R600vsG80 Benchmark Comparisons From Any Game Welcome!

[Xion X2]

Razor, it seems to me that most of your problems w/ me probably go back further than these most recent benchmarks that I've been running. I went back and looked at that thread that you linked from back around the time R600 was released and think that I probably overstepped my bounds a bit. So for that, I apologize.

I honestly try never to overreact, but I was probably a little out of line w/ some of my comments. I had been back and forth with many around here, as R600 vs G80 talk had gotten really heated, and by the time you and I started discussing it I was already tired of getting into arguments w/ folks about it.

So for the sake of getting this thread back on topic I am apologizing for that. I won't, however, apologize for my methodology with any of the benchmarking, because I still believe it is a reliable practice.

Concerning the Test Drive benches, I did everything that you originally asked for. After showing comparison screenshots of a GTX owner here w/ that game along with it running on my setup, you said that the shots weren't taken in the same location which rendered the comparison invalid. That different locations and different car models caused the framerate to "vary widely"--those were your words. Given that the game takes place on the entire island of Hawaii, it can be a little difficult to match locations perfectly, so what I did is bench the game in 3 different races from 3 different locations to see if the average framerate matched my screenshots, and there was only a 2.6fps variation between them which says, for the most part, "yes." Not only that, but during races you often have 3-4 cars on-screen at once, something that would make the hardware take more of a performance hit than a simple joyride in the shots that the GTX owner had posted.

I have played that game a ton and know that the framerate stays consistent throughout, as I told you and Jas back over on Rage. You guys didn't believe that, and then I showed it to clearly be the case w/ my benching at 3 different locales with different car models. Taking that data to heart, we then know that my screenshots w/ the GTX owner become more valid after seeing the framerate stays consistent in different locations w/ different models on screen at once. We then know that we should be able to snap shots in different locations without it affecting the accuracy of our measurements.

Now you may disagree with my methods, but I know my stuff. I have been gaming for god knows how many years.. started out on the Atari 400 and been building computers for the last 5-6 years. I started out on a computer when I was about 10 years old, programming in Basic. The first thing that I always do when I get a new game is optimize it for my hardware by testing it with FRAPS at different settings. So there was no doubt in my mind that Sammy's screens and mine were a reliable test to go by. They weren't a perfect testbed given that I was overclocked and he wasn't, or that he was running at a slightly lower res than me, but in general it was fairly accurate because they ended up balancing each other out, or coming awfully close to it.

Now I hope that you and I can put the past behind us and start being reasonable w/ each other from this point forward. But I warn you, if you insist on continuing to be an ass around this place or shoving your point of view in my face without backing up your points, then you can expect a dogfight, pal. I can debate with the best of them and know just a little about computers.

[Razor1]

ya know if ya said that in the begginning of your posts at Rage, this would have never came up :). I accept your apology, and I apologize as well for my behavior, I know I over stepped my bounds, I was trying to be not so abrassive with you in the r600 vs g80 thread, but as it kept going I just lost my cool, can only take so much ya know. Lets make sure it doesn't happen again.

I agree your methods aren't the greatest and I usually don't care to much about that unless the results are skewed, but the end results, I agree with it, TDU is close on both cards give or take 5-10%.

[sammy sung]

See but this is were I think your off base, why does a gaming benchmark need to be done in a scientific manor? Now I'm sure you'll call me stupid but most reviews sites are done in sterile environments, meaning those machines aren't used every day like a normal joe would. I take these shots more literaly than I would a review site, how ever what I do is take shots from post like these, compare them to my results and those of legit review sites. I only have a few sites I would really say are legit by the way.

How ever you want to throw it, these tests that Xion, Sammy Sung and others have done mean as much as a review at Anandtech. No they may not be consistent or 100% accurate in their methods by testing the same areas but you get the general idea of a game running on a system that wasn't steralized and built soley for benchmarking a product. Not to mention I'm sure Xion is constantly tweaking his system which makes things ever more reliable because that means that with a little adjustment I can make my 8800GTX or x2900 run better.

Razor1 you really just seem like a NVIDIA fan boy who can't have his card's honor insulted. I don't think that for one minute Xion thinks that the 8800GTX is crap or any NVIDIA product for that matter. Sorry but I think your really here just to cause trouble and nothing more, just my opinion.

Yup,no one ,not even the owners of the XT are saying that a standard unclocked xt can compete with a GTX.The fact that they are overclocking their cards 25% allmost to reach GTX peformance cannot be forgotten here but at the same time its the whole point.Its obvious that THEIR cards can compete,xions card looks a bit stronger than the other 2900 posters here.I agree that the 2900 offers less performance per watt and mhz but it also costs less,as xion has shown it can be a contender in the right hands.

[Xion X2]

fivefeet8-- I will be glad to bench Lost Planet again w/ HDR on high this time. The reason I didn't is because Medium already just about blinds me on the snow level because it's so bright. But I'll run a quick bench w/ frapslog and see how much it affects performance.

