Sign in with
Sign up | Sign in
Your question
Closed

Nvidia Fermi GF100 Benchmarks (GTX470 & GTX480)

Last response: in Graphics & Displays
Share
a b U Graphics card
January 20, 2010 2:16:33 AM

http://www.tomshardware.com/reviews/geforce-gtx-480,258...


"Crysis is perhaps the closest thing to a synthetic in our real-world suite. After all, it’s two and a half years old. Nevertheless, it’s still one of the most demanding titles we can bring to bear against a modern graphics subsystem. Optimized for DirectX 10, older cards like ATI’s Radeon HD 4870 X2 are still capable of putting up a fight in Crysis.

It should come as no shock that the Radeon HD 5970 clinches a first-place finish in all three resolutions. All three of the Radeon HD 5000-series boards we’re testing demonstrate modest performance hits with anti-aliasing applied, with the exception of the dual-GPU 5970 at 2560x1600, which falls off rapidly.

Nvidia’s new GeForce GTX 480 starts off strong, roughly matching the performance of the company’s GeForce GTX 295, but is slowly passed by the previous-gen flagship. Throughout testing, the GTX 480 does maintain better anti-aliased performance, though. Meanwhile, Nvidia’s GeForce GTX 470 is generally outperformed by the Radeon HD 5850, winning only at 2560x1600 with AA applied (though it’s an unplayable configuration, anyway)"




"We’ve long considered Call of Duty to be a processor-bound title, since its graphics aren’t terribly demanding (similar to Left 4 Dead in that way). However, with a Core i7-980X under the hood, there’s ample room for these cards to breathe a bit.

Nvidia’s GeForce GTX 480 takes an early lead, but drops a position with each successive resolution increase, eventually landing in third place at 2560x1600 behind ATI’s Radeon HD 5970 and its own GeForce GTX 295. Still, that’s an impressive showing in light of the previous metric that might have suggested otherwise. Right out of the gate, GTX 480 looks like more of a contender for AMD's Radeon HD 5970 than the single-GPU 5870.

Perhaps the most compelling performer is the GeForce GTX 470, though, which goes heads-up against the Radeon HD 5870, losing out only at 2560x1600 with and without anti-aliasing turned on.

And while you can’t buy them anymore, it’s interesting to note that anyone running a Radeon HD 4870 X2 is still in very solid shape; the card holds up incredibly well in Call of Duty, right up to 2560x1600."



It becomes evident that the GTX470 performs maybe 10% or less better than the 5850 on average, and the GTX480 performs maybe 10% or less better than the 5870 on average. Yet the power consumption of a GTX470 is higher than a 5870, and the GTX480 consumes as much power as a 5970.

The Fermi generation is an improvement on the GTX200 architecture, but compared to the ATI HD 5x00 series, it seems like a boat load of fail... =/



------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Original Topic:

http://www.hexus.net/content/item.php?item=21996


The benchmark is in FarCry2, which is an Nvidia favoring game.


==Nvidia GTX285==

...What you see here is that the built-in DX10 benchmark was run at 1,920x1,200 at the ultra-high-quality preset and with 4x AA. The GeForce GTX 285 returns an average frame-rate of 50.32fps with a maximum of 73.13fps and minimum of 38.4fps. In short, it provides playable settings with lots of eye candy.

==Nvidia Fermi GF100==
Take a closer look at the picture and you will be able to confirm that the settings are the same as the GTX 285's. Here, though, the Fermi card returns an average frame-rate of 84.05fps with a maximum of 126.20fps and a minimum of 64.6fps. The minimum frame-rate is higher than the GTX 285's average, and the 67 per cent increase in average frame-rate is significant...

==Lightly Overclocked ATI 5870==
The results show that, with the same settings, the card scores an average frame-rate of 65.84fps with a maximum of 136.47fps (we can kind of ignore this as it's the first frame) and a minimum of 40.40fps - rising to 48.40fps on the highest of three runs.

