and never get beaten in any game not even Metro ? I just don't understand the strategy at all.

read my previous post, but to elaborate with an example.

the gk110 is almost 2x bigger than the gk104.

at the current maturation of the 28nm process, there are still a lot of bugs. lets say per wafer, there are 50 bugs causing faulty chips.

now lets assume that per wafer of gk104, due to its small size, you can only get around 200 chips per wafer. now given that the gk110 is supposed to be nearly 2x bigger, you'll only end up with approximately 100 chips per wafer. (this is an example, the reality is we're talking a lot more chips per wafer.

now when you factor in the bugs, this means...

gk110 - 100 chips per wafer - 50 faulty chips caused by bugs = 50 usable chips x $500 sale cost per product (grossly overestimated since im not factoring in that its a lower cost when just buying a chip) = $25000.

gk104 - 200 chips per wafer - 50 faulty chips caused y bugs = 150 usable chips x $500 sale cost (once again im not factoring in the lower cost for buying just the chip) = $75000.

so at the current rate of maturation, nvidia wouldnt make as much money from releasing gk110 on an immature process compared to gk104. with gk104, they can release a product that surpasses the competition, make more money, whilst working on ironing out the bugs in the wafer before releasing a gk110 when it will be more profitable.

---TK--- · 03-23-2012, 21:59 | posts: 14,744 | Location: New Jersey, USA

Quote:

Originally Posted by Xtreme1979

Okay first bench ran was 3DMark 11 and here's my results:
http://3dmark.com/3dm11/3023193

gpu oc?
very good score regardless. think I hit 7300 single 580 give or take a few points

RagDoll_Effect · 03-23-2012, 22:03 | posts: 3,927 | Location: Australia

Quote:

Originally Posted by Xtreme1979

Okay first bench ran was 3DMark 11 and here's my results:
http://3dmark.com/3dm11/3023193

Cool, that's a great P score, is that at stock clocks?

regards,
RagDoll.

k1net1cs · 03-23-2012, 22:03 | posts: 3,328

Quote:

Originally Posted by Xtreme1979

Okay first bench ran was 3DMark 11 and here's my results:
http://3dmark.com/3dm11/3023193

I take it that's on stock clock / TDP?

[edit]
Lol, everyone's asking the same thing. =b
[/edit]

NAMEk · 03-23-2012, 22:05 | posts: 183 | Location: Lithuania

Quote:

Originally Posted by Sever

last time i checked, supercomputers generally dont use geforces, they use teslas. im sure nvidia will fix the GPGPU processing power in the tesla range whenever that launches.

read my previous post, but to elaborate with an example.

the gk110 is almost 2x bigger than the gk104.

at the current maturation of the 28nm process, there are still a lot of bugs. lets say per wafer, there are 50 bugs causing faulty chips.

now lets assume that per wafer of gk104, due to its small size, you can only get around 200 chips per wafer. now given that the gk110 is supposed to be nearly 2x bigger, you'll only end up with approximately 100 chips per wafer. (this is an example, the reality is we're talking a lot more chips per wafer.

now when you factor in the bugs, this means...

gk110 - 100 chips per wafer - 50 faulty chips caused by bugs = 50 usable chips x $500 sale cost per product (grossly overestimated since im not factoring in that its a lower cost when just buying a chip) = $25000.

gk104 - 200 chips per wafer - 50 faulty chips caused y bugs = 150 usable chips x $500 sale cost (once again im not factoring in the lower cost for buying just the chip) = $75000.

so at the current rate of maturation, nvidia wouldnt make as much money from releasing gk110 on an immature process compared to gk104. with gk104, they can release a product that surpasses the competition, make more money, whilst working on ironing out the bugs in the wafer before releasing a gk110 when it will be more profitable.

If you are 100% guaranteed of your statements of immaturation, then my bad and thanks for heads up

Xtreme1979 · 03-23-2012, 22:11 | posts: 1,227 | Location: Bay City, MI

Quote:

Originally Posted by k1net1cs

I take it that's on stock clock / TDP?

[edit]
Lol, everyone's asking the same thing. =b
[/edit]

Yes, everyone sorry stock GPU clocks. BTW here is Heaven 3.0 max everything:

k1net1cs · 03-23-2012, 22:16 | posts: 3,328

I think I just farted rainbows. Tesselated rainbows.

---TK--- · 03-23-2012, 22:16 | posts: 14,744 | Location: New Jersey, USA

add some gpu mhz. about 100 can you post back? be carefull a photoshopped 680 burnt the vrm`s from just a 100mhz increase lol

harkinsteven · 03-23-2012, 22:17 | posts: 1,537 | Location: Ireland

Someone test The Witcher 2 with ubersampling please

fr33k · 03-23-2012, 22:23 | posts: 1,784 | Location: USA

Quote:

Originally Posted by ---TK---

add some gpu mhz. about 100 can you post back? be carefull a photoshopped 680 burnt the vrm`s from just a 100mhz increase lol

please don't continue the troll thread.

Noisiv · **Memory**: 4GB OCZ 5-4-4-4-12

Quote:

Originally Posted by Sever

last time i checked, supercomputers generally dont use geforces, they use teslas. im sure nvidia will fix the GPGPU processing power in the tesla range whenever that launches.

I dont quite follow you.... I didn't even mention GeForce.
But they will proly use same GK110, and although designing it for Tesla, offer it through GeForce SKU.

And Nvidia won't fix GPGU, nothing is broken on GK104 - but they will design chip around general computing this time, not around graphics like for GK104.

Quote:

Originally Posted by Sever

l

so at the current rate of maturation, nvidia wouldnt make as much money from releasing gk110 on an immature process compared to gk104.

WTH are u talking about

one GF100 through Quadro 6000 is selling for $4,000 (four thousand dollars)

---TK--- · 03-23-2012, 22:27 | posts: 14,744 | Location: New Jersey, USA

Quote:

Originally Posted by fr33k

please don't continue the troll thread.

sure thing back seat mod