I'll also be posting the Snow level benches shortly; I have them saved already and just need to host them.

[Razor1]

you should be able to turn down HDR levels in the source console, look for something called HDR level or HDR commands.

env_tonemap_controller
SetAutoExposureMax
SetAutoExposureMin
SetBloomScale


these are the ones you should play around with

[Xion X2]

Yup,no one ,not even the owners of the XT are saying that a standard unclocked xt can compete with a GTX.The fact that they are overclocking their cards 25% allmost to reach GTX peformance cannot be forgotten here but at the same time its the whole point.Its obvious that THEIR cards can compete,xions card looks a bit stronger than the other 2900 posters here.I agree that the 2900 offers less performance per watt and mhz but it also costs less,as xion has shown it can be a contender in the right hands.
Overclocking my card makes a very noticeable difference in Lost Planet. On the Snow level, I've seen an increase from 30-40% in some places where the snow flurries are flying all around you. It must really stress the hardware.

But Test Drive flies even at stock speeds.. averages right at 50fps which looked to be about what you were getting on your stock GTX when in driver mode, Sammy.

We should bench that one again but this time at the same res and at stock speeds. I have a feeling they'd be almost identical.

[sammy sung]

Yup,no one ,not even the owners of the XT are saying that a standard unclocked xt can compete with a GTX.The fact that they are overclocking their cards 25% allmost to reach GTX peformance cannot be forgotten here but at the same time its the whole point.Its obvious that THEIR cards can compete,xions card looks a bit stronger than the other 2900 posters here.I agree that the 2900 offers less performance per watt and mhz but it also costs less,as xion has shown it can be a contender in the right hands.


EDIT:
Regarding TDU i will do the same tests as Xion as soon as i get my mobo back,i took my screenies in the replay mode where more of the maps is shown/rendered and at speed up to 400khm so its not entirely a reliable comparison.Lets use a the same car or atleast one in the same class and race the same sp race and start the fps counter when the race starts ,with lets say a 3 minute duration.Would make it more reliable in my eyes.

[sammy sung]

Overclocking my card makes a very noticeable difference in Lost Planet. On the Snow level, I've seen an increase from 30-40% in some places where the snow flurries are flying all around you. It must really stress the hardware.

But Test Drive flies even at stock speeds.. averages right at 50fps which looked to be about what you were getting on your stock GTX when in driver mode, Sammy.

We should bench that one again but this time at the same res and at stock speeds. I have a feeling they'd be almost identical.


Yea ,i hope to get the mobo back next week,with luck i´ll be up and running again the same day if the green god is willing this time.And its not a surprise its doing well with TDU since it was developed for ati hardware to begin with,i'm a 360 owner nowadays and enjoy it immensly:)

[Obie]

Did anyone read this article (http://www.pcmag.com/article2/0,1895,2147064,00.asp)? I found it interesting to say the least. Honestly, are their any benchmarks in Vista? What about SLI and CF in Vista?
Here, look at the results (http://common.ziffdavisinternet.com/util_get_image/16/0,1425,sz=1&i=162825,00.gif) using Vista! :jammin: I honestly didn't know about that. However, I am still curious of single card results in Vista be it dx9 or dx10.

[Razor1]

yeah that benchmark is quite old, they did that along time ago probably a month there was a press release on it actually.

[sammy sung]

Did anyone read this article (http://www.pcmag.com/article2/0,1895,2147064,00.asp)? I found it interesting to say the least. Honestly, are their any benchmarks in Vista? What about SLI and CF in Vista?
Here, look at the results (http://common.ziffdavisinternet.com/util_get_image/16/0,1425,sz=1&i=162825,00.gif) using Vista!:super: I honestly didn't know about that. However, I am still curious of single card results in Vista be it dx9 or dx10.

Havent seen them but generally CF has often been more efficient than SLI,this however shows some inconsistensies with no info about wich drivers that were used and a odd gap in performance where the lower clocked nvidia rig scored higher than the higher clocked rig.The ati cards are the 1gb xtx version with no info about gpu clocks and so on,not much are known about their performance yet so this looks a tad unreliable.

[Xion X2]

Lost Planet - Retail | DX9 / XP
1680x1050, Highest Settings (HDR = medium)
8xAF, AA N/A
2900XT @ 900|975 | E6600 @ 2.4GHz | Catalyst 8.39


Again, cinematics were not benched in frapslog; they were skipped. What I did was play the snow mission up until you get to the cave, so the recorded framerate you see is based on gameplay.

Snow level
Avg: 55.3fps
http://i12.tinypic.com/6gv2xjc.jpg

Some screens:
http://i14.tinypic.com/6h767av.jpg
http://i19.tinypic.com/67hjimh.jpg
http://i7.tinypic.com/53f7l2f.jpg
http://i19.tinypic.com/4qqa343.jpg
http://i7.tinypic.com/628w39e.jpg
http://i13.tinypic.com/675sk9e.jpg
http://i11.tinypic.com/4q37mnl.jpg

[Xion X2]

Same settings except HDR on high. Still the Snow level. Takes ~4.5fps hit:

http://i7.tinypic.com/66w4h8i.jpg

I had a few screens that I snapped, but now I can't find them, sorry. Personally I can't tell a difference between the two and would rather leave it off since it costs a little in performance.