==5970==
Average frame-rate increases to 99.79fps with the dual-GPU card, beating out Fermi handily. Maximum frame-rate is 133.52fps and minimum is 76.42fps. It's hard to beat the sheer grunt of AMD's finest, clearly.

Even after taking into account the Nvidia-favoring of FarCry2, the results are not too shabby...in line with what we've been expecting - that the GF100 is faster than the 5870. It will most likely be faster in other games, but to a smaller degree... Now the only question is how much it will cost...
a c 271 U Graphics card
January 20, 2010 2:28:12 AM

Bluescreendeath said:
Now the only question is how much it will cost...


And how long until second gen?
Score
0
a b U Graphics card
January 20, 2010 2:41:24 AM

They could have sent an email to Hardware Canucks or NVIDIA to find out the system configuration...
Score
0
Related resources
January 20, 2010 2:55:22 AM

+1 to Randomizer; or just do a couple of test with a GTX285 till they get the same results as Nvidia and do the test with the HD5870 with those settings.
Score
0
a b U Graphics card
January 20, 2010 3:06:48 AM

Another thing I notice is they have the Physics set to very high which would further tilt the results to the Nvidia based card with the use of the Ageia physx engine -plus who knows if Nvidia had a seperate physx card installed as well or not to skew the result - I'll wait for some real benches to see what the real performance winds up being !!
Score
0
a b U Graphics card
January 20, 2010 3:27:14 AM

Far Cry 2 doesn't used PhysX so that has no effect on the result.
Score
0
a b U Graphics card
January 20, 2010 3:48:03 AM

Nice find! I think it's only going to get more interesting when more games are bench-marked.
Score
0
a b U Graphics card
January 20, 2010 4:55:30 AM

Far cry2 uses havok to render physics....so fc2 is certainly benefited by ati cards
Score
0
a b U Graphics card
January 20, 2010 5:14:30 AM

looks about right.

one test is hardly conclusive, but i think its probably going to be about right. better performance than a 5870, though probably not enough to justify the extra cost (fair enough, we dont know the cost yet, but im sure it will be much higher).

given that fermi is also unlikely to OC as well (heat output) it doesnt look so great right now.

we'll see though.
Score
0
a b U Graphics card
January 20, 2010 5:29:13 AM

GF100 pricing has to be competitive with the 5870 or no one is going to bother with it.
Score
0
a b U Graphics card
January 20, 2010 6:06:56 AM

i dont think it can be though. given the delays, die size and timing they won't be able to price competitively with ATI.

especially since ATI will almost certainly lower the price of the 58xx cards when fermi releases.

thats exactly why i don't see fermi being successful (relative the the 5870 at least).
Score
0
a b U Graphics card
January 20, 2010 8:06:14 AM

I don't care about HC after they said a
Quote:
SIMULATED
5870 would perform like this.
Score
0
a b U Graphics card
January 20, 2010 9:19:43 AM

Any models names out yet?
Also would like to know that the prices, HD5870 surely gonna come down to at least $330 - $350! And i am pretty sure that something equivalent to HD5870 from nVidia would costs somethig around $450.

Aren't they targeting the HD5970?
Score
0
a b U Graphics card
January 20, 2010 9:30:24 AM

Quote:
Wait till march.


Or June (latest rumor) for mainstream products.
Score
0
a b U Graphics card
January 20, 2010 9:32:17 AM

Eagerly waiting... but cant resist speculating! :D 
Score
0
January 20, 2010 10:11:58 AM

Don’t forget guys that these benchmarks are done with the GTX360 GF100. The GTX380 GF104 is going to be about 15% faster than the GTX360.
Score
0
a b U Graphics card
January 20, 2010 10:13:26 AM

michaelmk86 said:
Don’t forget guys that these benchmarks are done with the GTX360 GF100. The GTX380 GF104 is going to be about 15% faster than the GTX360.


Please show us how do you know these benchmarks are done with the GTX360. From what I know nVidia doesn't have a naming scheme yet.
Score
0
a b U Graphics card
January 20, 2010 10:24:24 AM

michaelmk86 said:
Don’t forget guys that these benchmarks are done with the GTX360 GF100. The GTX380 GF104 is going to be about 15% faster than the GTX360.


