R600vsG80 Benchmark Comparisons From Any Game Welcome!

[SLINROB]

The drivers I could believe but not the RAM. That won't make a 10FPS difference. You're all supposed to be benching @ 2.4GHz too, and you all have E6600s. Are you all benching at stock speeds on the Conroes like I am?

No, that run for me was at 3.3ghz with my CPU. Im back to a overclock now, had it running stock long enough and was ready to go back to a overclock. lol.

And I didnt mean the ram alone would make up for it, I realize that, I was meaning a combination of different things. From what I understand from someone else this demo gained next to nothing with the CPU being overclocked anyway because its solely GPU bound.

Get that bios tool man so you can overclock your CPU.

[Xion X2]

Backlash is running vista and massive card overclocks and if i look closely at robs cpu flag line it doesnt look like it says 2400 mhz,could be a crap pic perhaps.Madpistol isnt running the same res as me either and im not running the same driver as rob.Go figure that one out.AND i have slowass ram running at 533mhz.
Yeah, I'm taking the overclocks into account, but some of the stock speeds are really different from each other.

[sammy sung]

Yeah, I'm taking the overclocks into account, but some of the stock speeds are really different from each other.


I edited my post,looks like his cpu flag line says 3200mhz or something similar.
My GTX oc score is lower than his default,would be good if madpistol could run it again at 1680-1050 ,kinda hard to tell now.

[SLINROB]

I edited my post,looks like his cpu flag line says 3200mhz or something similar.


Its 3286 according to CPUZ. Might as well say 3.3;)

[Xion X2]

No, that run for me was at 3.3ghz with my CPU. Im back to a overclock now, had it running stock long enough and was ready to go back to a overclock. lol.

And I didnt mean the ram alone would make up for it, I realize that, I was meaning a combination of different things. From what I understand from someone else this demo gained next to nothing with the CPU being overclocked anyway because its solely GPU bound.

Get that bios tool man so you can overclock your CPU.
Well I'm going to, but it still won't help for comparison's sake. Not everyone can go to 3.6GHz on their processors. Sammy's already mentioned that his will only go to 2.7GHz with his current RAM that he's using, and Madpistol only has his OC'd to 3.2GHz if I recall correctly.

2.4Ghz is just a simple baseline that we all can use and is reachable for everyone. You can set up a profile easily in your BIOS to set your overclock back the way that it was.

Anyway, you all can run whatever clocks you like, but if we're going to be benching between the two cards I just think the CPU speeds should be equivalent is all. Kind of unfair, otherwise.

[SLINROB]

Well I'm going to, but it still won't help for comparison's sake. Not everyone can go to 3.6GHz on their processors. Sammy's already mentioned that his will only go to 2.7GHz with his current RAM that he's using, and Madpistol only has his OC'd to 3.2GHz if I recall correctly.

2.4Ghz is just a simple baseline that we all can use and is reachable for everyone. You can set up a profile easily in your BIOS to set your overclock back the way that it was.

Anyway, you all can run whatever clocks you like, but if we're going to be benching between the two cards I just think the CPU speeds should be equivalent is all. Kind of unfair, otherwise.


Im at 3286. No need for me to be at 3.6ghz, i did it for a while but did not like running that much VC through my chip.

Im not gonna be setting my CPU back to stock anymore since this last benchmark, you know how finicky this board is mike and if I mess with it too much........well you know.

[SLINROB]

Oh hell! hold on and I will run it at 2.4ghz real quick but after this Im back to a permanent OC.:D

[sammy sung]

I could try the 165.01 driver, comparing dates in the inf files between that and the 160.03 im using now it looks like the 165's is a older set so i would prefer to stay on these as theyve worked out great for me sofar.

[Xion X2]

Im at 3286. No need for me to be at 3.6ghz, i did it for a while but did not like running that much VC through my chip.

