GTX 680 Release Date?

[BoFox]

Do you people really expect a card with 256bit bus to be significantly faster than 7970? 7970 loses 13% of its performance when you lower its bandwidth by 33%. GK104 would need to be a hell of a lot better than tahiti and/or have way better z-culling algorithms for that to be true.

Well, I'm all for more bandwidth too, but HD 7870 has a 256-bit bus, and actually beats HD 7950 with its 384-bit bus in a few games. When you compare the specs, the 7870 seems to be far more efficient specs-wise, even with much lower bandwidth. It's something else that hinders the efficiency of 32 ROPs with an odd number of 384-bit bus in correlation, so I do not think GTX 680 would be trailing too far off behind when it comes to bandwidth efficiency.

Unless GTX 580 was already SEVERELY bandwidth-limited, it certainly is possible to add 50% more GPU muscle and still get 30-40% increase in overall performance for example (with GTX 680 having equal bandwidth).

 

[railven]

ripping ppl off!!!
thats what they do. remember???

Nvidia is trying to make money selling graphics cards!!!! how dare them!!! They are gonna sell a mid range chip for over $500? what!!!!

1) if you dont like the price, dont buy it!!!
2) if its faster than a 580, faster than 7970, do you think its midgrade?
3) if you only want to pay 250$ for a graphics card, than buy a 250$ gpu!!!
4) this is just stupid!

it is a nice spin, but there are facts that just arent gonna let this spin get very popular.

1) we pay for cards based on their performance.
2) people payed $550 for the 7970
3) they will pay it for a card that is faster
4) People are used to paying high dollar for nvidia cards.
5) this spin is laughable

Besides. if its called midrange, it just makes AMD look worse. Nvidia is a stronger brand name and typically a higher premium. The gk104 has a chance of actually having a better price/performance ratio than the 7970. There is no way your gonna sway customers away by saying nvidis is ripping them off when they are giving a better deal than AMD. Its not possible!

nd if AMD drops prices on the 7970, nvidia can still sell cards with a high price. Thats nvidia , thats what they do!!!!!!

Nvidia is the intel of GPUs. They have a stronger brand name. its just how it is. its how its been

This whole post feels like dejavu! I swear I said about 90% of this before but was told something about significant node change, new architectures, and then accused of being a corporate cheerleader because I didn't mind paying for the "premiums."

I still agree with 90% of the things being said here. Good to see it's okay to pay $550 for a "mid-range" card haha Since I'll be buying two this generation

 

[ocre]

This whole post feels like dejavu! I swear I said about 90% of this before but was told something about significant node change, new architectures, and then accused of being a corporate cheerleader because I didn't mind paying for the "premiums."

I still agree with 90% of the things being said here. Good to see it's okay to pay $550 for a "mid-range" card haha Since I'll be buying two this generation

Wait, i am gonna blow everyones mind!!!! If the gk104 is midrange then:

the HD7970 is midrange too!!!!

They are close to the same size die. I guess AMD started seriously ripping off their customers for a while now!!!!

 

[Olikan]

Well, I'm all for more bandwidth too, but HD 7870 has a 256-bit bus, and actually beats HD 7950 with its 384-bit bus in a few games. When you compare the specs, the 7870 seems to be far more efficient specs-wise, even with much lower bandwidth. It's something else that hinders the efficiency of 32 ROPs with an odd number of 384-bit bus in correlation, so I do not think GTX 680 would be trailing too far off behind when it comes to bandwidth efficiency.

Unless GTX 580 was already SEVERELY bandwidth-limited, it certainly is possible to add 50% more GPU muscle and still get 30-40% increase in overall performance for example (with GTX 680 having equal bandwidth).

the 7950 don't use the all bandwith, the 7970 just get a hard bandwidth-limit at 1.1GHz.
the 7870 have better caches, that's why it is better than a 6970 with lower bandwidth.

it's hard to say about kepler, as known, it have lower bandwidth, probably lower caches since the die is small... but, it's a new design...so, who knows?

 

[railven]

Wait, i am gonna blow everyones mind!!!! If the gk104 is midrange then:

the HD7970 is midrange too!!!!

They are close to the same size die. I guess AMD started seriously ripping off their customers for a while now!!!!