Wrong. The GF100 is the high end part, read the whitepaper.
Score
0
a b U Graphics card
January 20, 2010 10:32:11 AM

But I read it on the internets, and besides, why would nVidia PR be wrong?

j/k

This is how its going. Relax, Fermi will get here eventually, and these puny crumbs wont matter then
Score
0
January 20, 2010 10:32:54 AM

I never doubted Fermi will be faster than 5870. However the question is not whether but BY HOW MUCH.

Giving the size and price it will come (probably around 600 if not more) it should be at least 50% faster or near 5970 to be able to sell at that price. If its only 20% faster, then its a loss, as AMD will come out with updated Cypress in May-June time (could be even earlier) and that will be 285/4890 all over again. Nvidia will have about 1-2 months to sell the card then it will be getting into NON profit area again.

I really hope Fermi is 50% faster than 5870 (and I speak on average). This will probably drive prices of 5870 to $350 and 5970 to probably $500 - that is if nVidia can price Fermi at $500 which is so highly doubtful
Score
0
a b U Graphics card
January 20, 2010 10:42:34 AM

Nvidia aimed for the 5970 but fell quite a bit short. If 40% faster in Far Cry 2 is the best it can do (and surely Nvidia would show us the best?), then that is a long way short of the 5970.

It might catch it in some TWIMTBP games and games with heavy tessellation (assuming the tessellation is as good as claimed), but it will fall half way between a 5870 and 5970 in most games I think. If that ends up true, it's going to be a hard sell vs the 5970 at the same price.

Nvidia might just decide that being fastest in the 'ultra enthusiast' top end sli market is good enough this series. 3x GF100's should beat 2x5970's...maybe, and if you have a nearby nuclear power plant.
Score
0
a b U Graphics card
January 20, 2010 11:25:35 AM

Look, we know nothing from this. The best new stuff is the tesselation stuff, and thats being debated as to its scalability. We simply dont know enough.
Totally new arch, new approach, doing double triangle setup (actually doing 4x, but it being halved by using only half hot clock speeds). We have no idea how itll perform in bungholios compared to real world. I point to the past, and to physx as well.
Cant compare it to last gen, cant compare it to ATI, we know very little more than we did before, other than maybe itll perform like most has guessed it would anyways
Score
0
a b U Graphics card
January 20, 2010 11:30:40 AM

I actually expect a $350-$400 price point. Based on how they've re-designed the chip (the latest anandtech article), it should also be far cheaper to make lower-end parts by simply cutting the excess hardware, so I'm just as interested in the 360GT/GTX vs the 5850/5770...
Score
0
a b U Graphics card
January 20, 2010 11:47:36 AM

Well, Im thinking the 360 will be a full chip, just neutered down. I expect there to be plenty of them mas well, because the top cards will be hard to come by, if rumors are true, plus, unlike last gen, their other card, non gpu cards, thier failure rates didnt enter in to the highest end, as they has a highest end, unlike this gen, where the highest end chips have to be used for both the first and second iterations for their gpgpu chips, as the high end is 448, not 512, which points to how hard a full chip is going to be, plus, it uses everything from 448 on up for both the highend and second card for the gpgpu, making for even fewer for gpu usage only.
In other words, the 512 cards, for gpu, will be in short supply
Score
0
a b U Graphics card
January 20, 2010 11:50:22 AM

But ... but....but it will have PhysiX and ... a....aaa.....tessalation....and ..aa...it will be Green!!!!
Score
0
a c 130 U Graphics card
January 20, 2010 12:19:53 PM

Yes i watched the video of this the other day, it looks good of that there can be no doubt. Only thing is its (as said) an Nvidia demo set up, so caution as ususal is heavily advised before getting too carried away with it. If all is as reported it looks good.
Relativly speaking there isnt long to go now untill we start getting some numbers out, followed by the cards and pricing, thats what will make or break this card, how its priced vs its perf.