Im not gonna be setting my CPU back to stock anymore since this last benchmark, you know how finicky this board is mike and if I mess with it too much........well you know.
Yeah, I can understand if you don't want to. That board is really stupid when it comes to swapping back and forth between settings.

I'm not trying to be a prick, but I know you can see where I'm coming from. Overclocking a CPU can make a big difference w/ some of this stuff, and a better baseline for comparison is to just use stock speeds since everyone with half a brain cell can get their processors to stock speeds. :D

If you can save a profile for your current overclock (I could do it on that board, if I remember right) then you should be able to just load the profile back in after you're through benching and it will set everything back like it was.

[SLINROB]

Yeah, I can understand if you don't want to. That board is really stupid when it comes to swapping back and forth between settings.

I'm not trying to be a prick, but I know you can see where I'm coming from. Overclocking a CPU can make a big difference w/ some of this stuff, and a better baseline for comparison is to just use stock speeds since everyone with half a brain cell can get their processors to stock speeds. :D

If you can save a profile for your current overclock (I could do it on that board, if I remember right) then you should be able to just load the profile back in after you're through benching and it will set everything back like it was.

No problem, Here are my results running stock everything....CPU and all.



Just like I thought its solely GPU bound. No physics or AI going on and thats why.



http://i197.photobucket.com/albums/aa160/SLINROB/sfdgsdfgdfg.jpg

[sammy sung]

Crap this heat is annoying,its 7:28 am here now and ist relatively cool atm but idle temps are 7-9 c higher across the board than my normal 32cpu 25nb and 51 gpu.2 days ago at 5 am when the sun had pounded my window all afternoon(106F in the sun),turning my apartment into a greenhouse in the process they were 47 35 and 68 :thumbdwn:

This is how it looks now at 26 c ambient temps:

http://img231.imageshack.us/img231/8505/temphr1.jpg

[Xion X2]

No problem, Here are my results running stock everything....CPU and all.


Cool. I wonder what drivers MadPistol is running to score 9FPS lower than you at the exact same settings and hardware on this test?

I wonder why the 165.01s were never official releases on Nvidia's site? They seem to be offering much better performance than any other Nvidia driver out there.

Mad--are you sure that you have everything set up the same as us? Make sure that you have AA and AF set to "application preference" in your control panel and transparency AA turned off. Also make sure that you're running the Direct3D graphics instead of OpenGL. It gives you the option of choosing either in the benchmark.

[sammy sung]

Cool. I wonder what drivers MadPistol is running to score 9FPS lower than you at the exact same settings and hardware on this test?

I wonder why the 165.01s were never official releases on Nvidia's site? They seem to be offering much better performance than any other Nvidia driver out there.

Mad--are you sure that you have everything set up the same as us? Make sure that you have AA and AF set to "application preference" in your control panel and transparency AA turned off. Also make sure that you're running the Direct3D graphics instead of OpenGL. It gives you the option of choosing either in the benchmark.

ok ill try them and compare to the 160.03,back asap.

EDIT: Weird,theyre no longer available at guru3d.And another thing Xion ,madpistol is NOT using the same settings as us,think his test is 1600x1200.

[Madpistol]

The test was done on the 158.22's. The only reason I'm using these now is because I've been trying to run up my 3dmark scores some, and the 158.22's are the latest approved driver by FM.

However, that is an interesting find that my computer is actually scoring lower than SLINROB even though our systems are virtually identical. I will have to try out those new drivers at some point.

Just ran through the settings in NV control panel, and everything seems to be normal. AA and AF are application controlled and Transparency AA is set to off. Nothing looks screwy. It just seems that the 165's are a lot better.

Oh well. I'm tired. Good night. ;)

EDIT: I just made a good observation. Ya'll are testing @ 1680x1050, while i'm testing @ 1600x1200.

1680x1050 = 1764000 pixels

1600x1200 = 1920000 pixels = higher res

After some quick calculations, I figured out that 1600x1200 has ~ 8.84% more pixels than 1680x1050.

