1Ghz 5870 18% faster than GTX480 in Heaven with Tessellation off!

[Mr Fujisawa]

Why haven't they said which card it is? if that's the 470 then fine, but if that's the 480, ouch, it better be 5870 price.

 

[Rroff]

Not sure why Rroff has come up with the idea GT212 was supposed to be up against Evergreen, it wasn't. It was supposed to be out a LONG time ago now, really over a year now,

GT212 would have been up against evergreen and same generation, tho originally it was supposed to be out April/May 2009 then delay til Nov and finally aborted in its high end configuration and all we have are some rubbish low end parts based on GT216.

it was a very minimal die shrink, it wouldn't have been about clock speeds but about making it profitable at the same price. It almost certainly would have been a simply replacing of the GT200 with a more profitable version of the core that was maybe 60% of the size of the GT200b and would have been the midrange card with Fermi above it.

The GT212 wasn't a massively faster core, nor was it designed as anything but a 240 shader design, it was a like for like replacement and theres no way in hell, not a single tiny chance that Nvidia planned such a massive speed bump on GT212 that a card with the same number of shaders would be considered "next gen".

Your thinking of the GT216 - the 212 was originally specced for 384SP on the high end GTX part, with an "ultra" 480SP part and a 384SPx2 GX2 part on the cards to be released later when the ATI 5 series launched. The GT216 was to be the respun more economical/better profit 200 series replacement.

If Rroff seriously thinks Nvidia meant a 240 shader GT212 was supposed to be the next gen card and they pluked Fermi out of a different idea and shoehorned it in........ well.

Do you actually know anything about GT212 at all? (I'm not saying I'm an expert on it or my info is entirely correct - its just very different information to what I've heard - and my source for the information formely worked for nVidia)

 

[Ricardox1]

It wont beat the 5970

 

[molemole0]

impressive

 

[Trunks9486]

The tragic thing is that all comparisons are made with the 5870. Sure they are both single GPU but ultimately Fermi was supposed to deliver the crown back to nvidia, and as it seems now it will barely out perform Cypress.

5970 is going to murder this thing and is already available. Not to mention that the rest of the 58xx range can get a price cut if needed.

 

[RavenXXX2]

Was looking to purchase a 480 but now it looks like another 5870 is the way forward..

 

[Locky]

Get another 5870,

I can't see fermi being anything decent, I was hoping for another 8800 case for the industry to move on but it doesn't look like the case tbh.

 

[nickrose]

Get another 5870,

I can't see fermi being anything decent, I was hoping for another 8800 case for the industry to move on but it doesn't look like the case tbh.

+1

You can count on one hand with 3 fingers missing the number of great cards Nvidia have produced of late.

Gtx280 and 8800 series.

And I used to be a Nvidia fan, until the 5870's came along.

Cannot see fermi being much better.

 

[Rroff]

Its not like ATI have made any more great cards of late...

8800GTX, 8800GT/GTS250, GTX260, GTX285/295.
4850, 4870/4890, 5770, 5850/5870/5970.

Tri-SLI GTX480 would need some serious cooling but not much will touch it for performance

 

[Pendu]

Rroff mate you need to stop making baseless predictions. You knowing scaling is no way near perfect in either SLI or Crossfire, Tri-SLI for 480GTX could be no better than 2x 5970.

 

[Rroff]

Even if we assume I know nothing more than the man on the street its hardly a baseless prediction... if we look at 2x 4870X2 V 285GTX tri-SLI benchmarks it would give us a good indication - sans CPU power - of where the gains will stack up. multiGPU techniques like AFR don't magically gain or lose efficency over different generations of cards that are based around the same type of shader/core setup and the performance delta is similiar.

 

[Orangey]

Tri-SLI GTX480 would need some serious cooling

Chunky Delta fan feeding into a custom wind tunnel duct to be precise.

Great idea, shame it's out of necessity though.

 

[mmj_uk]

5970 is going to murder this thing and is already available.

No kidding? that's like saying Phenom X4 will murder Phenom X2.

 

[Neil79]

All this talk of SLI GTX260 makes me wonder if i should get another lol

 

[maxilive]

I don't really read much about all this ATI vs Nvidea jazz but if any of these numbers are close to how the 480 is going to preform then thats disappointing. I think the 5870 came out in October? So 4-5 Months down the line and still not much improvement?

Saying that I suppose the 5870 has seen a 10% increase etc in some games with all these driver updates. I hope the Nvidea card is actually good at some things though, for example better crossfire scaling, better performance with triple screen setups or generally more consistent framerate (5870 is pretty good for that already).

If not, then oh dear Nvidea have got some work to do!

 

[arc@css]

Its not like ATI have made any more great cards of late...

8800GTX, 8800GT/GTS250, GTX260, GTX285/295.
4850, 4870/4890, 5770, 5850/5870/5970.

