Comparison Revisited, Methods Refined: Nvidia GeForce GTX 480/470 vs. ATI Radeon HD 5870/5850

[dividebyzero]

I think you'll find that 16, 32 x AA etc are forced through the driver in general, which is probably a little outside the scope of a general review (no offense TS). I think the general idea was to test the cards at a level playable for all, using game menu settings which I think was achieved. My own personal queries regarding game selection aside, I think that the settings in general couldn't have been racked up too much higher without severely limiting the playability of the games. For instance
The next step up in IQ for Metro 2033 is basically 4xMSAA/16xAF ~20fps for the GTX480, ~14fps GTX470, ~3fps HD5850/5870...maybe a big % win for the GTX's but not playable by any means....a textbook example of a hollow victory
Crysis Warhead at the tested 4xAA/0xAF is what I would consider barely playable for the lower two cards. Going up a step to 4xAA/16xAF makes that a certainty.
As for 16+, I don't know if nVidia and ATI can be directly comparisoned-as you probably know.
While the new antialiasing for nVidia looks pretty good having seen it first hand, how do you compare an nVidia setting in say Modern Warfare 2 @ 16 (or 32) x S+TMSAA/16xAF (hybrid AA) with anything thats available to an ATI card ?

 

[grvalderrama]

How about you turn up the AA on some of your benchmarks and see which gpu actually does better when not playing at 4x or 8x. Seems like you are taking alot of what the GF100's do so well by curbing AA to 4x in games like RE5 and BC2. Let's see how well each card does if you force 24x or 32x in the same games, at least 16 plz. No more tests with AF off or AA crippled to make the 5870 look like it can compete.

Well, it should! Otherwise, there's no good reason to buy it!

 

[Guest]

I wouldn't call 4x or 8x AA "crippled". It would probably be akin to saying 'highest playable settings'. As was mention earlier. Hollow. Empty suit.

If the settings were maxed and brand A beat brand B by an "amazing" 98% it wouldn't matter if it's unplayable.

 

[Guest]

Very nice review.
So I see Fanboy's on both sides still trolling each other here.
to gain some perspective on this review i went to newegg and price checked the 480 vs the 5870.
Pretty interesting and confusing prices for those who aren't looking at the specs of the cards.

Looking at these prices, the ATI 5870 prices range from $399.00 to ""$499.00"" for the 1GB 256 bit cards.

The Nvidia 480... ranges in price from ""$499.00"" to $529 for the 1.5 GB 384 bit cards

The cards from ati for $399 are lower end 5870's with lower clocks for memory and core.
and would compare with the 470.
I find it funny the prices range $100 on the ATI

Looking at the 480's prices there is a $30 price range difference.

The review here show's the 480 beating the 5870 on every front. and they did it on (beta) drivers vs ATI's 5th gen drivers. yes fanboy's 10.5 means 5 patches and most of them were to fix issues with frame rates on a few of the games tested here. I think when Nvidia brings new drivers to the table the performance will totally kill the 5870... ( the betas already beat them)
I build systems with both cards....

Budget systems get the ATI/AMD setup. ( price plays a very large role when people are buying a system)

For the Gaming enthusiast it's Nvidia/Intel system .
You don't need water cooling for a 480.... anyone who is spending $499 on a video card has already spent the money for a good case with proper cooling and a good power supply.

 

[Guest]

Best Comment, ever! Specially the part when fanboys were taught how much better power and heat the G80 did now a small minority of them completely flipflop their position on that. It's laughable and memorable how much a hypocrite some can be.

 

[grvalderrama]

Very nice review.
So I see Fanboy's on both sides still trolling each other here.
to gain some perspective on this review i went to newegg and price checked the 480 vs the 5870.
Pretty interesting and confusing prices for those who aren't looking at the specs of the cards.

Looking at these prices, the ATI 5870 prices range from $399.00 to ""$499.00"" for the 1GB 256 bit cards.

The Nvidia 480... ranges in price from ""$499.00"" to $529 for the 1.5 GB 384 bit cards

The cards from ati for $399 are lower end 5870's with lower clocks for memory and core.
and would compare with the 470.
I find it funny the prices range $100 on the ATI

Looking at the 480's prices there is a $30 price range difference.

