New arma2 benchmark scores 5870 vs 480 with OC

[RogueBlade 0]

http://alienbabeltech.com/main/?p=16997&page=16

[Leon86 13]

Overclocking a 480, there's a good idea. Guess it didn't get hot enough on standard clocks :)

It looks like the 480 is pretty much on par with the 5870.

[SWAT_BigBear 0]

Not impressed, since my "average fps" is only 1-2 digits lower of the 2560x1600 score.

Edited April 13, 2010 by SWAT_BigBear

[MavericK96 0]

Well it's not like ArmA2 is super GPU-dependent...

[Avneet 10]

You have to count ATI already has it's drivers matured. Remember how much FPS gain there was when ATI had it's new drivers. Now imagine the fact that Nvidias drivers are still the first ones. I see the 480 going much further.

[Leon86 13]

Well it's not like ArmA2 is super GPU-dependent...

That all depends on settings. Put your 3D res on 200%, AA on, anisotropic filtering and postprocessing on max. That'll be a slideshow for sure.

[DaSquade 0]

Uhm, is it me, but they don't mention what version of arma2 they run. From the info i'm even forced to say they run the demo.

So not really a review you can put your money on imho although it might give you an indication.

Like above mentioned, what settings, mission etc?

Sadly ArmA isn't your average, plug and go game...

[bangtail 0]

Not impressed, since my "average fps" is only 1-2 digits lower of the 2560x1600 score.

The 480 and the 5870 will demolish a 285. Even though A2 is not very GPU dependant, both those cards have significantly more horsepower.

Minimum FPS are very important and the 285 just doesnt cut it at that res.

http://www.hardocp.com/article/2009/09/22/amds_ati_radeon_hd_5870_video_card_review/8

Couldn't find any reviews of the 285 vs 480 vis a vis Arma 2 but you can expect the 480 to maintain a better minimum FPS than the 5870 across the board.

Edited April 13, 2010 by BangTail

[SWAT_BigBear 0]

The 480 and the 5870 will demolish a 285.

Oh, I'm sorry...didn't realize I said the 285 was better..:rolleyes:\

For the price tag, I would have expected a lil lot more.

Btw, I do not need to read links concerning my pc, If it didn't perform properly for me, don't ya think I would build another one?

[galzohar 31]

You can probably ignore any benchmark that does not specify that they are using Arma 2 1.05.

[MavericK96 0]

That all depends on settings. Put your 3D res on 200%, AA on, anisotropic filtering and postprocessing on max. That'll be a slideshow for sure.

True. It said they used "maximum details" with AA on "high" for the tests. I don't know what exactly that entails, but based on the scores I'm guessing it was CPU-bound at that point. The 480 is considerably faster than the 5870 in most games, and at least marginally (5-10%) better in pretty much all games.

[RogueBlade 0]

True. It said they used "maximum details" with AA on "high" for the tests. I don't know what exactly that entails, but based on the scores I'm guessing it was CPU-bound at that point. The 480 is considerably faster than the 5870 in most games, and at least marginally (5-10%) better in pretty much all games.

Considerably faster? sources? I've been reading benches all week, can't be bothered to start posting comparison charts but the 480 is definitely not "considerably faster". Yes it wins here and there. One example of the 5870 being faster is that island looping bench in crysis, I forget what it's called. 5870 won that which was pretty surprising as cyrsis is supposed to be an nvidia optimized game.

The only case where the 480 it is considerably faster is with extreme tessellation on something like the heaven benchmark. Thing is, games won't even be making use of tessellation until the next gen of dx11 cards start coming out.

[MavericK96 0]

Considerably faster? sources? I've been reading benches all week, can't be bothered to start posting comparison charts but the 480 is definitely not "considerably faster". Yes it wins here and there. One example of the 5870 being faster is that island looping bench in crysis, I forget what it's called. 5870 won that which was pretty surprising as cyrsis is supposed to be an nvidia optimized game.

The only case where the 480 it is considerably faster is with extreme tessellation on something like the heaven benchmark. Thing is, games won't even be making use of tessellation until the next gen of dx11 cards start coming out.