Mactronix
Score
0
a b U Graphics card
January 20, 2010 12:38:32 PM

Quote:
Fixed.


well the fix is only partial....I think you should read the third paragraph...
http://news.softpedia.com/news/Havok-Will-Not-Support-G...

It says Ati supports HAVOK but instead of running it on its GPU it optimizes the code to run HAVOK faster on the cpu...So its the ati which optimize the code ,not nVidia..Therefore FC2 runs better with ATI
Score
0
January 20, 2010 12:51:37 PM

Performance aside, I think the price will jolt the faint of heart. They will sell as many as they make because they have borderline occult followers. I dont think they will be able to price it for market penetration though and I think they will be expensive enough that ATI can maintain their current prices if they choose and still win this round soundly.
Score
0
January 20, 2010 1:17:34 PM

yep. the simple fact remains they really don't have to and will still sell their 5xxx series as fast as they make them.
Score
0
a b U Graphics card
January 20, 2010 2:28:42 PM

Does anyone have a picture of the setup? Hardware canucks talks about it being a 512 SPU unit or 448 unit and possibly underclocked for demo purposes etc. However most of the main demo units (outside the Unigine model) were watercooled, like the ones in the Rocket Sled and the previous 'HD5870 v Fermi' videos.

There's really no way to compare a 'theoretical' HD5870 to a specific model Fermi, even with a GTX285 to compare to. Especially after the FUD that it was an HD5870 it was up against originally.

Also I'll ask again why was Physics on? That's fine for testing a game for a game, but terrible for testing GPUs as it puts more focus on the CPU and system resources than you'd want for straight up GPU vs GPU comparison.

Still waiting for something more than shaky-cam and paper. [:jaydeejohn:7]

Score
0
January 20, 2010 2:41:33 PM

those numbers basically mean nothing GGA. they can do or say any number of things to slant the results the way they want and it is only one game title eve if it is legitimate.
Score
0
January 20, 2010 3:07:22 PM

jsc said:
GF100 pricing has to be competitive with the 5870 or no one is going to bother with it.

Nvidiots will buy them, just like they bought the 8800 ultras.
Score
0
January 20, 2010 3:24:09 PM

michaelmk86 said:
Don’t forget guys that these benchmarks are done with the GTX360 GF100. The GTX380 GF104 is going to be about 15% faster than the GTX360.


I know this has already been covered, but GF100 doesn't necessarily seem to be the "high-end" card as some of you put it, I think that is more just the general naming scheme for the line of cards. I know Fermi is the name for the architecture and the NV100 is the name of the chip itself.

I imagine the high ends will be the GF180, GF170, and then you'll have your 160, 150, 140's.... I'm purely speculating of course so I could be totally wrong, but that'd be my guess.

Oh and the GTX380, 360, etc. are going to be the Tesla series names, not the regular gaming card GeForce ones, which is what I guess most people on here care about if they care at all.
Score
0
a b U Graphics card
January 20, 2010 3:25:27 PM

If theres anything that may be seen as decent from all this is, claiming what thetre claiming, being as late as they are, if they do the pooch with perf, after being late, this will do more harm than good, so, in a sense, I think itll be halfways decent, or at least what we saw last gen, 4870 vs 280, tho again, I also feel, anything less is a huge failure.

Theoretically, it should be better, because of the new arch, but if its worse, bad news
Score
0
a b U Graphics card
January 20, 2010 3:30:51 PM

LOL , late, late, where are the deadlines ? What DX11 game are AT I know itall fanbois playing that the rest of us are missing out on ?
Score
0
January 20, 2010 3:31:54 PM

well JDJ i dont know about better. i think it will be faster and buggy myself. wont keep people from buying them and crying about it later though.
Score
0
January 20, 2010 3:32:26 PM

Sounds like a failure in the making. Anytime a tech company spends more effort on hyping and marketing their products than working on them, then its almost a guaranteed mediocre and disappointing product.
Score
0
a b U Graphics card
January 20, 2010 3:38:12 PM

RealityRush said:
I know this has already been covered, but GF100 doesn't necessarily seem to be the "high-end" card as some of you put it, I think that is more just the general naming scheme for the line of cards. I know Fermi is the name for the architecture and the NV100 is the name of the chip itself.


