[vrzone] GTX670Ti in March

[thilanliyan]

well even if only matches the current gtx580 but fills in the gtx570 price slot which is 350 bucks then I dont think that is much of an accomplishment at all. and if it cost more than the current gtx570 while only matching or barely beating the gtx580 then that is a complete failure imo.

I don't think AMD or nV are sandbagging on purpose. The process tech has had troubles at 40nm and now it seems at 28nm as well. It just seems to be getting harder and harder to die shrink without issues. You probably ARE going to wait a while to double GTX570 performance for the same price.

 

[blackened23]

Wow, I guess there is no convincing you that Kepler isn't late just because Tahiti launched. Is this a hard concept to wrap your mind around? Or is it something else that I'm missing?

Well, didn't nvidia state publicly that Kepler would be released in 2011 by their original roadmap? I'm sure nvidia is on their own schedule (ie when its done) but you have to think that competetors like to answer each other tit for tat. Know what I mean?

Like I said, if BigK is all that i'll be all about it. There was no denying that GF110 was a better part than the 6970, so if BigK is better i'll definitely buy it...but blaming AMD for any of this is ridiculous IMO

 

[OILFIELDTRASH]

So this card will cost $300 and be faster than a 7970, right ? ()

Have you bought your 7970's yet?

 

[3DVagabond]

Since it's been quiet over the weekend here's another rumor:

http://vr-zone.com/articles/report-nvidia-gk104-releasing-in-march-as-geforce-gtx-670-ti/14952.html

From that article:
As the nomenclature suggests, GK104 / GTX 670 Ti is more of a successor to the GF114 / GTX 560 Ti. The flagship GTX 580 sucessor is at least half a year away, and will not be released till Q3 or Q4 2012. Till then, AMD is expected to hold on to the single GPU crown with Radeon HD 7970 with the GTX 670 Ti and HD 7950 following behind.

Not saying any of it's true, but if anyone is going to accept part of it as factual...

Also, Nvidia's usage of the high performance process rather than AMD's low performance process should make the GK104 strong if the architectures are similar in strength per die mm^2.

Do you have a source for this? The reason I ask is I've read that AMD hasn't revealed what 28nm process they are using is, but that HPL isn't suitable for GPU's only HP is.

 

[thilanliyan]

Wow, I guess there is no convincing you that Kepler isn't late just because Tahiti launched. Is this a hard concept to wrap your mind around? Or is it something else that I'm missing?

Not that Kepler has to launch at the same time as Tahiti BUT, didn't JHH say that Kepler yields are low, which means whatever card was going to be released is or is going to be late?

 

[Keysplayr]

Well, didn't nvidia state publicly that Kepler would be released in 2011 by their original roadmap? I'm sure nvidia is on their own schedule (ie when its done) but you have to think that competetors like to answer each other tit for tat. Know what I mean?

Like I said, if BigK is all that i'll be all about it. There was no denying that GF110 was a better part than the 6970, so if BigK is better i'll definitely buy it...but blaming AMD for any of this is ridiculous IMO

I'm sure you wouldn't want AMD to pre-maturely launch their products just to beat out Nvidia to market. You'd want it when it was ready for prime time. I know you, you are a perfectionist.

 

[Keysplayr]

Not that Kepler has to launch at the same time as Tahiti BUT, didn't JHH say that Kepler yields are low, which means whatever card was going to be released is or is going to be late?

It could. It could also mean 165 other things if you really try and think of them.

 

[blackened23]

I'm sure you wouldn't want AMD to pre-maturely launch their products just to beat out Nvidia to market. You'd want it when it was ready for prime time. I know you, you are a perfectionist.

Fair enough. I'm assuming GK110 or 112 will be worth the wait, no doubts here. The specs alone are pretty monstrous.

 

[toyota]

You mean set the bar low, like oh lets see, GTX 480 499$ for 10% performance over the 5870 while generating 2x more power and heat? Like that?

Now its suddenly *amd's* fault that NV can't get their stuff out on time. Wonderful. I'm not trying to slight nvidia here, but blaming AMD for NV's failure is ridiculous. (disclaimer: no, we don't know if these rumors are true) Stop shifting the blame, NV are likely late to the game just like the last go around. The blame lies solely with them and its somewhat ironic that so many people consider nvidia infallible.

If NV puts their money where their mouth is and blows the 7970 away, trust me, i'll be all about it. I'll buy 2 of those suckers on day 1 and enjoy it. But blaming AMD for NV's prices and lateness? Come on dude.

and I did not think the gtx480 or gtx470 were good cards when all factors were considered. I got desperate and jumped on the first decent gtx470 deal that happened and ended up returning it because it was not worth the heat, power consumption and noise. Nvidia has its share of issues and being late for the last two gens is certainly a major issue.

again though right here and now, AMD has not set the bar very high for a next card at its price points. and also again that is going to allow Nvidia to sell what was intended as gtx660 cards for gtx670 or maybe even gtx680 price points. Nvidia is only going to take advantage of this situation so yes BOTH companies are to blame for the prices we will have.

 

[Quantos]

Not that Kepler has to launch at the same time as Tahiti BUT, didn't JHH say that Kepler yields are low, which means whatever card was going to be released is or is going to be late?

Not necessarily late. Low yields could mean low availability and/or higher prices.

 

[Grooveriding]

Wonders never cease. The hard part is, it's killing you that you don't know what's coming.

Looooool.

Good for a laugh as always. It's a vrzone rumour, same source saying nvidia is also late in their high end card until September, shouldn't you be scrambling to try and debunk this one as well ?