The review here show's the 480 beating the 5870 on every front. and they did it on (beta) drivers vs ATI's 5th gen drivers. yes fanboy's 10.5 means 5 patches and most of them were to fix issues with frame rates on a few of the games tested here. I think when Nvidia brings new drivers to the table the performance will totally kill the 5870... ( the betas already beat them)
I build systems with both cards....

Budget systems get the ATI/AMD setup. ( price plays a very large role when people are buying a system)

For the Gaming enthusiast it's Nvidia/Intel system .
You don't need water cooling for a 480.... anyone who is spending $499 on a video card has already spent the money for a good case with proper cooling and a good power supply.

I mean no disrespect, but I believe you don't get it, Nvidia had plenty of time to improve their cards (and drivers) and leave 5870 and 5850 on the dust, instead they've created cards that just compite against them, and barely win.... and just to realize that you can boil water on your card! Not to mention the power consuption...
Are they great card? Yes they are. Are they efficient? No, they are not... But more important, they could have been more efficient.
And for the record, for the gaming enthusiasts there's 5970, since they don't seem to care about budget...

 

[grvalderrama]

I wouldn't call 4x or 8x AA "crippled". It would probably be akin to saying 'highest playable settings'. As was mention earlier. Hollow. Empty suit.

If the settings were maxed and brand A beat brand B by an "amazing" 98% it wouldn't matter if it's unplayable.

What's the point of "an amazing 98%" if you cannot enjoy the game? More so, what's the point of spending all that money on high-end cards and computer if you can't enjoy a game?

 

[Guest]

Well I agree with the other guy. The 480 beats the 5870 for the same price. Go look it up.
And about the boiling water part, Your funny. A good case with good fans will do the job just right.

The ATI Fanboy's Can't Stand The Fact That This Review Shows The ATI Card Losing.

Quotes from customers who purchased these 480 cards on newegg................................................................................

GPU is 40 C idle, 70 C on load. overclocked i7 920 system is in mid 40's.

no problem idle around 45c full load around 76c with 1200 antc keep cool them down lots! crysis around 45fps on max with 4X AA and Merto 2033 highest with 4X AAA around 40 to 60fps with DX 11. win 7 rated 7.6 both!


from all the reviews i've read about it getting way too hot.. LIES this card is running under 70 Degrees at fan speed of 50% also movies, games, everything.. looks just amazing, props to this card =)


Some complain that the card runs hot and is loud, but i'm coming from a 4850X2 so this sounds and feels like a child's nursery in a freezer in comparison

Not as hot running as they say, i have bad ventilation in my Mid-case and it idles at 46c on a hot day and 42c on a cool day (south FL) never goes above 79c with fan at 100% (rather protect it than whine about noise)
Use MSI afterburner for its excellent progressive fan profile system. (look for Beta at this time)


I don't know why everyone is complaining about the heat. It's not hotter then my GTX 280 was. Idle around 44-46 Playing BF2 with 80% fan goes up to around 76-80 so it's not bad at all.

 

[red1776]

The 480 beats the 5870 for the same price. Go look it up.

where are you finding the GTX 480 and the 5870 for the same price?
I did look it up (since you reference Newegg) and the 5870 can be had for $400 while the 480 starts at $500 and goes up. You led with, and the seeming crux of your point was that the 480 could beat the 5870 at the same price. it in fact does not. (and by your own reference).
gotta love it when a green fanboy goes around complaining about red fanboys. It upholds my long standing assertion that many times if you want to know what others are up to...listen to what they are accusing you of doing. Rock on NvFanboy....rock on.


http://www.newegg.com/Product/Product.aspx?Item=N82E16814125316&cm_re=hd_5870-_-14-125-316-_-Product

http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&DEPA=0&Order=BESTMATCH&Description=gtx+480

 

[grvalderrama]

The ATI Fanboy's Can't Stand The Fact That This Review Shows The ATI Card Losing.

from all the reviews i've read about it getting way too hot.. LIES this card is running under 70 Degrees at fan speed of 50% also movies, games, everything.. looks just amazing, props to this card =)

So, I'm supposed to believe what you say and not take into account Techspot's review? It's hard to do that... even more if you are just a guest, I cannot rely on your opinion. Besides, why would they ALL lie? Can they all be lying?? Are you kidding?

Furthermore, I am not an ATI fan (see my system spec...), and I do recognise when Nvidia or Ati do a great job on their products, and you should do it too. Maybe you just can't stand that Ati did it much better than Nvidia this time, can you?

Calling Techspot people liers... what an ungrateful Guest!