If you blow people's mind, kudos. Mid-range/High, etc. I always separate by cost - not die size, GPU code name, or something I only know if I visit a forum and then have to remove my heatsink, clean of TIM, and look closely.

$550 - gets you a good card (AMD side.) If nVidia delivers on that front, kudos! The more choices the better.

 

[BoFox]

Wait, i am gonna blow everyones mind!!!! If the gk104 is midrange then:

the HD7970 is midrange too!!!!

They are close to the same size die. I guess AMD started seriously ripping off their customers for a while now!!!!

And even worse if the "mid-range by designation" GK104 actually beats AMD's high-end flagship product.. All of this argument of it being mid-range makes NV sound even more marvelous.

What if OBR is really right? He probably is with the Czech prices, but it could also be indicative that NV is going to set it at $399 here..

Perhaps it is because NV plans on doing a beefed-up GK104 "Ti" very shortly, with 8x6-pin power connectors (which the board already supports).

And then the monster Kepler is probably due within 3-4 months from now (if Charlie is correct about it being taped out last month, and if things go well).

$399 sounds like a pretty good price-point, similar to when HD 5870 came out (with the prices creeping up anyways, making the AIBs real happy).

Anything new so far with the leaks?

 

[Vesku]

$400 would be great, would bring price/performance back into line for a node change.

 

[BoFox]

the 7950 don't use the all bandwith, the 7970 just get a hard bandwidth-limit at 1.1GHz.
the 7870 have better caches, that's why it is better than a 6970 with lower bandwidth.

it's hard to say about kepler, as known, it have lower bandwidth, probably lower caches since the die is small... but, it's a new design...so, who knows?

Tahiti has 50% larger L2 cache than Pitcairn. The B3D guys are pretty clueless (with Baumann only giving a cryptic hint/clue), as I have already pointed out here: http://alienbabeltech.com/abt/viewtopic.php?f=73&t=24402&start=20#p60418 (The last post over there is my attempt to make sense of it..)

 

[Awkward]

And even worse if the "mid-range by designation" GK104 actually beats AMD's high-end flagship product.. All of this argument of it being mid-range makes NV sound even more marvelous.

What if OBR is really right? He probably is with the Czech prices, but it could also be indicative that NV is going to set it at $399 here..

Perhaps it is because NV plans on doing a beefed-up GK104 "Ti" very shortly, with 8x6-pin power connectors (which the board already supports).

And then the monster Kepler is probably due within 3-4 months from now (if Charlie is correct about it being taped out last month, and if things go well).

$399 sounds like a pretty good price-point, similar to when HD 5870 came out (with the prices creeping up anyways, making the AIBs real happy).

Anything new so far with the leaks?

...

It's their fastest offering out now, it's priced as such, and there is no immediate successor in sight. It's not a mid-range product.

And then the monster Kepler is probably due within 3-4 months from now (if Charlie is correct about it being taped out last month, and if things go well).

Pretty much sums up the worth of your post.

 

[superjim]

The entire "is GK104 mid-range?" debate started from the rumor that NV decided to bump up the 104 to its flagship 680 card after seeing benchmarks of the 7970, leaving GK110 as a refresh/next-cycle product (i.e. 780). If I was one to throw money down on any of the rumors around Kepler, I'd probably do so on this one. It makes complete sense from a financial perspective. If the competition (7970) is selling at $550 and you can make your planned $300 mid-ranged part meet this performance, that's money in the bank. Charlie wrote an article about this (GK104 profit margins) 1-2 weeks ago.

 

[railven]

The entire "is GK104 mid-range?" debate started from the rumor that NV decided to bump up the 104 to its flagship 680 card after seeing benchmarks of the 7970, leaving GK110 as a refresh/next-cycle product (i.e. 780). If I was one to throw money down on any of the rumors around Kepler, I'd probably do so on this one. It makes complete sense from a financial perspective. If the competition (7970) is selling at $550 and you can make your planned $300 mid-ranged part meet this performance, that's money in the bank. Charlie wrote an article about this (GK104 profit margins) 1-2 weeks ago.

From my recollection on this forum, "paying $550 for a mid-range part" stemmed at the low performance gains HD 7970 had versus previous node shifts and the increased price tag. The paying a premium arguement.

