Don't get me wrong, I do contribute to charities and am good with my mates. But personally I'd rather do what I do with graphics cards and be generous in other ways. Horses for courses.
Last edited:
-
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.
Quick Question Regarding nVidia's Next Gen...
- Thread starter Overslop
- Start date
Don't get me wrong, I do contribute to charities and am good with my mates. But personally I'd rather do what I do with graphics cards and be generous in other ways. Horses for courses.
Soldato
I don't contribute to charities, im not really into the 3rd person charity.
I personally don't like upgrading unless i need to on some components or feel that i could do with more performance or if there is a new feature that i want.
New performance just being there does not compel me.
Last edited:
Same with me. I always feel like I could do with more performance though!
2 GTX 480s are brilliant of course, but there is always room for improvement. I game on a 40" 1080p screen (large pixels) so the more AA the better. There are quite a few titles that even my setup can't run perfectly. If I'm not getting a perfect 60FPS V-Sync then I turn down settings until I do. Upgrading means fewer settings needing to be reduced. As much as that though, I can just sell and do without a high end card for a few weeks...then buy back the same type of cards/next gen for a fair amount less!
Soldato
Same with me. I always feel like I could do with more performance though!
2 GTX 480s are brilliant of course, but there is always room for improvement. I game on a 40" 1080p screen (large pixels) so the more AA the better. There are quite a few titles that even my setup can't run perfectly. If I'm not getting a perfect 60FPS V-Sync then I turn down settings until I do. Upgrading means fewer settings needing to be reduced. As much as that though, I can just sell and do without a high end card for a few weeks...then buy back the same type of cards/next gen for a fair amount less!
What! No 1080p 60fps maxxed out needs an upgrade pronto, you should have said earlier
.What! No 1080p 60fps maxxed out needs an upgrade pronto, you should have said earlier
Associate
- Joined
- 2 Sep 2010
- Posts
- 350
this has answered my question .
.this has answered my question
Which is?
Yeah atm it would be 5 and a half months minimum if a test run went out tomorrow.
I've got a weird feeling tho we may see a refresh about 4 months before Kepler
it makes absolutely no sense but there we are. Probably see a shuffling of the lineup before that as well to try and stimulate sales.
Last edited:
so you don't think that Nvidia will release a full fat gf104,(edit) or just a clocked one to give us a 475 and then a possible full fat gf100 to give us a 485.
even if they are only mediocre increases over the 470/480 it will give them something to go up against the new 6870 and the 6950/70
even if they are only mediocre increases over the 470/480 it will give them something to go up against the new 6870 and the 6950/70
Last edited:
Soldato
I don't think we will see a full fat GF100, maybe a marketing stunt but even so, there would be little point as the card would still be irrelevant if it isn't the fasted single GPU.
Who knows though, maybe they can get the power consumption down enough for it to be viable.
May be we could get a full fat GF104 if yields will allow, and there isn't any major increase in heat similar to GF100.
As above - we won't see a "full fat" 106 it was never designed for that direction and doubtful on a GF100 - possibly a very limited cherry picked run to get the "headline" lead over ATI.
I'm thinking we might see slightly tweaked, overclocked cards based on the 104 tho.
I'm thinking we might see slightly tweaked, overclocked cards based on the 104 tho.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
The problem is, a less than 10% shader bump on the 480gtx isn't going to do much, infact with the same fairly low bandwidth on Nvidia Fermi's it would likely be a bit more bandwidth starved than a 480gtx so would likely not scale well.
Though end users are overclock memory ok, Nvidia's seemed very reluctant to up memory speeds, though maybe it kept them down on the gf104 just to keep it further away from a 470gtx and leave a larger amount of performance for a 384shader version.
The thing is, a 475gtx in the past 3 months could have made Nvidia a lot of money. Instead of £200 470gtx's that lose a bucketload, they could have been selling 15% faster full GF104's for maybe £200, and making more profit, instead they've stuck to three minimal profit/massive loss making cards in the two 460's and 470 cards.
The last three months, when AMD had nothing between £120 and £200 worth buying, is the time Nvidia most likely would have capitalised on a full GF104 IF they could.
WHich really does suggest they can't make full gf104's, which is pretty shockingly bad.
Why with a awful yield having issues gf100 they decreased the granularity by increasing cluster size I don't know. Whatever they are doing at 40nm, they are doing wrong with lots of failed parts on a waifer, going from 32 shaders on the GF100 to 16, or staying at 32 on the GF104 would likely have increased yields and increased the amount of shaders they could release with a decent yield.
At this point I really doubt we'll see a 384 shader GF104 or a 512 shader GF100 except in "extreme stunt" quantities, IE a launch, and a few dozen around the world, then nothing.
I've been saying it for 2 years now, what we need is Nvidia with a small core strategy/design, so they can actually compete on price with AMD. At the moment AMD aren't making huge margins on anything but the 5870, the 5850 is pretty expensive to make but I'm sure could come down another £15, theres just no need to, Nvidia can't go lower than it has now and with the GF104 can't come close to matching 5850 performance.