 

[dividebyzero]

re "Guest" post #83

red, grvalderrama,

I really wouldn't waste my time on an obvious troll, although I suspect this Guest is more a ham-fisted attempt at misdirection.
If this is an nV***** it's blatantly the worst I've seen for some time...but I suspect it's an AMD troll since every fact can be easily refuted. A little more sublety might have achieved a better response.
* How difficult is it to refute the pricing of the individual cards ?
* The poster has, or had, supposedly, a GTX 280, HD 4850X2 and now a GTX4xx ? But no ! Poster doesn't mention he has that card...just that other do.
* Plays Metro 2033 and doesn't know that the default Analytical AA (AAA) setting is disabled when applying 4xAA (MSAA)

BTW Does this sentance even make sense ?

but i'm coming from a 4850X2 so this sounds and feels like a child's nursery in a freezer in comparison

So, bs statements that a five-year-old could refute, bombastic nonsensical statements and a quote which doesn't support the posters argument. Retarded fanboy for sure...red or green, we'd rather you ****** off and got your jollies elsewhere

 

[red1776]

red, grvalderrama,

I really wouldn't waste my time on an obvious troll, although I suspect this Guest is more a ham-fisted attempt at misdirection.
If this is an nV***** it's blatantly the worst I've seen for some time...but I suspect it's an AMD troll since every fact can be easily refuted. A little more sublety might have achieved a better response.
* How difficult is it to refute the pricing of the individual cards ?
* The poster has, or had, supposedly, a GTX 280, HD 4850X2 and now a GTX4xx ? But no ! Poster doesn't mention he has that card...just that other do.
* Plays Metro 2033 and doesn't know that the default Analytical AA (AAA) setting is disabled when applying 4xAA (MSAA)

BTW Does this sentance even make sense ?


So, bs statements that a five-year-old could refute, bombastic nonsensical statements and a quote which doesn't support the posters argument. Retarded fanboy for sure...red or green, we'd rather you ****** off and got your jollies elsewhere

:haha:, yes to all, but like I heard elsewhere, I guess I'm easily baited since they opened the doors to drive by 'unidentified guest' comments.
and I have no idea what the comment:

but i'm coming from a 4850X2 so this sounds and feels like a child's nursery in a freezer in comparison

could possibly mean. I mean I own a HD 4850 X2 and it runs cooler than frog in heat during a hailstorm.

 

[Guest]

So according to this article

Ati 5850 >>>>>>>>>> GTX470

 

[dividebyzero]

Yup, according to the article I'd say that about sums it up.
But like most reviews, it's a snapshot of a few games with a particular driver set and a particular resolution, for example :
Adding Sims 3 and Crysis or testing the reviewed games at 2560x1600 would result in
5850 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>GTX 470
While a review benchmark suite of CoD:MW2, Far Cry 2, HAWX, Battleforge, Batman:AA and World in Conflict for example would have resulted in
GTX 470 >>>>>>>>>>>>>> HD 5870
It's also very much the case that modern games have a multitude of graphics settings and it's generally noted in reviews that "maximum detail" or "maximum settings" etc are used, but even these absolute parameters usually don't convey the exact nature of the settings used. For example some reviews of BFBC2 don't enable HBAO (similar to high dynamic lighting that diffuses shadowing dependant upon light source) as I'm pretty certain the TS review has, while other reviews using the same "max details" setting will enable it- knocking around 20-25% off the ATI cards framerate while affecting the nVidia card to a much lesser degree- hardly surprising since nVidia has been involved in its evolution
So it's pretty much a case of YMMV.

 

[Guest]

The only issue with that dividebyzero is that most of the games you have listed run at well over 60fps, even over 100fps with these graphics cards. So while yes the GeForce GTX 470 might beat the Radeon HD 5850 by 10-20fps when testing with World in Conflict for example, the fact that the Radeon HD 5850 is still spitting out over 80fps makes it a hollow victory I feel.

Its fair to say any of these graphics cards smash CoD:MW2, Far Cry 2, HAWX, Battleforge, Batman:AA and World in Conflict with ease, even at 2560x1600. So I am guessing this is the very reason why the review does not include these titles the second time around.

 

[red1776]

The only issue with that dividebyzero is that most of the games you have listed run at well over 60fps, even over 100fps with these graphics cards. So while yes the GeForce GTX 470 might beat the Radeon HD 5850 by 10-20fps when testing with World in Conflict for example, the fact that the Radeon HD 5850 is still spitting out over 80fps makes it a hollow victory I feel.