Tri-SLI GTX480 would need some serious cooling but not much will touch it for performance

I disagree there.

I would say the 5870 / 5970 are great cards, as are some of the nvidia ones there that you have listed but to be fair it has been a long while since their release. Regarding the ATI 5000 series, impressive performance by all reviews and accounts of most users. A few teething troubles granted, some even major ones but that is to be expected with new tech.

Hope the new nvidia cards are not too steeply priced and peform comparitively well to the 5000 series.

 

[johny dolan]

If fermi can only match the performance of the ati cards that have been out already 6 months then that's very disappointing indeed.

If the performance is same but Fermi uses much more watts and runs much hotter and costs more to buy then its a massive fail.

 

[Rroff]

All this talk of SLI GTX260 makes me wonder if i should get another lol

If you can find one cheap and play at 1920x or above its deffinatly quite a performance boost a bit late in the day now tho for starting into 200 series SLI. Crysis aside (where performance is closer to an OC'd 5850) my overclocked GTX260 SLI will match or beat an overclocked 5870 in DX9 and 10 games.

 

[drunkenmaster]

Even if we assume I know nothing more than the man on the street its hardly a baseless prediction... if we look at 2x 4870X2 V 285GTX tri-SLI benchmarks it would give us a good indication - sans CPU power - of where the gains will stack up. multiGPU techniques like AFR don't magically gain or lose efficency over different generations of cards that are based around the same type of shader/core setup and the performance delta is similiar.

THe problem being of course that a 285gtx wasn't about the same speed asa 4870, it wasn't really close, it was more than often 20-25% ahead, so 3 cards vs 2x4870x2's isn't surprising.

The problem is, you're just acting as if the performance gap between the 4870/gtx285 is maintained when comparing a 5870/480GTX, it almost certainly won't be, and even worse the chances of anyone being able to buy 3x 480gtx's are slim to none, 3x470gtx is more likely but then you lose even more performance.

Its very likely 2 x5970 will trounce 3x 480gtx's and utterly destroy 3x 470gtx's.

As for the GT212, it makes NO sense in any way at all, the fact that GT212 was NEVER going to be a dx11 part means wasting resources turning it into some grandious 480SP ultra part is utter nonsense, I've seen ONE rumour about it being 384sp's all stemming from the same source, one random website news story with nothing else in 6 months before or after backing it up.

THe GT212 was just so not going to be some uber part, and Fermi was always their DX11 part, and they taped it out middle of last year, which means they most certainly did plan to have it out last year, they just failed. You don't make a 480 or 384sp part with dx10.1, that radically different at a massive design cost thats supposed to be the high end card for at most, 5 months before you replace it with Fermi and make it too expensive to compete in the midrange and too slow to compete with the 5870/GT300 cores. Likewise, what company would produce a 285GTX, build a FERMi thats going to be about 80% faster, and then in between for a ridiculously short time build a part that will be 50% faster than the 285gtx, by doing so you'd simply make Fermi look like a pathetically small increase and make it seem like a poor upgrade.

So other than the fact I've only seen a single very dodgey rumour suggesting it was anything more than a shrink of a GT200b, it makes no business sense at all, would have ended up being a ridiculously short lived high end part at a high cost that offered no benefits at all.

Remember at that point the 285gtx was faster than a 4890, and made a profit by costing more, the 260GTX wasn't faster than a 4890 and made no profit, Fermi was slated to be out around the same time as AMD's next gen. Why would they make an expensive large core part that would replace a high profit high end part for no reason? Answer is, they wouldn't. They wanted to replace the entire line up top to bottom with cheaper to make parts because the 285gtx was low sales anyway, the highest end is, it was the 260gtx and below suffering on cost/margins and those were the high volume parts, theres no reason to spend money and time improving a high profit part in the 285gtx and every single reason to replace the low profit parts.

I would say theres a 3% chance even Nvidia were stupid enough to make a card that would offer 80% of Fermi's performance, to be sold over a 4 month period, then be discontinued at ridiculous cost. But then Nvidia are a company stupid enough to do that, it still doesn't mean the GT212 was intended to compete with a 5870, there is 0% chance Nvidia wanted the gt212 to compete with Evergreen, AMD had a DX11 part out in 2009, Nvidia tried their hardest to get a DX11 part out in 2009, to pretend Nvidia didn't intend them to compete is nothing but crap, while theres a slim chance Nvidia were dumb enough to make the gt212 overpowered, they were not dumb enough to try to launch a DX11 part and leave it as Tesla/GPGPU only and not for it to compete with AMD's DX11 part.

I'm not sure how you can suggest otherwise while keeping a straight face. Its basically the same as suggesting AMD intended the X1950pro to compete with the 8800gtx........... utter crap basically.

 

[johny dolan]

And how much would 3x 480gtx's cost even if it was possible to buy 3.