Then the rumors trickled out about GK104 becoming GTX 670, then into GTX 680, then confirmed. During the begin you'd see people deny the claim because GK104 isn't the successor to GF110, it's the successor to GF114 so it can't be called GTX 670 or GTX 680.

Fast forward to it becoming official and a lot of negatives changed. Now that both sides are charging a "premium" for their "top end" card regardless of what is under the hood, it's perfectly OK - it's just ironic how we got here to now calling GTX 680 a "mid-range" part.

And I agree with you, these companies both are in it to make bank. If the GK104 competes with HD 7970, launch it at a similar price point (I'd personally prefer they start a price war, but - that's me being a selfish consumer. )

 

[blackened23]

....they had to do it out of necessity because GK100 could not pass engineering sample status due to interconnect problems. Personally I find anyone suggesting that an "unreleased product" is really the high end, even though it won't be released until the end of the year, to be pretty funny. We can make a laundry list of unreleased products and proclaim them high end, such as haswell, sea islands, etc.

That is why GK104 is the flagship, for now.

 

[Olikan]

Tahiti has 50% larger L2 cache than Pitcairn. The B3D guys are pretty clueless (with Baumann only giving a cryptic hint/clue), as I have already pointed out here: http://alienbabeltech.com/abt/viewtopic.php?f=73&t=24402&start=20#p60418 (The last post over there is my attempt to make sense of it..)

but pitcairn have 3 CUs per L1...

anyway, if it's not the caches, what the hell is?

 

[JTsyo]

Are the other Keplars cards also expected to be released with the 680 or are they going to be a few weeks later?

 

[tviceman]

....GK100 could not pass engineering sample status due to interconnect problems.

Where did you get that fancy info?

 

[Keysplayr]

....they had to do it out of necessity because GK100 could not pass engineering sample status due to interconnect problems. Personally I find anyone suggesting that an "unreleased product" is really the high end, even though it won't be released until the end of the year, to be pretty funny. We can make a laundry list of unreleased products and proclaim them high end, such as haswell, sea islands, etc.

That is why GK104 is the flagship, for now.

This looks like a cut and paste. It also has a ring of an older S|A article.
http://semiaccurate.com/2011/09/20/exclusive-nvidia-has-two-keplers-in-house-but-not-the-big-one/
Who did you quote this from?

 

[badb0y]

If GK100 is in fact in house and ready to be released in 4 months why call this card a GTX 680? Wouldn't it be better to humiliate AMD by releasing a card named GTX 660 Ti that beats a HD 7970 and then later on release a GTX 680 that completely roflstomps the HD 7970?

The only logical conclusion is that GK100 is not ready for at least another 5 months or more so nVidia decided to bump up the GK104 to a GTX 680 since it was competitive with a HD 7970.

Pure speculation so I can be completely wrong here.

 

[Arzachel]

Where did you get that fancy info?

Yeah, I believe GK100 simply didn't yield, not much more to it.

If GK100 is in fact in house and ready to be released in 4 months why call this card a GTX 680? Wouldn't it be better to humiliate AMD by releasing a card named GTX 660 Ti that beats a HD 7970 and then later on release a GTX 680 that completely roflstomps the HD 7970?

The only logical conclusion is that GK100 is not ready for at least another 5 months or more so nVidia decided to bump up the GK104 to a GTX 680 since it was competitive with a HD 7970.

Pure speculation so I can be completely wrong here.

They couldn't sell the 660Ti for >500$

 

[notty22]

....they had to do it out of necessity because GK100 could not pass engineering sample status due to interconnect problems. Personally I find anyone suggesting that an "unreleased product" is really the high end, even though it won't be released until the end of the year, to be pretty funny. We can make a laundry list of unreleased products and proclaim them high end, such as haswell, sea islands, etc.

That is why GK104 is the flagship, for now.

Is that not what you just said in this post ? lol
The first bold was the rhetorical question and the second bold your own answer.

 

[BoFox]

Just to rub it in to those with annoying anti-680 agendas that's making this thread tiresome to read, I'll share some more news here:

Looks like Kepler will indeed support DX11.1, if GPU-Z (the new version supporting GTX 680) is correct..