The whitepaper clearly states....

Quote:
"GF" denotes that the chip is a Graphics solution based on the Fermi architecture. "100" denotes that this is the high end part of the "GF" family of gpu's.
Score
0
a b U Graphics card
January 20, 2010 3:40:15 PM

Quote:
Nvidia aimed for the 5970 but fell quite a bit short. If 40% faster in Far Cry 2 is the best it can do (and surely Nvidia would show us the best?), then that is a long way short of the 5970.

It might catch it in some TWIMTBP games and games with heavy tessellation (assuming the tessellation is as good as claimed), but it will fall half way between a 5870 and 5970 in most games I think. If that ends up true, it's going to be a hard sell vs the 5970 at the same price.

Nvidia might just decide that being fastest in the 'ultra enthusiast' top end sli market is good enough this series. 3x GF100's should beat 2x5970's...maybe, and if you have a nearby nuclear power plant.



Again, if the GF100 is 40% faster then the 285 across the board, it beats the 5870, and possibly the 5970. And I don't think the chip will be nearly as expensive as the rest of you do.
Score
0
a b U Graphics card
January 20, 2010 3:48:17 PM

I think ATI should make some vids on the 5890, say its 50% better than the 5870, put up some slides etc, make some white papers on the changes needed for the power etc, and Im sure some people would believe it.
Whats current pricing on the 285? Bump that aways up, only makes sense right? Its 40% better right? No way this card comes out at 400$, or 450$, its a sure 500$ or more card. If the perf is less than 40%, maybe less, depending on how much less
Score
0
a b U Graphics card
January 20, 2010 3:52:37 PM

gamerk316 said:
Again, if the GF100 is 40% faster then the 285 across the board, it beats the 5870, and possibly the 5970. And I don't think the chip will be nearly as expensive as the rest of you do.


Every indication we see points to a very large chip that will probably need heavy cooling. We know they have had horrible yields and still are. Nvidia aren't exactly well known for selling cheap chips.

Why would you believe these cards are going to be cheap?
Score
0
January 20, 2010 3:54:15 PM

Haven't people learned not to trust bench markings of products from companies that makes them...
Score
0
a c 130 U Graphics card
January 20, 2010 3:57:02 PM

Time will tell ladies and gents.
Nvidia have always been PR junkies anyway, thats a corperate decision and has nothing to do with how the card will perform when it gets here.
People can talk all they like about die size and power usage etc, but that just dosent matter in the end of the day.
People have been doing the card down using unconfirmed rumour,specs, and benchmark vid's for ages now.
Benchmarks that have been done are stated to be on engineering samples anyway so the perf still isnt known yet, i would guess its going to be up on what we have seen so far at any rate.

Im prepared to give a large slice of benefit to Nvidia untill this card is PROVEN to be rubish, im just a novice as far as actual transistor level theory is concerned but i saw plenty of reasons to be excited by what was being offered right from the get go, others not so and are moving their position around now that its seems like it is going to be a decent performer after all.

Mactronix
Score
0
January 20, 2010 3:59:08 PM

notty22 said:
LOL , late, late, where are the deadlines ? What DX11 game are AT I know itall fanbois playing that the rest of us are missing out on ?


Dude.... stop using the term fanboy, I'm getting sick and tired of reading people using that term. There's a reason calling someone a "fanboy" on EOCF is a bannable offense.

There ARE games out right now that use DX11 that some people want to play, and he CLEARLY meant they are late in releasing DX11 support, nothing more. And unless you're going to go back in time and stall ATI, that's the way it is.

jennyh said:
The whitepaper clearly states....

Quote:
"GF" denotes that the chip is a Graphics solution based on the Fermi architecture. "100" denotes that this is the high end part of the "GF" family of gpu's.