[Xion X2]

Some more random shots that I didn't include but thought were awesome:

http://i11.tinypic.com/4p9eyqe.jpg

http://i15.tinypic.com/53rwec8.jpg

http://i14.tinypic.com/67rdob8.jpg

http://i9.tinypic.com/4veae6c.jpg

[fivefeet8]

fivefeet8-- I will be glad to bench Lost Planet again w/ HDR on high this time. The reason I didn't is because Medium already just about blinds me on the snow level because it's so bright. But I'll run a quick bench w/ frapslog and see how much it affects performance.


HDR at HIGH shouldn't make anything brigher. It simply increases the color precision. The game uses Int10 HDR for it's Medium setting and FP16 for it's HIGH setting.

[Razor1]

well it could be a bug in the drivers that cause the over exposure, but he can use the console commands to drop the glare amounts if its happening.

[Xion X2]

well it could be a bug in the drivers that cause the over exposure, but he can use the console commands to drop the glare amounts if its happening.
Didn't see anything out of the ordinary w/ it on high.. it looked the exact same to me. Visually I could see no difference but performance was a little slower.. around 4.5fps on average according to the bench I posted above.

[Razor1]

well there are small differences from Int 10 to FP tonemapping, the color ranges for FP tonemapping are much greater they are preceptible to the eye but its not like its easy to see ;). Now there are instances where Int 10 will give artifacts where contrast is high, check Humus's int 10 or was it int 16, don't remember, demo for examples, when the sun goes partly behind a wall, the edge where the tone map glare cuts off sometimes gets blockly.

[Obie]

Havent seen them but generally CF has often been more efficient than SLI,this however shows some inconsistensies with no info about wich drivers that were used and a odd gap in performance where the lower clocked nvidia rig scored higher than the higher clocked rig.The ati cards are the 1gb xtx version with no info about gpu clocks and so on,not much are known about their performance yet so this looks a tad unreliable.
Unfortunately there aren't any SLI G80 vs CF R600 benchmarks using Vista. I hope to see some soon using Halo 2, Shadow Run and Lost Planet (all in Vista). That way it's clear which dual gpu setup is doing it's job better.

Xion, good job!! :thumbsup:

[sammy sung]

Lost Planet - Retail | DX9 / XP
1680x1050, Highest Settings (HDR = medium)
8xAF, AA N/A
2900XT @ 900|975 | E6600 @ 2.4GHz | Catalyst 8.39


Again, cinematics were not benched in frapslog; they were skipped. What I did was play the snow mission up until you get to the cave, so the recorded framerate you see is based on gameplay.

Snow level
Avg: 55.3fps


hehe you gotta love the japanese,its never too cold for cleavege:D

[Xion X2]

hehe you gotta love the japanese,its never too cold for cleavege:D
LOL.

There are no words for how awesome this game is, though. The demo didn't even break the surface. If you want a game that will really push your card to the limit, this is it. I'm running a few degrees hotter than I do on Oblivion with everything maxed.

I can't believe some of the stuff I'm seeing in this game. Truly amazing in both graphics and gameplay.

[Razor1]

Xion this is why I said we really shouldn't use Dirt as a good measurement of next gen games

http://translate.google.com/translate?u=http%3A%2F%2Fwww.computerbase.de%2Fart ikel%2Fhardware%2Fgrafikkarten%2F2007%2Ftest_ati_r adeon_hd_2400_xt_hd_2600_xt%2F&langpair=de%7Cen&hl=en&safe=off&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

the 320 mb gts gets beat down by a 1950xtx, how many times do ya think we would actually see that ;)

Gotta wait a month or two, till both camps get drivers up to snuff for Vista and Dx10, there are quite a bit of improvements on the horizon, but seems like more bug fixing on AMD's part, of course that is the first step before optimization, nV has been bug fixing alot longer, and lately they have been increasing performance as well with the 160 series of drivers.

[-Aerows-]

great Lost Planet screenshots with some humongous green alien boss looking thing in them



Is that in the demo? I do *not* recall seeing that..that looks great!

Particularly:

http://i9.tinypic.com/4veae6c.jpg

You know a game looks good when you feel the urge to go WTF!!?? is *that* thing?

[Xion X2]

It's the final version, and man, let me tell you something. I'm just a couple of missions in and have already seen some stuff since then that either matches that or puts it to shame.

This is one of the most badass games I have ever played. It is just filled with crazy aliens, incredibly detailed landscapes and huge boss fights. It's just awesome. It's like playing through a movie or something.

[JasonPC]

Unfortunately it's just way too short. I'm almost done with it :(

The multiplay scene is not too hot right now either.