You were already warned to knock it off with the flamebaiting and inflammatory posting. Sorry you couldn't resist the temptation to see if we were serious...not to rub it in but you just talked yourself into a vacation, you will be released from vacation 2-days after Kepler's release, however far off that may be.

Public Service Announcement for the rest of the readership - do not post in kind as BoFox has done unless you wish to join him in his vacation.

The baiting and flaming in the run-up to Kepler release has been unacceptable, it ends now.

Administrator Idontcare

 

[notty22]

Interesting a boost reading , instead of shader clocks.

Here is the 6.0 gpu-z link
http://www.techpowerup.com/162639/TechPowerUp-GPU-Z-v0.6.0-Released.html

 

[Keysplayr]

Just to rub it in to those with annoying anti-680 agendas that's making this thread tiresome to read, I'll share some more news here:

Looks like Kepler will indeed support DX11.1, if GPU-Z (the new version supporting GTX 680) is correct..

Does the Unigine results shot display Intel HD Graphics by default if you don't turn off the integrated video in the machines BIOS? Cause according to that screenie, an Intel HD3000 chip is kicking the snotzky out of a 7970.
I'd also like to add that results can be skewed when run in windowed mode. So those numbers are probably not accurate.

 

[SlowSpyder]

If GK100 is in fact in house and ready to be released in 4 months why call this card a GTX 680? Wouldn't it be better to humiliate AMD by releasing a card named GTX 660 Ti that beats a HD 7970 and then later on release a GTX 680 that completely roflstomps the HD 7970?

The only logical conclusion is that GK100 is not ready for at least another 5 months or more so nVidia decided to bump up the GK104 to a GTX 680 since it was competitive with a HD 7970.

Pure speculation so I can be completely wrong here.

They're calling it the GTX680 (presumably) because they can charge more for it (go premiums). No one wants a $500 GTX660Ti, it would make me wonder what the GTX680 can do, maybe I should wait. But if they call it the GTX680 and price it appropriately compared to the 7970 (assuming they perform close to one another) it can be sold as their current flagship and for a lot of money.

I think newer more advanced fabrication processes are going to come slower than what we were used to in the past. How long have we been on 40nm GPU's now? AMD and Nvidia don't want to blow their load on 28nm already, by tweaking/updating Tahiti and letting the GK110 bake longer these companies can sell these $500+ GPU's twice. AMD has an 8970 cooking, Nvidia will release the GK110 as the GTX685 or GTX780. <-- My guess as to what is going to happen based on nothing but my opinion.

 

[3DVagabond]

Well, I'm all for more bandwidth too, but HD 7870 has a 256-bit bus, and actually beats HD 7950 with its 384-bit bus in a few games. When you compare the specs, the 7870 seems to be far more efficient specs-wise, even with much lower bandwidth. It's something else that hinders the efficiency of 32 ROPs with an odd number of 384-bit bus in correlation, so I do not think GTX 680 would be trailing too far off behind when it comes to bandwidth efficiency.

Unless GTX 580 was already SEVERELY bandwidth-limited, it certainly is possible to add 50% more GPU muscle and still get 30-40% increase in overall performance for example (with GTX 680 having equal bandwidth).

Check reviews where they are running the same drivers on the 7950 as the 7870, not the 7950 release drivers, and I think you'll find the 7950 beats the 7870 pretty consistently. Also the 7950 is severely clock limited @ 800MHz. If you set the two cards up with the same drivers and clock speeds the 7950 whipes the floor with the 7870.

This isn't to say that the 680 will or won't beat the 7970. Just that trying to say the 7870 is in any way on par with the 7950 isn't accurate once you compare apples to apples.

 

[BoFox]

Does the Unigine results shot display Intel HD Graphics by default if you don't turn off the integrated video in the machines BIOS? Cause according to that screenie, an Intel HD3000 chip is kicking the snotzky out of a 7970.
I'd also like to add that results can be skewed when run in windowed mode. So those numbers are probably not accurate.

Wow, I didn't know that HD3000 could do DX11 tessellation! :awe:

Maybe it was windowed so he could show GPU-Z and EVGA Preicison on the same screen? Dunno.. Do you by any chance happen to know the "real stuff" because of being a Focus Member? I know it's wise to try to keep people from being over-hyped and then therefore disappointed, but I'm just curious, man!