My mistake jennyh. I wonder then.. does that mean the low end cards are going to be GF40, GF50, etc.? That would make sense as well I guess.

Quote:

Nvidia Fermi GF100
Min = 64.60 fps
Avg = 84.05 fps
Max = 126.20 fps

AMD Radeon HD 5970
Min = 76.42 fps
Avg = 99.79 fps
Max = 133.52 fps

Based on average fps scores, the 5870 is 19% faster than Nvidia's GF100. So, it doesn't beat the 5970.


Wait... what??
Score
0
a b U Graphics card
January 20, 2010 4:00:35 PM

juncwil said:
Haven't people learned not to trust bench markings of products from companies that makes them...

Last time I looked at the ATI propaganda they claimed to be working with Dirt2 developers for a cutting edge dx11 game.
Score
0
January 20, 2010 4:02:38 PM

notty22 said:
Last time I looked at the ATI propaganda they claimed to be working with Dirt2 developers for a cutting edge dx11 game.


Well you shouldn't believe them too much as well until you see solid proof.
Score
0
a b U Graphics card
January 20, 2010 4:14:50 PM

Thats the question SS. They certainly need to do this, and looks to be designed to this way, and current discussions are, each GPC is tied to a tri setup, so, halving them, for a midstream part, like a 5770 say, it also halves the tesselation abilities possibly, which will cripple nVidias ability to do tesselation, possibly, in their mid cards.
Another thing is, and this is regards to bungholio marks as well. Tiangle setup is doubled on Fermi, 4x actually, but only a half of the hot clock, so essentially doubled, so, when say heaven is run, if it uses small 8 bits or less, itll shine, whereas in games, if they arent limited to 8 and less, the entire amount is 32, regardless, anything over that, and its latency bound.
If you have tons of 8 bit cranking in the bungholio, itll kill it, up to a point, or do as well as it can do, much like we see with 2 other thimngs, theres benches out there that actually use almost all of the 5 way shaders on ATI cards, and in those benches, they destroy nVidia cards, having only 1, or furmark, where we see ATI cards having to throttle, because its being asked to do what a game never would, and the numbers being thrown out on furmark are incredible, but the heat issues are whats brought up.

So, at this point, until we know more, what weve seen in the heaven bench needs to come with caution, it may be the real deal, and what we see there will be useable in games, but, even so, the nVidia mid cards will automatically be cut in half, per GPC, and the ATI mids will look better in Heaven bench, even if they arent, so we need to wait
Score
0
a b U Graphics card
January 20, 2010 4:32:48 PM

Cheap Fermi - an oxymoron?

Speaking from an economic standpoint, the top-end Fermi cards simply won't be cheap and will therefore be limited to a very small portion of the market. NVIDIA needs to derivate down to the mid-range as soon as possible, to make the GPUs more appealing to Joe Average who is willing to shell out $150, and not $599, on a new card.
http://www.hexus.net/content/item.php?item=21996&page=3
Score
0
January 20, 2010 5:42:53 PM

they will JDJ. they will get that G92 in some brand new shiny plastic!
Score
0
a b U Graphics card
January 20, 2010 6:41:22 PM

Quote:
From the Link: Farcry 2 testing

Nvidia Fermi GF100
Min = 64.60 fps
Avg = 84.05 fps
Max = 126.20 fps

AMD Radeon HD 5970
Min = 76.42 fps
Avg = 99.79 fps
Max = 133.52 fps

Based on average fps scores, the 5970 is 19% faster than Nvidia's GF100. So, it doesn't beat the 5970.

...As for prices, we'll wait and see. Obviously ATI is the one setting the prices this round. Nvidia has a hard sell to make with when their chips are 40% larger (and therefore at least 40% more expensive to manufacture).


You can't do that comparision though. I can easily point to other games where 40% performance over a 285 beats a 5970. Even then, a single card would be trading blows with AMD's best dual-card, which isn't that bad a decision to be in...
Score
0
    • 1 / 8
    • 2
    • 3
    • 4
    • 5
    • More pages
    • Next
    • Newest
!