Check out the Guru3D review. Definitely a lot of games where the 480 is ~10-15+ FPS higher than the 5870. The margin closes as you increase resolution, but nonetheless, that is what the results show. Then again, I don't know what you consider "considerably faster". To me, 10 FPS gain is pretty considerable. Also, they tested Crysis in that review and the 480 was faster at all resolutions, so I'm not sure what benchmark you're referring to.

And yes, in games like Metro 2033 with DX11 and tessellation, the 480 has a very significant lead over the 5870.

[bangtail 0]

Considerably faster? sources? I've been reading benches all week, can't be bothered to start posting comparison charts but the 480 is definitely not "considerably faster". Yes it wins here and there. One example of the 5870 being faster is that island looping bench in crysis, I forget what it's called. 5870 won that which was pretty surprising as cyrsis is supposed to be an nvidia optimized game.

The only case where the 480 it is considerably faster is with extreme tessellation on something like the heaven benchmark. Thing is, games won't even be making use of tessellation until the next gen of dx11 cards start coming out.

The 480 is faster and in some cases by quite a bit (SLI scales way better than CF on the 4x0 series). This all comes at a cost. For the amount of power it pulls and heat it creates, the gains are not that great.

That being said, it is the fastest single GPU card on the market (although that probably won't last too long if ATI has anything to say about it).

Of course, you could always just buy 2 of these ;)

http://www.tweaktown.com/articles/3230/sapphire_radeon_hd_5970_toxic_4gb_video_card/index.html

Edited April 13, 2010 by BangTail

[RogueBlade 0]

[bangtail 0]

It's a well known fact that DX10 Crysis favours ATI.

Fermi is faster overall, them's the facts. It's a shame they couldn't get the heat and the TDP right but it is the fastest single GPU card.

Minimum FPS is what counts and Nvidia cleans up in both Single GPU and Multi GPU in Crysis Warhead

Edited April 14, 2010 by BangTail

[Bush 0]

Minimum FPS is what counts and Nvidia cleans up in both Single GPU and Multi GPU in Crysis Warhead

No it's not. There's much more to it, a simple minimum fps number do not have to reflect the whole framerate in a game. A minimum fps figure could just be a short single drop in the whole benchmark and you can’t judge anything by that.

Yes, GTX480 is faster than 5870 in most cases (not all) but at what cost?

More than 50% less performance/watt and a much higher price, almost the same as a 5970 which also draws less power than a single GTX480.

Edited April 14, 2010 by Bush

[MavericK96 0]

More than 50% less performance/watt and a much higher price, almost the same as a 5970 which also draws less power than a single GTX480.

What are you talking about? The 5870 is $400, the GTX 480 is $500, and the 5870 is upwards of $700.

[bangtail 0]

No it's not. There's much more to it, a simple minimum fps number do not have to reflect the whole framerate in a game. A minimum fps figure could just be a short single drop in the whole benchmark and you can’t judge anything by that.

Yes, GTX480 is faster than 5870 in most cases (not all) but at what cost?

More than 50% less performance/watt and a much higher price, almost the same as a 5970 which also draws less power than a single GTX480.

Do some research :rolleyes:

I bought my 2 PNY 480s (havent shipped yet) for $539 each, that's nowhere near the 5970 which retails for ~$700. You're wrong on the TDP as well, the 5970 draws 294w while the 480 draws 250w. The 5970 is also notoriously hard to find (this is 5 months after it's release and that availability will deteriorate further now) as Nvidia have reportedly jammed up TSMC with the 4x0 series production.

http://www.guru3d.com/article/geforce-gtx-470-480-review/13

http://www.guru3d.com/article/radeon-hd-5970-review-test/10

Minimum FPS is the single most important number in determining acceptable average FPS.

The 'short drop' as you describe it can effectively kill immersion so it is, in fact, critical to those of us who demand a constant, uninterrupted experience. The 480 provides a better min FPS than the 5870.

I really like ATI btw and it took me a long time to order the 480s (which is why I will probably have to wait 2 weeks for them). I concede that they use a lot of power and run fairly hot (although people who have them are saying that they don't run nearly as hot as some reviewers initially claimed, probably due to an updated BIOS).

