Early Fermi Benchmarks

But not by much, which is impressive considering its a single GPU card versus a dual GPU solution.

And the 5870 almost always has better minimum frame IIRC, which is far more important that peak fps.
 
But the GTX 285 didn't. The 275 did, but it was kind of a gimped card. I was expecting the GTX 480 to have an 8-pin and a 6-pin.

That was a picture of the 480 wasn't it? Or was I looking at a picture of the lesser card?
The GTX 280 has a 8 and 6 pin connector. The 285 has two 6 pin connectors.
 
I wouldn't hold my breath.

With all the problems they had trying to ship the current batch the 2 gpu on one card is another way of burning money like they are losing money on their single cards now as well. Not to forget the ever so important PCI-E spec which wont allow such a card with the present specs.

They can write what they want at fudzilla it is not going to happen.
 
http://www.fudzilla.com/content/view/18038/1/

See what they can do in that period of time to get under the PCI-e spec. By the time the dual card comes out, they'll probably have 475 and 485.

Either they manage to massively reduce leakage in a full respin or they wait for PCI-Express 3.0, which maybe has more power headroom. If anybody knows more about PCI-E 3, I'd love a link. Wikipedia doesn't mention power.
 
always got an excuse eh? just eat it up for once =p unless Nvidia pays you, you have no reason to be this loyal to a brand name

Actually, I think he's right. CoP is basically just Stalker CS with a few changes and additions to make it utilize some DX11 features. It isn't a showcase for DX11 at all. Using CoP as an example of what ATI and DX11 can do is somewhat irrelevant.
 
The GeForce 470/480 needs to perform well in current, released games like Bad Company 2 and AvP. This thing isn't going to win any awards based on performance in future games.


Bingo!!

In fact, Fermi's performance in Battlefield Bad Company 2 will make me a repeat customer of Nvidia or a new customer of ATI.
 
Bingo!!

In fact, Fermi's performance in Battlefield Bad Company 2 will make me a repeat customer of Nvidia or a new customer of ATI.

Couldn't agree more. BC2 is making my system bleed. I'm worried my q6600@3ghz isn't enough as I seem to be cpu limited even with my 8800gt's
 
The way I read it, was the card seems to have some torque, but relatively low horsepower. Just like it's younger brother(8800gtx), it seems to chug a long at many different resolutions and many different game qualities. Problem is, the 8800gtx brought the horsepower and the torque. I wonder if Charlie was right on this card. Because of the low pass rate, parts of the card were disabled. Maybe in another year, when the failure rate and pass rate are flipped, will we see a refresh of this card that is what Fermi was supposed to be.

Then again, I don't believe anything until Kyle has it in his hands and shows us "real world gameplay advantages"
 
Couldn't agree more. BC2 is making my system bleed. I'm worried my q6600@3ghz isn't enough as I seem to be cpu limited even with my 8800gt's

For what it's worth, my system in my signature below has trouble with BFBC2 at 2560x1600 with 4x FSAA and 8xAF and my GTX 285. Hence one of the reasons I use my Acer 23" H233H (1920x1080 native).

If I were you, I would go with the videocard first and see if that helps. Can you push your processor to 3.4 or thereabouts?
 
Those Crysis benchmarks are funny, How they underline the minimum FPS.

The stutter to sub 5 fps on the 5800 based card is most likely caused by the lower amount of VRAM, swapping textures. Especially at those high resolutions when you crank up the AA.
These stutters have been well documented and will be eliminated with the 2GB 5870.
 
i think that depends what games you're talking about.

if you talking about dx9 games then im sure that the 5970 will be much faster but dx9 is history to be honest.

Game developers and console manufacturers would beg to differ. DX9 has a few more years in it according to them. However, it's long since past time to move on.

Methinks a $299-349 GTX 470 is in my future...
 
Yeah, looks like the 5870 easily has the lead in most games( not FC2 though, but we know that's heavily nvidia biased ) *until* it runs out of vram. Ironically, most of these 'benchmarks' appear to be run at very high resolutions with tons of AA, as if they purposely tried to get the 5870 to run out of steam.

While the results are still relevant, it would be far more interesting to see benchmarks where vram wasn't an issue for either card.
 
bh8r9.jpg

Looks like nVidia included some bling bling diamonds on these cards -- no wonder they're so expensive. I'm going to rip the diamonds off my card and sell them at a profit.

1.) Buy Fermis
2.) Rip diamonds off
3.) PROFIT!
 
Bingo!!

In fact, Fermi's performance in Battlefield Bad Company 2 will make me a repeat customer of Nvidia or a new customer of ATI.

I couldn't have said it better myself. I really hope NV pulls a dildo out of my pants


edit: I meant to say a rabbit of their hat
 
Last edited:
I couldn't have said it better myself. I really hope NV pulls a dildo out of my pants


edit: I meant to say a rabbit of their hat

I can understand the relationship between a hat with pants, both are items of clothing. But a rabbit and a dildo?:confused:

Not getting on you, you gave me a good belly laugh!:D
 
now I am wondering. wouldnt it be funny to make a video card that goes like
300 fps every 1 second to 0 fps 2 second. average will be 100 fps but you will feel a pause 2 seconds.

This is why, in general, the average frame rate is a bad number to go by. I like the reviews here because they give graphs of frame rate vs time (or point in time within a game) and a subjective impression of playability. This is much better in terms of knowing how well a card will actually perform on a game then something like average frame rate.

At the very least, find the expectation value, aka the mean or average, of the distribution and then remove all data that is more or less than one standard deviation from the mean. Then find the average of the remaining data and report that. Thus, extreme values will not skew the average one way or the other. Of course I think minimum frame rates are important too because even a seconds of stuttering frames will have a negative impact on game experience.

