• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The first "proper" Kepler news Fri 17th Feb?

Right from the presentation for Kepler and Maxwell Nvidia made it clear that this first initial Kepler card would not be high and and would not be the replacement for Fermi.

That's news to me? :confused:

Where did you see / hear / read that?


I agree. The GTX 680 is only a x80 series by name. It was never intended to be the true successor to the 580 and it is only that now because Nvidia had no choice.

While it's clear from the architecture that GK104 wasn't originally supposed to be the high-end part, I don't think that Nvidia "had no choice" but to release it as the high-end card. The choice was provided to them by the 'less-than-stellar' performance of the 7970 - giving them the option of competing with their upper mid-range part.

Had the 7970 performed better (or even just been released at 1100Mhz...) I think Nvidia would have gone with the 670Ti label, and given something of a 'paper launch' preview of the GK110. As it is, they have the option of marketing GK104 as a high-end card. Which is a shame for us, the consumer, as it means a higher pricetag.
 
That's news to me? :confused:

Where did you see / hear / read that?




While it's clear from the architecture that GK104 wasn't originally supposed to be the high-end part, I don't think that Nvidia "had no choice" but to release it as the high-end card. The choice was provided to them by the 'less-than-stellar' performance of the 7970 - giving them the option of competing with their upper mid-range part.

Had the 7970 performed better (or even just been released at 1100Mhz...) I think Nvidia would have gone with the 670Ti label, and given something of a 'paper launch' preview of the GK110. As it is, they have the option of marketing GK104 as a high-end card. Which is a shame for us, the consumer, as it means a higher pricetag.

Do you see the lack of the GK100?? I do. The GK110 moniker indicates it to be a second generation Kepler part like the GF110 was the second generation Fermi part. If anything it probably means the big Kepler was delayed anyway. You forget that big Kepler will be the basis of the Nvidia compute cards such as the high end Tesla and Quadro ranges and the profit margins on them are much more than mere gaming cards. If they had the GK100 or GK110 up and running at the very least they would be talking about its non-gaming performance.

So what I think happened is the following:
1.)The high end big Kepler hit issues. Probably the GK100 was canned and development moved over to the GK110,which meant a delay.
2.)Nvidia knew AMD would release their parts quicker than them and had less delays.
3.)Nvidia knew AMD moved over to a more a compute focussed architecture. They expected a very large performance jump over the HD6900 series(more than 40%).
4.)The smaller GK104 was more or less fine(or not delayed as much as big Kepler).
5.)Nvidia expected that the GK104 would be competing with GCN based cards lower than the HD7900 series and would have no answer to the HD7970.
6.)The HD7900 series was released and then they realised that a higher clockspeed version of the GK104 should be able to compete with or exceed the HD7970.
7.)Hence the comments they made and the pricing.
 
Last edited:
That's news to me? :confused:

Where did you see / hear / read that?

On the presentation wall.

What I meant of course was that the initial Kepler cards would not be high end. Initially they were meant to be laptop parts and mid ranged desktop parts.

However, they are now high end with names and a price tag to match, so will now replace Fermi completely.

The true (in my opinion of course !) replacements will be the high end parts. That will (to me at least) be the true replacement for Fermi and not the mid ranged/tweaked and rebadged 680.

While it's clear from the architecture that GK104 wasn't originally supposed to be the high-end part, I don't think that Nvidia "had no choice" but to release it as the high-end card. The choice was provided to them by the 'less-than-stellar' performance of the 7970 - giving them the option of competing with their upper mid-range part.

Had the 7970 performed better (or even just been released at 1100Mhz...) I think Nvidia would have gone with the 670Ti label, and given something of a 'paper launch' preview of the GK110. As it is, they have the option of marketing GK104 as a high-end card. Which is a shame for us, the consumer, as it means a higher pricetag.

Yeah I agree completely. Pretty much because that seems to have happened.

I did read that AMD were having "Fermi type" problems with their 28nm so maybe that was correct?

Or it could just be that AMD did pretty much what they had to to take the lead and nothing more? I guess we will find that out over the coming weeks/months.
 
Last edited:
That's news to me? :confused:

Where did you see / hear / read that?




While it's clear from the architecture that GK104 wasn't originally supposed to be the high-end part, I don't think that Nvidia "had no choice" but to release it as the high-end card. The choice was provided to them by the 'less-than-stellar' performance of the 7970 - giving them the option of competing with their upper mid-range part.

Had the 7970 performed better (or even just been released at 1100Mhz...) I think Nvidia would have gone with the 670Ti label, and given something of a 'paper launch' preview of the GK110. As it is, they have the option of marketing GK104 as a high-end card. Which is a shame for us, the consumer, as it means a higher pricetag.

I can't agree more. This is the cause of it all.

And consequently, while pricing is important as far as purchasing is concerned, there is no reason to be concerned about underperforming tech at all. If anything the tech is performing better than expected.
 
I can't agree more. This is the cause of it all.

And consequently, while pricing is important as far as purchasing is concerned, there is no reason to be concerned about underperforming tech at all. If anything the tech is performing better than expected.

If these benchmarks prove to be accurate then in games the 680 is faster than the 580 by 33,36,10 and 35 percent so an averge of ~29%

How does this compare to previous Nvidia launches (using various review sites at 1080p)?

GTX480 was 41% faster than GTX285 so already this launch is worse than fermi by a large chalk.

GTX280 was 50% faster than 9800GTX.

8800GTX was 70% faster than 7900GTX


So looking like the worst high end Nvidia single GPU in recent memory.

The pricing and the performance increase over the previous high end are the most important metrics. The point straxusii makes still stands though. Unless the reviews show that the GTX680 actually performs much better than in the preview,the performance jump observation still stands.

