GTX 480 and GTX 470 Reviews

[zalbard]

So with increase in resolution the ATI's card performance actually increases instead of decreasing??

Must be just some mistake.

[LightSpeed]

But its a fact that ATI's competitiveness increases with resolution. All the review sites either said it, or its very apparent in their graphs. And i dont mean raw FPS, i mean percentage wise the 5870 closes the gap with the 480 the higher the resolution goes

[SKYMTL]

I find this very weird

So with increase in resolution the ATI's card performance actually increases instead of decreasing??

To me it looks like they only ran one benchmark run and left it at that. The first time through the Stone Giant benchmark will always lead to lower performance due to a buffering issue or something. The benchmark needs to be run twice or three times for every change in settings in order to get accurate results.

[Chickenfeed]

I can't confirm their 1280x800 results however I can confirm their 5870 numbers at 1920x1200. Do note that diffusional depth of field doesn't work on ATI hardware using the current press release beta build (as far as its effect of performance, that is unknown ). A final public version of the demo should be released in the next few weeks. Their engine uses PhysX for what its worth. None the less this further cemments GF100s prowess at tessallation.

EDIT: Their 1280x800 ATI results are a bug on their end. A 5870 gets around 50fps avg at this resolution.

[annihilat0r]

Don't any of you guys find it strange that before the reviews, there were 2560 (plus some AA) benchies everywhere that showed a massive Fermi lead, and that everyone had attributed it to the extra memory? Now we have the reviews and the situation is actually the opposite.

On a irrelevant note, I think GTX 400 series' performance in minimum FPS numbers shouldn't be missed. 15% over 5870 doesn't mean much, but even higher min framerates are in the equation, and I personally care more about the min FPS than the max FPS.

[jam2k]

[Manicdan]

^nice, im actually laughing out loud, but it should also put on a george forman grill for comparison

[Helloworld_98]

I wonder what temps you'd get at idle with a passive 480...

[japamd]

Don't any of you guys find it strange that before the reviews, there were 2560 (plus some AA) benchies everywhere that showed a massive Fermi lead, and that everyone had attributed it to the extra memory? Now we have the reviews and the situation is actually the opposite.

Maybe GF100 ROPs are bottlenecking it as GF100 still process only 32 pixels per clock which is G200' same rate, while having twice more stream processors and even more shading power. Please someone correct me if I'm wrong.

On a irrelevant note, I think GTX 400 series' performance in minimum FPS numbers shouldn't be missed. 15% over 5870 doesn't mean much, but even higher min framerates are in the equation, and I personally care more about the min FPS than the max FPS.

That's great indeed, but without a graph showing how the frame rates are fluctuating it's hard to say if it's good or if it's only a quick disk hiccup or something like that. If sustained min FPS are better on GF100 that would be a very nice improvement, as you said, better than higher avg FPS as they're already on top of roof for most games.

http://img36.imageshack.us/img36/1957/gpuo.png

[Chickenfeed]

That's great indeed, but without a graph showing how the frame rates are fluctuating it's hard to say if it's good or if it's only a quick disk hiccup or something like that. If sustained min FPS are better on GF100 that would be a very nice improvement

Exactly what I've been thinking. It would be nice to see more proper graphs and not merely a static min avg max type deal.

[damha]

New benchmark!

[damha]

This just in, new information from Wikipedia:



So thats what it was. Blasted Fermilab.

[Manicdan]

i dont keep up on everything, who wins in single gpu on LN2, 5870 or 480?

[snoro]

480 on ln2 win at about every benchmark for the moment except vantage( might be some driver issue) and 01 which have not been tested by anyone for the moment.For 03, 05 and 06 a stock gtx480 can beat a overclocked hd5870 (950core,1300mem).

[LiquidFiction]

i dont keep up on everything, who wins in single gpu on LN2, 5870 or 480?

5870 By a LONGSHOT

[Power5]

Nvidia also needs to add a this sign to the heat sink to keep from getting sued:

Same reason they can't put this sticker on the HS is the reason they could not have a full cover shroud. Early testing with dual card setups revealed that the upper 480 became fused to the lower 480 when the shroud melted and dripped onto the lower PCB like candle wax. When turned off, the two become welded together. Instead they chose to cut a hole in the hood.

I read through most of this thread and must of missed it, but what fan speed are the temps recorded at?

Our server is also getting hammered a lot. This is just craaazy people!!!

http://lab501.ro/placi-video/nvidia-...imal-magnetism

Guess what....

That is a classified, there is plenty of room for another card in there.

[Russian]

I love how almost every single GTX 480/470 review in the world is posted on that list... except one. hahahaha

[Sushi Warrior]

5870 By a LONGSHOT

shamino hit 1150 already, and the card isn't even out. I would go with a 5870 now, but we haven't even SEEN the potential of Fermi (and its wide, leaky, channels).

[Manicdan]

what the highest clocks the 5870 got?

[zalbard]

what the highest clocks the 5870 got?

Something around 1450.

[Monstru]

manicdan - GTX 480, from what I can see now, will reign 2k3/2k5/2k6 tops, once GPU gets over 1000MHz. Doesn't matter if ATI is at 1500MHz. In Vantage however, HD5870 kicks ass!!!

[zalbard]

Quite an interesting "review" - check it out.

[jam2k]

[noroger]

hitler reviews fermi:
http://www.youtube.com/watch?v=VpZXhR1ibj8

[insurgent]

I'm wondering why there are no fermi reviews for tech report, anyone know?