The benchmarks are as much about the strengths of the card for future considerations. There may not be a difference in 20 FPS in the games they test right now, but the next round of games when 20 fps is the difference between playable and not, you might want the 470. In the case of these DX11 cards, i would say that its mostly about games not yet released that need those "hollow victory FPS.

 

[Steve]

Right but we all know that it does not always work that way. For example the evidence suggests quite the opposite. While the GeForce GTX 470 cleans up in those older games mentioned it’s only as fast as the Radeon HD 5850 in newer games such as Battlefield Bad Company 2, Aliens vs. Predator and Metro 2033 for example. I think he has a point...

Games are unlikely to change significantly over the next 12 months and after that point the Radeon HD 5850 and GeForce GTX 470 will likely become redundant and any half serious gamer will upgrade to a graphics card from whichever manufacturer is offering the best deal at the time.

Resident Evil 5 is a perfect example of this as it was featured in this article. The Radeon HD 5850 beat the GeForce GTX 470 but both provided minimum frame rates of over 60fps at 1920x1200 using maximum in-game quality settings with 8xAA enabled. Yes the Radeon HD 5850 won but in this game it doesn’t really matter. Like you say that margin can be translated to other more demanding games, but not always, which is why we tried to test with as many games as possible.

Also we are planing a few similar articles and for reasons mentioned by other readers we will leave games out such as Modern Warfare 2 since they present very little challenge for these graphics cards.

dividebyzero is right though the settings and games selected for testing do play a critical role and its difficult to include more games than we have, especially when changing how we test the games. Since running these tests 2 or 3 times more to accommodate the usual tested resolutions would have been a huge task we stuck with a single resolution. Therefore we felt it made sense to test at 1920x1200 as this is a middle ground.

As for the settings we went with maximum in-game quality settings with reasonable levels of AA when possible. Those suggesting we tested with 16x and 32x need to get back on their meds, did you not see the frame rates in games such as Metro 2033? I think they were low enough myself.

Thought I might add that HBAO was enabled when testing Battlefield Bad Company 2.

 

[dividebyzero]

The only issue with that dividebyzero is that most of the games you have listed run at well over 60fps, even over 100fps with these graphics cards..

HAWX at 1920 and 2560 at max detail including AO
World in Conflict at 1920 and 2560 , Very High detail
Far Cry 2 at 1920 and 2560, Max detail, 8xAA
Battleforge (DX11) at 1920 and 2560 , max detail (note minimum framerares also)
Kind of puts paid to that theory

So while yes the GeForce GTX 470 might beat the Radeon HD 5850 by 10-20fps when testing with World in Conflict for example, the fact that the Radeon HD 5850 is still spitting out over 80fps makes it a hollow victory I feel.

I would agree....if the 5850 was spitting at 80+ fps, but since it's somewhere south of 60 then probably not. And by extrapolation you also seem to have decided that multi-monitor gaming resolutions are irrelevant

..Its fair to say any of these graphics cards smash CoD:MW2, Far Cry 2, HAWX, Battleforge, Batman:AA and World in Conflict with ease, even at 2560x1600..

If you call 35+ fps "smashing with ease" then I guess so

..So I am guessing this is the very reason why the review does not include these titles the second time around.

And you seem to be missing the point of my post, which was that depending on the game benchmark suite, resolution and game IQ, either (or any) of these cards could be seen as a "benchmark winner", and that with the myriad of games and settings available, there are no clear cut "winners".
If, by your metric, that any "extra" fps over and above playable are superfluous then for the most part any card above a HD5770/5830 or GTX465 falls into that category.

 

[red1776]

@ Steve,
True it gets rather game specific, But taking a big picture overview I think gives you a good guide to the capabilities of the cards. you would probably have a better inside knowledge of whats coming down the road, but why do you not expect games to change much over the next 12 months? it appears that DX11 is catching on and the Nvidia cards have done much better in every test of 'heavy,and medium tessellation' that I have seen. That seems to be the definition of future considerations for graphic cards at this point.

 

[Steve]

Yes you are right there will be more DX11 games but I am not sure heavy tessellation effects are going to be enjoyed on a GeForce GTX 470 anyway. I also think we are yet to really see the bigger picture which will come as more truly DX11 games are released. Still looking at the demanding games we have right now there is little difference between the Radeon HD 5850 and GeForce GTX 470 in terms of performance while the biggest difference is in the price.