Making excuses that the tech looks shiny and the die looks tiny,etc means bugger all TBH IMHO for a high end GPU with a price which matches it. It does not matter that it might have been cheaper but it isn't though. That might be good for Nvidia and their profit margins but in the end is of no importance to the consumer. The same goes with AMD and their HD7970,HD7950,HD7870 and HD7770 pricing. Maybe it will be great if you are shareholder in AMD or Nvidia,but I am not so I am going to skip the current generation until the price and performance boost for the price looks better.
 
Last edited:
The pricing and the performance increase over the previous high end are the most important metric.

It depends whether you're looking at it from a consumer perspective, or a technological perspective. It's important to distinguish between the two...


From a consumer perspective this is the high end card. No ifs, no buts - it's the GTX680 with a pricetag to match. From that perspective, it is no more impressive than the 7970 was: A modest improvement in performance over the previous generation, with a high-end pricetag.

From a technological perspective it seems to be a fairly impressive chip from what we've seen so far. Small die size, relatively low power, and able to roughly match the performance of the 7970, using 'only' a 256-bit memory bus. We already know that Nvidia is preparing a larger variant, and from GK104 we can extrapolate a fair amount about how that chip will perform.


If you're only interested in the 'right here and right now', i.e. you're going to buy a new card on March 23rd come what may, then the consumer perspective is the only one you need be concerned about. If you're planning a GPU upgrade more strategically, and/or don't mind waiting six months, then the technological perspective becomes more important. Of course, I fully expect AMD to throw a 28nm GCN refresh into the mix in late 2012 / early 2013 as well!
 
If these benchmarks prove to be accurate then in games the 680 is faster than the 580 by 33,36,10 and 35 percent so an averge of ~29%

How does this compare to previous Nvidia launches (using various review sites at 1080p)?

GTX480 was 41% faster than GTX285 so already this launch is worse than fermi by a large chalk.

GTX280 was 50% faster than 9800GTX.

8800GTX was 70% faster than 7900GTX


So looking like the worst high end Nvidia single GPU in recent memory.

it hasan't even been released yet............... facepalm
 
It depends whether you're looking at it from a consumer perspective, or a technological perspective. It's important to distinguish between the two...


From a consumer perspective this is the high end card. No ifs, no buts - it's the GTX680 with a pricetag to match. From that perspective, it is no more impressive than the 7970 was: A modest improvement in performance over the previous generation, with a high-end pricetag.

From a technological perspective it seems to be a fairly impressive chip from what we've seen so far. Small die size, relatively low power, and able to roughly match the performance of the 7970, using 'only' a 256-bit memory bus. We already know that Nvidia is preparing a larger variant, and from GK104 we can extrapolate a fair amount about how that chip will perform.


If you're only interested in the 'right here and right now', i.e. you're going to buy a new card on March 23rd come what may, then the consumer perspective is the only one you need be concerned about. If you're planning a GPU upgrade more strategically, and/or don't mind waiting six months, then the technological perspective becomes more important. Of course, I fully expect AMD to throw a 28nm GCN refresh into the mix in late 2012 / early 2013 as well!

Well I see it from a consumer perspective. The tech looks very interesting but in the end its all about value for money. For example the HD7870 2GB looks a nice card from a tech perspective - tiny die,good performance and low power consumption. However,it is £270 and there are cards such as the GTX560TI 448 which are now over £70 cheaper and offer most of its performance,and this comes from a SFF PC fan who really likes the lower power draw offered by the HD7870 and the GTX680.

There is also an issue with predicting performance of the larger chips too. We really don't know how well the GK110 will perform relative to the GK104,ie, what sacrifices for gaming performance effiency have been made for other markets. I don't know TBH. We still don't know the full performance of the GTX680 yet!!:p

About your last paragraph too. If the current high end is not enough of an increase over the existing for a given price then just don't upgrade until:
1.)The cards drop in price so the performance boost looks more in line with the price.
2.)The cards are replaced with ones which perform better for a given price.

That is a simple consumer based view. For example look at the GTX480. There are people who skipped this initially,and when the GTX580 was released either bought cut price GTX480 cards or spent their money on full priced GTX580 cards.
 
Last edited:
the consumer will worry about two things

1...... how much more powerful is it than a 7970

2......the price

you can bet your last dollar that it'll be over 400 quid, therefore if it is only 15% faster then it's a waste of money buying.... because the 7970 might only be 300quid by july !....... let alone the 7950, because that's already dropping in price and has 3GB

The best bet looks like the 7950 !!!!!

after all, if i was after muscle right now i'd get the GTX 590.... therefore the additional 8 fps doesn't bother me much, it's the price only....... this new Nvidia card might be seen as another GTX 590, too expensive and a waste of time.

This new card is only interesting if it's cheap and way more powerful than the competition, which it definitely wont be.... Nvidia look ****ed to me

because AMD look ready to drop their prices.... but Nvidia are never cheap on release, i dont think there's enough money around right now to justify a card, that's 150 quid more expensive that the competiotion, but only 8FPS faster...... only a mug would go down this route !

dont forget that i'm going from an 8800GTX up to one of these monsters, so i'm not worried too much about FPS..... SIMPLY BECAUSE I KNOW ANY ONE OF THESE WILL BE HEAVEN compared to what i have now !!!!!!!

i'm only gaming at 35fps..... but this will soon be 100 to 120! so who cares about the odd 8....... money, it's always money !
 
Last edited:
I would still wait until the GTX680 reviews are out to make a final judgement though. For instance we don't know if the GTX670 will be simply an underclocked GTX680(or not). It could explain all the noise about the magic BIOS.
 
Back
Top Bottom