If you want the fastest single GPU card, it's the 480. Heat and TDP be damned :D

Edited April 14, 2010 by BangTail

[jblackrupert 14]

3-way SLI and the temps look fine

http://www.youtube.com/watch?v=tpnB96BuICQ

[Bush 0]

Do some research :rolleyes:

I bought my 2 PNY 480s (havent shipped yet) for $539 each, that's nowhere near the 5970 which retails for ~$700.

In Europe the GTX480 is quite close to the 5970 in price, just a bit more than 50€ difference. And at least here in Sweden you can find both in stock in some online shops.

You're wrong on the TDP as well, the 5970 draws 294w while the 480 draws 250w.

Yeah but not according to the reviews out there:

Minimum FPS is the single most important number in determining acceptable average FPS.

The 'short drop' as you describe it can effectively kill immersion so it is, in fact, critical to those of us who demand a constant, uninterrupted experience. The 480 provides a better min FPS than the 5870.

But it's not that easy that you can just look at the minimum number.

For example it could be a case where the 5870 have one drop to 20fps in the whole benchmark and the rest is above 30fps, then the GTX480 have 3 drops to 25fps. You see what I mean, you can’t judge the performance by just looking at the minimum fps number. It’s better to look at a diagram over the whole benchmark where you can see all the fps drops.

Something like this is much more interesting than just a minimum number:

[galzohar 31]

Since there is no such thing as "FPS" in a given point in time, the definition of minimum FPS can actually change between different benchmarks. After all, all you can count is how many frames were rendered in the last X seconds or how much time have passed since the last frame was rendered. In between you could say all graphic cards have 0 FPS and infinite FPS for a infinitesimally small amount of time when the frame was rendered, thus have 0 minimum FPS and infinite maximum FPS. But all of this does not matter - what matters is whether or not you get low FPS for sustained periods of time.

If you get low FPS for less than 1/2 a second for the whole benchmark but the rest of the time your FPS is higher, then I'd say it's better. But if you get FPS drops for 1+ seconds every 10-20 seconds then that's a different matter.

A graph like the one in the above post gives a lot more info than just minimum/maximum/average, but it still misses some info like over what duration was the "momentary" FPS averaged (as there is no such thing as real momentary FPS, only X frames rendered within Y seconds). Though seeing the "jumps" in values are at least 1 frame, I assume the measurements were done every second (aka "how many frames were rendered during the last second").

One last thing, though - If you're going to look at only 1 number - Look at average FPS. After all, there aren't a lot of things that would make 1 card have more game-breaking FPS drops yet manage to maintain higher average FPS, but having some random element cause an FPS drop that lowered the minimum FPS for one card but not the other in an insignificant way is quite likely. The only thing better than "average FPS" is some kind of a graph. Saying "minimum FPS is the most important" is quite ignorant especially when you don't know how that minimum FPS was measured. At least average FPS has only 1 way in which it can be measured (total number of frames divided by total number of seconds).

Edited April 14, 2010 by galzohar

[bangtail 0]

But it's not that easy that you can just look at the minimum number.

For example it could be a case where the 5870 have one drop to 20fps in the whole benchmark and the rest is above 30fps, then the GTX480 have 3 drops to 25fps. You see what I mean, you can’t judge the performance by just looking at the minimum fps number. It’s better to look at a diagram over the whole benchmark where you can see all the fps drops.

I'm not in Sweden so I wouldn't know about prices there :D

No point is discussing this any further. I'll trust G3D over tpucdn.com (whatever that is) and I'll do my own testing when the cards arrive.

---------- Post added at 06:44 AM ---------- Previous post was at 06:42 AM ----------

Since there is no such thing as "FPS" in a given point in time, the definition of minimum FPS can actually change between different benchmarks. After all, all you can count is how many frames were rendered in the last X seconds or how much time have passed since the last frame was rendered. In between you could say all graphic cards have 0 FPS and infinite FPS for a infinitesimally small amount of time when the frame was rendered, thus have 0 minimum FPS and infinite maximum FPS. But all of this does not matter - what matters is whether or not you get low FPS for sustained periods of time.