I still believe if you were to buy either graphics card right now the Radeon HD 5850 would be the better choice. How this choice should change in 12 months time is difficult to say and a lot of that will come down to driver development and support for newer games.

The GeForce GTX 260 for example came home really strong taking it to the Radeon HD 4870 in newly released games just before those cards were phased out. So I can only guess as to how things will change towards the end of their life.

 

[dividebyzero]

@Steve
Would you care to speculate on the four known reviews then that definitely featured HBAO on 257.15 /Cat 10.5 based reviews of BFBC2 ?

All resolutions at 1920, all max detail + HBAO
........................HD5850..........GTX470............HD5870..........GTX480
Legion HW.........43 fps.............45fps..............52fps...............58fps
Xbit....................44.2................47.6................51.6.................59.7
HW Canucks.....42.55..............45.43...............50.67...............not tested
Techspot...........56...................47...................66....................68

Except for Legion Hardware who haven't stated their testing method, all benchmarking was done under gameplay conditions and recorded by FRAPS (TS -1min, Xbit -3 runs averaged, HWC 5 minutes)
Oddly enough Hexus and Tom's Hardware also tested BFBC2 using 257.15 and Cat 10.5 at "max detail" (not stated whether HBAO enabled) with similar results:
Hexus : 53.27fps (5850), 53.4 (GTX470), 62.33 (5870), 68.8 (GTX480)
(tested on 30sec gameplay, recorded using FRAPS)
Tom's : 58.13fps (5850), 62.24 (GTX470), 69.44 (5870), 76.57 (GTX480)
(tested on 2 min 25sec gameplay, recorded using FRAPS)

Just seems really, really odd that testing by the other sites showed the GTX470 to have between 100 and 108% of the performance of the 5850, yet TS shows 84%

 

[Steve]

Looking at the results it seems that only the GeForce GTX 470 is out of whack in the scene that we tested. The scene used for testing in this article is less demanding than the one normally used as I found it easier to accurately measure a minute of game play. Normally I use the water fall scene. So this would explain the frame rates being higher on all cards, except for the GeForce GTX 470. I will look into this and see if I can work out why the GeForce GTX 470 is so low here.

 

[dividebyzero]

A pretty concise article on GPU Benchmarking methodology at Hardware Canucks

 

[Guest]

I gladly pay a few bucks extra from the HD5850 for the GTX470 to be able to have a flawless gaming experience in Linux too, I use Linux a lot and I'm always at the bleeding edge of Xorg/Kernels/Wine and ATI are always months behind in their support.

With Nvidia I'm able to use the absolutely latest stuff to it's full effect, gaming or video. Linux becomes more and more interesting because it's generally more painless to use for all other stuff than gaming if your reasonably computer literate, and it's more secure, and a native Steam-client is on it's way which is nice even though the Windows-client works fine under Wine.

ATI have interesting hardware, but Linux-support is terrible so they are promptly off my list.
(I currently have and use two HD4***-cards so it's not that I'm unfamiliar with ATI-cards, but I prefer my Nvidia cards because of the aforementioned reasons)

 

[Guest]

I actually have a GTX 470 and I love it. Yeah I'm with watercooling and the card is running on 850 core and 925 memory and I think that it beats even the GTX 480: Crysis 50 (even 60 at some points) fps at 1920x1080 very high w/o AA, BBC2 at 60-70 fps everything maxed out with 4xMSAA and 4xTRSAA, etc. I don't think that there is a game which can't be run on this card. I've paid the same price as for a HD 5870 and I'm pretty sure that an OC'd 5870 will be outperformed by my card (Unigine Heaven 2.0 with extreme tessellation - 40fps same resolution 1920x1080). If you guys prefere ATI, ok. Yeah it will be cheaper and don't get me wrong, I'm sure that almost every card has the right buyer for it if you get what I mean. I think that for the bucks I paid I've got the best bang. If my budget was smaller I would certanly buy a HD 5850 or 5870 w/o water. These are great cards, but if you have the money and want to play the newest games as they were ment to be played go for the latest hardware. In the other case you have to optimize your machine.

P.S. I almost forgot the Fermi cards demand a good CPU. Don't go for it with a Core 2 Duo or something like that. Overclock your CPU ))