Oh wait, it's mildly positive for nv, got it.

Stepping out of nv fcg lala land...


I believe this one like the one about no high end till late this year. Should be a good card, gtx580 performance for hopefully the $3XX price bracket.

 

[Grooveriding]

Have you bought your 7970's yet?

Heck no, I am grabbing a couple of the $300 nvidia cards that are faster than the 7970, are these the ones ?

 

[DeeJayeS]

Wow, I guess there is no convincing you that Kepler isn't late just because Tahiti launched. Is this a hard concept to wrap your mind around? Or is it something else that I'm missing?

I'm reminded of a line from the Lord of the Rings movie, when Gandalf tells Frodo: "A wizard [Kepler? ] is never late, nor is he early. He arrives precisely when he means to."

/queue laughter

 

[tincart]

I think people have diverging views of what "late" means.

 

[Lonyo]

Wow, I guess there is no convincing you that Kepler isn't late just because Tahiti launched. Is this a hard concept to wrap your mind around? Or is it something else that I'm missing?

Party starts at 8pm.
You only plan to arrive at 10pm. You arrive at 10pm as planned.

Part 1) Are you late to the party?


Part 2) Did you inform the host of the party of when you planned to arrive? Does the answer to this change the answer to part 1?

 

[thilanliyan]

It could. It could also mean 165 other things if you really try and think of them.

But do those other things really matter? Fact is that AMD has their first 28nm cards out and nV doesn't, which sucks for consumers.

nV not actually stating a release date doesn't make the fact that they are slower to the market than AMD easier to bear, especially for the people that are willing to consider both camps.

 

[blackened23]

I think people have diverging views of what "late" means.

For this discussion, its this:

I'd say its late. Like I said earlier however, I have a gut feeling that GK110 / 112 or whatever it will be called will be worth the wait. Rumor has it that NV finally made efficient parts, so i'm anxious to see how the GPU wars turn out later this year.

 

[3DVagabond]

For this discussion, its this:

I'd say its late. Like I said earlier however, I have a gut feeling that GK110 / 112 or whatever it will be called will be worth the wait. Rumor has it that NV finally made efficient parts, so i'm anxious to see how the GPU wars turn out later this year.

If you look at that chart Fermi was late. So, that means Kepler isn't. Even though it was repeatedly said that one design being late wouldn't effect the other. After all, they were started years ago and are being developed by different design teams.

Add to that if rumors are true that BigGK will debut as the 780. So, then it'll be early.

 

[Vesku]

Even if Kepler ends up being incredibly similar to AMD 28nm in terms of performance per mm2, NVIDIA will EVENTUALLY be bringing us the biggest chip that can be made on the fab process. Where as AMD has stuck to optimal sizes. So assuming they don't end up canceling their top end product, they will eventually have the fastest single chip you can buy. Granted it might cost $1000 but the only near par performance card alternative from AMD will be dual chip.

In the near term, however, this looks like the 5800 series story part deux.

 

[BD231]

I'll be interested in 580 performance when its a 160 to 200 watt part.

 

[Grooveriding]

For this discussion, its this:

I'd say its late. Like I said earlier however, I have a gut feeling that GK110 / 112 or whatever it will be called will be worth the wait. Rumor has it that NV finally made efficient parts, so i'm anxious to see how the GPU wars turn out later this year.

I am expecting GK110 to be amazing. AMD has not done a big die in a long time and I doubt they will now. So, I think when we see the real high end Kepler with a 450mm2+ die this fall, it is going to be really fast. Hopefully 7970 +40%, until then it will be whatever the full GK104 core/clock part is and 7970, likely trading blows depending on the game.

August/September until we get a card faster than a 7970. I think I am getting a couple of the GK110s unless AMD breaks te mold and gives us a big die. I can live with 580 tri sli performance for another six months.

 

[VulgarDisplay]

I am expecting GK110 to be amazing. AMD has not done a big die in a long time and I doubt they will now. So, I think when we see the real high end Kepler with a 450mm2+ die this fall, it is going to be really fast. Hopefully 7970 +40%, until then it will be whatever the full GK104 core/clock part is and 7970, likely trading blows depending on the game.

August/September until we get a card faster than a 7970. I think I am getting a couple of the GK110s unless AMD breaks te mold and gives us a big die. I can live with 580 tri sli performance for another six months.

In the meantime I'll be happy with my 7970 performance for the next 6 months. Especially since when it's overclocked it can be +25% faster than stock fairly easily.

 

[Lonyo]

For this discussion, its this:

[snip]

I'd say its late. Like I said earlier however, I have a gut feeling that GK110 / 112 or whatever it will be called will be worth the wait. Rumor has it that NV finally made efficient parts, so i'm anxious to see how the GPU wars turn out later this year.

https://twitter.com/#!/nvidia/status/25141305492

@TheMadJack here you go: Keple0r - 2011, Maxwell - 2013 #gtc2010

(AMD was also late, since they said 2011, but they have now launched, so while they launched late, they have done it. NV are late and still haven't launched)

 

[bunnyfubbles]

For this discussion, its this:

I'd say its late. Like I said earlier however, I have a gut feeling that GK110 / 112 or whatever it will be called will be worth the wait. Rumor has it that NV finally made efficient parts, so i'm anxious to see how the GPU wars turn out later this year.

considering Fermi never made it to 2009 and got pushed back to 2010, large in part because of fabrication process woes that have ultimately continued to this day, I'd say we're right on schedule if we adjust that 2011 Kepler projection for that delay.

 

[blackened23]

3dvagabond is nostradamus