If you get low FPS for less than 1/2 a second for the whole benchmark but the rest of the time your FPS is higher, then I'd say it's better. But if you get FPS drops for 1+ seconds every 10-20 seconds then that's a different matter.

A graph like the one in the above post gives a lot more info than just minimum/maximum/average, but it still misses some info like over what duration was the "momentary" FPS averaged (as there is no such thing as real momentary FPS, only X frames rendered within Y seconds). Though seeing the "jumps" in values are at least 1 frame, I assume the measurements were done every second (aka "how many frames were rendered during the last second").

One last thing, though - If you're going to look at only 1 number - Look at average FPS. After all, there aren't a lot of things that would make 1 card have more game-breaking FPS drops yet manage to maintain higher average FPS, but having some random element cause an FPS drop that lowered the minimum FPS for one card but not the other in an insignificant way is quite likely. The only thing better than "average FPS" is some kind of a graph. Saying "minimum FPS is the most important" is quite ignorant especially when you don't know how that minimum FPS was measured. At least average FPS has only 1 way in which it can be measured (total number of frames divided by total number of seconds).

It is the most important when if the average FPS are above 24 FPS, I should have quantified that. If your FPS is dropping below 24-30, then the minimum FPS is the most important factor and the more it drops below the minimum acceptable FPS (24-30) the more it is going to negatively affect the average. Minimum FPS is the FPS that will affect your enjoyment below a certain point (24-30 FPS). That is why enthusiasts consider it the most important number when considering with FPS.

Average FPS is very important but not game changing providing, again, that it is consistently above the acceptable norm. I have to quantify that again as I think that is where the argument is coming from. Obviously, if your max fps and average fps are low, then they can be just as telling but I have parameters for gaming that I rarely deviate from and a 25 max fps isn't going to fly for me. I'll admit, my opinions here are highly subjective but they are based on my own extensive experience. That experience has consistently reinforced the fact that minimum fps is the fps most likely to hamper your gaming experience when it drops below an accepted number.

From Tweakguides

There is definitely a 'minimum' FPS - a point at which, if the FPS in a game becomes low enough, your eyes and brain will begin to notice the fact that an animated image on the screen no longer looks smooth; it becomes quite apparent that it is a series of still images being displayed in sequence, like a slideshow. However the exact minimum number of frames required to keep things smooth in a game is not a set scientific value; since the human eye is involved, it will vary from person to person, and importantly it also varies by game type.

In my experience, the baseline for relatively smooth graphics is around 25 frames per second, if consistently rendered without stuttering.

Edited April 14, 2010 by BangTail

[Bush 0]

No point is discussing this any further. I'll trust G3D over tpucdn.com (whatever that is) and I'll do my own testing when the cards arrive.

I’m no "fanATIc", I never said that 5870 was better than GTX480 performance wise. You just don’t seem to understand that you can’t just look at the minimum fps and draw a conclusion that the framerate will be much more stable in the whole benchmark.

What is tpucdn.com? Never heard of it.

It is the most important when if the average FPS are above 24 FPS, I should have quantified that. If your FPS is dropping below 24-30, then the minimum FPS is the most important factor and the more it drops below the minimum acceptable FPS (24-30) the more it is going to negatively affect the average. Minimum FPS is the FPS that will affect your enjoyment below a certain point (24-30 FPS). That is why enthusiasts consider it the most important number when dealing with FPS.

No it's not the most important number. There is so much more to it than that.

[bangtail 0]

I’m no "fanATIc", I never said that 5870 was better than GTX480 performance wise. You just don’t seem to understand that you can’t just look at the minimum fps and draw a conclusion that the framerate will be much more stable in the whole benchmark.

What is tpucdn.com? Never heard of it.

No it's not the most important number. There is so much more to it than that.

You don't even know where you are copying and pasting from. That really instills confidence in your claims :rolleyes:

If your average fps is 40 and your max fps is 60 but your minimum fps is 18, guess which one is going to affect your viewing experience.

I'll give you a hint, it's not average or max.

Obviously if your average or max fps are below the accepted norm (24-30), then you have a point but I don't/won't play games where my max fps is 25.

I never said minimum fps was the only number to take into consideration. I said it was the most important number and it is AFAIAC.

Edited April 14, 2010 by BangTail