But giving a graph is better. For total awesomeness the data should be loaded in an xml file or an excel file and open for downloads. That would let people with a more mathematical inclination do their own analysis. But that might be asking a bit too much.:p
 
5970 with BC2
2560x1600
everything on highest,
2x msaa

average i'd say is about 85-90fps


What are your other specs?

The 5970 is interesting, but it has been real difficult to find one and they are super expensive. I certainly don't mind paying a premium though for performance.
 
What are your other specs?

The 5970 is interesting, but it has been real difficult to find one and they are super expensive. I certainly don't mind paying a premium though for performance.

how are they super expensive?

they are cheaper than a 2 x 5870 with binned GPU, I really don't think its that expensive...
 
we'll see what you think of the 480's price, you'll probably find some way to justify it costing nearly as much as a 5970 while being slower =p

the fastest single gpu card on the market can cost as much the company wants to charge

the fastest dual gpu card on the market can cost as much the company wants to charge


that's just how it is, and always has been.
 
the fastest single gpu card on the market can cost as much the company wants to charge

the fastest dual gpu card on the market can cost as much the company wants to charge


that's just how it is, and always has been.

Unless its ATI doing the charging, then it's too much right? Don't hold back now.
 
I can understand the relationship between a hat with pants, both are items of clothing. But a rabbit and a dildo?:confused:

Not getting on you, you gave me a good belly laugh!:D

Yea... exactly. I would have went for a hamster reference here myself :eek:

I think his point was nVidia will have to come with some serious performance if they want to sell any cards -- a point I don't necessarily agree with.

nV will do fine on sales, even if they don't kill in performance and are overpriced. Fanboys will buy them and others will buy them for their supposed unique features (tessellation, CUDA, Physx). If they are able to one-up ATI and make using three screens easier and better scaling for multi-screen gaming (without a $100 dongle), they'll sell even more. Isn't SLI supposed to work with multi-screen gaming with the same scaling as a single screen out of the box? I might buy a pair of 480s in this case.
 
Unless its ATI doing the charging, then it's too much right? Don't hold back now.

i never said it that way,

i said it as in general that 700 dollars is alot for a video card, regardess if its from nvidia or ati.

maybe we should turn down the fanboy radar down a bit, shall we?
 
5970 with BC2
2560x1600
everything on highest,
2x msaa

average i'd say is about 85-90fps

That's about what I'm seeing. I've got the same card and I'm using the same settings.

$700 dollars for a video card is not expensive?:confused:

They aren't $700 if you shop around. I picked up mine at Microcenter for $599.99. Compared to dual 5870's in Crossfire I'd call the 5970 a solid value.
 
DualOwn said:
the fastest single gpu card on the market can cost as much the company wants to charge

the fastest dual gpu card on the market can cost as much the company wants to charge


that's just how it is, and always has been. And i think the logic you have on 480 vs 5970 is completely off.

When 2 items are priced similarly, they are expected to perform the same. It doesn't matter if it's single or dual gpu.
 
the fastest single gpu card on the market can cost as much the company wants to charge

the fastest dual gpu card on the market can cost as much the company wants to charge


that's just how it is, and always has been.

so if the company (ati) charges $699 it's too much, but if NV charges that much for their card, it's not ?

480 vs 5970 makes perfect sense, if you want the fastest possible card, and the 480 can't beat the 5970, then the 5970 is still the fastest card.

Nvidias fastest offering (480) vs ATIs fastest offering(5970) makes sense, to everyone except for fanboys who are in denial. It was the same deal with the 7950gx2 vs 1950xtx, the 1950xtx smacked the 7800gtx around, and even the 7900, but the 7950gx2 was faster(even though drivers were horrible) it was the single fastest card you could get.

The days of fastest single GPU are slowly coming to an end,
 
If they are able to one-up ATI and make using three screens easier and better scaling for multi-screen gaming (without a $100 dongle), they'll sell even more. Isn't SLI supposed to work with multi-screen gaming with the same scaling as a single screen out of the box? I might buy a pair of 480s in this case.

Features become more important when Nvidia does it.
Having to buy two cards for three monitors hurts the customer because most will only be able to buy one card at launch.
ATI users with Eyefinity setups will probably pass up on Fermi because of this.
 
so if the company (ati) charges $699 it's too much, but if NV charges that much for their card, it's not ?

480 vs 5970 makes perfect sense, if you want the fastest possible card, and the 480 can't beat the 5970, then the 5970 is still the fastest card.

Nvidias fastest offering (480) vs ATIs fastest offering(5970) makes sense, to everyone except for fanboys who are in denial. It was the same deal with the 7950gx2 vs 1950xtx, the 1950xtx smacked the 7800gtx around, and even the 7900, but the 7950gx2 was faster(even though drivers were horrible) it was the single fastest card you could get.

The days of fastest single GPU are slowly coming to an end,

i know what you trying to say but it cant be changed the way you want it to look like.

there is always going to be the fastest single gpu and the fastest dual gpu on the market.
 
i know what you trying to say but it cant be changed the way you want it to look like.

there is always going to be the fastest single gpu and the fastest dual gpu on the market.

You're the one changing your tone. When you were talking about ATI you said it was too much. When you replied to an nVidia comment you toned down your rhetoric and said "a company can charge whatever it wants to"
 
Yeah, looks like the 5870 easily has the lead in most games( not FC2 though, but we know that's heavily nvidia biased ) *until* it runs out of vram. Ironically, most of these 'benchmarks' appear to be run at very high resolutions with tons of AA, as if they purposely tried to get the 5870 to run out of steam.

While the results are still relevant, it would be far more interesting to see benchmarks where vram wasn't an issue for either card.

I think its pretty fair, split second drops to low frame rates don't affect the average by much at all. Its just funny to me that they have to point it out specifically.
 
Back
Top