If AMD manage to push similar performance from a 336mm2 5850 core, into a circa 200mm2 core(i'm not sure that is the case, but looking increasingly likely) then Nvidia are quite literally boned. They can sell a 200mm2 core in a £150 with a healthy profit and beat 460gtx performance easily, while Nvidia would make a decent loss at £150 for something 20-30% slower.
Though end users are overclock memory ok, Nvidia's seemed very reluctant to up memory speeds, though maybe it kept them down on the gf104 just to keep it further away from a 470gtx and leave a larger amount of performance for a 384shader version.
The thing is, a 475gtx in the past 3 months could have made Nvidia a lot of money. Instead of £200 470gtx's that lose a bucketload, they could have been selling 15% faster full GF104's for maybe £200, and making more profit, instead they've stuck to three minimal profit/massive loss making cards in the two 460's and 470 cards.
The last three months, when AMD had nothing between £120 and £200 worth buying, is the time Nvidia most likely would have capitalised on a full GF104 IF they could.
WHich really does suggest they can't make full gf104's, which is pretty shockingly bad.
Why with a awful yield having issues gf100 they decreased the granularity by increasing cluster size I don't know. Whatever they are doing at 40nm, they are doing wrong with lots of failed parts on a waifer, going from 32 shaders on the GF100 to 16, or staying at 32 on the GF104 would likely have increased yields and increased the amount of shaders they could release with a decent yield.
At this point I really doubt we'll see a 384 shader GF104 or a 512 shader GF100 except in "extreme stunt" quantities, IE a launch, and a few dozen around the world, then nothing.
I've been saying it for 2 years now, what we need is Nvidia with a small core strategy/design, so they can actually compete on price with AMD. At the moment AMD aren't making huge margins on anything but the 5870, the 5850 is pretty expensive to make but I'm sure could come down another £15, theres just no need to, Nvidia can't go lower than it has now and with the GF104 can't come close to matching 5850 performance.
If AMD manage to push similar performance from a 336mm2 5850 core, into a circa 200mm2 core(i'm not sure that is the case, but looking increasingly likely) then Nvidia are quite literally boned. They can sell a 200mm2 core in a £150 with a healthy profit and beat 460gtx performance easily, while Nvidia would make a decent loss at £150 for something 20-30% slower.
I'm pretty confident that Nvidia will milk their Fermis till early 2012, then they will come out with cards with mahoosive TDP and the world will blow up into trillions of pieces in an incredible fireworks show at launch. For great justice.
They could always re badge the GTX 460 to a GTX 580 and try to sell a few more..Worked that well before it looks like ATI are considering it if some of the rumours are to be believed.
ATI is not considering to re-brand any of their mainstream graphics. It's the low-end that gets re-branded and there might be several reasons why you'd want to sell the good ol' product while maintaining it recognizable on the market. Think AMD EOLing ATI brand and stopping production of Cypress cores (already happened), their new advertising campaign we've yet got to see and then the popularity of Juniper cores on the budget market and with companies like Dell and Apple. It has nothing to do with confusing unaware customers.
Soldato
Yep if below is true, looks like Nvidia margins are screwed, IMO Fermi will soon be in limited production with the few that are produced (or remaining), priced out of the market.
Soldato
171.09 * 134.61 = 230.30 ?
That's some epic maths there...
I'm guessing that the dimensions are supposed to be 17.109, 13.461, 18.418 and 18.349. In this case you get 230mm^2 for Barts. Since Caymann is ~50% more GPU than Barts, and there is a firm division into modular units, you can look at around 345mm^2 for Caymann.
Regarding the next-gen nvidia GPU, this is Kepler, and is scheduled for release in the second half of 2011. I would not expect the same level of delays for Kepler as were associated with Fermi, since the largest part of the architectural redesign has already taken place with Fermi.
No idea whether we will see a Fermi refresh beforehand. I would doubt it though.
That's some epic maths there...
I'm guessing that the dimensions are supposed to be 17.109, 13.461, 18.418 and 18.349. In this case you get 230mm^2 for Barts. Since Caymann is ~50% more GPU than Barts, and there is a firm division into modular units, you can look at around 345mm^2 for Caymann.
Regarding the next-gen nvidia GPU, this is Kepler, and is scheduled for release in the second half of 2011. I would not expect the same level of delays for Kepler as were associated with Fermi, since the largest part of the architectural redesign has already taken place with Fermi.
No idea whether we will see a Fermi refresh beforehand. I would doubt it though.
Soldato
If that's the case then it shows just how shockingly superior AMD's architecture really is considering it's a smaller design than G104 yet not only beats it, but also beats the much larger GF100 as well!
Nvidia NEED Kepler to not only arrive on time, but to be competitive as an architecture...