But to be fair, I will run the benchmark @ 1680x1050 right quick.

[Madpistol]

Done. Here's the results:

http://img522.imageshack.us/img522/4892/main2007061001334589yr3.jpg

So the 165's are yielding a 1.1 fps increase, not 9. That seems a little bit more resonable.

[sammy sung]

Yea i kinda figured that was the case Madpistol and your scores are more in line with ours now:)

Tried the 165.01 now but they simply refuse to run in full screen 1680x1050 with 4xAA,keept locking up it i try.I could run it in windowed mode but thats all

http://img87.imageshack.us/img87/2277/unigineni7.jpg

[Xion X2]

Well like you said before, Madpistol, you learn something new every day. I had no idea that 1600x1200 would take so much more of a performance hit than 1680x1050.

Well at least we know now that your system is running like it should. Thanks for going back and benching again. :)

[Trademark]

I'll see if I can get my 26inc monitor positioned today, I'll put my cpu clock to 2.40Ghz aswell. Because the results are a "slight bit" unfair lol

[SLINROB]

I'll see if I can get my 26inc monitor positioned today, I'll put my cpu clock to 2.40Ghz aswell. Because the results are a "slight bit" unfair lol

This benchmark is all GPU. Going from 2.4ghz to 3.3ghz on my CPU gained me 1 whole FPS lol.

Its like this because its just a graphical show case which contains no AI or Physics/Stuff blowing up....etc..


I can see Xions point though that if we are benchmarking for comparisons with our 8800GTX's vs his 2900XT then the CPU needs to be at 2.4ghz so all things are equal.


I really was not trying to be unfair about it the first time I ran the test, I only became interested in it because I had never heard of it before and thought I would give it a run.



Well, just to make sure everyone knows all my other test were done at CPU stock speed and I even went back and ran this benchmark at stock speed as well, Im glad I did because it proved to me that the demo was GPU limited as I was thinking.


From here on out guys Im back at 3.3ghz and am not touching anything, this board is very very finicky and I dont like messing with the bios too much.

[sammy sung]

Did some 3dmark runs with the 165.01 drivers vs older runs with 97.92's i did in january.This doesnt really support claims of huge speed increases in nvidias later drivers,not with 3dmark anyway .Re-ran them with the same settings with my cpu at 2.7ghz /8800GTX stock.Not any huge gains but the 165.01 is the fastest set sofar.Some improvements just the same and in my experience the biggest gains have been made in the minimum fps on the G80 drivers,i dont have any sudden drops into the teens anymore in any of the games i play.

3DMark01 97.92 :49506 165.01 :50332

3DMark03 97.92 :35829 165.01 :38776

3DMark05 97.92 :15696 165.01 :15947

3DMark06 97.92 :10421 165.01 :10818


I also found my oldest saved bench of 3DMark01 with the awesome score of 2565 :p

http://service.futuremark.com/compare?2k1=490546

Date: 2001-03-19
CPU: Intel Pentium III 1132 MHz
GPU: NVIDIA GeForce2 GTS/Pro/Ti
OS: Microsoft Windows ME
Res: 1024x768 32bit

I have scores saved from GeForce2 GTS,GeForce3,GeForce4 Ti 4600,GeForce FX 5900 Ultra,RADEON X850XT up to the 8800GTX with cpus from pentium 3 ,xp1800,xp2700 in various clocks.Interesting how much progress have been made over the years.

[Xion X2]

You need to test it out with actual games, though. Go bench Oblivion on both drivers and see how much of a difference there is. Also bench Rainbow Six Vegas if you have it, because I know for a fact that the game used to slow down to 45FPS on a single GTX any time that you put the camera on your ally soldiers. It doesn't do that anymore, as SLINROB showed earlier. It flies along at 70+FPS like it doesn't even phase it.

Trust me, later drivers (158.22 and later) have made some very significant performance increases in games across the board. This isn't some lame BS excuse that I'm making up in order to put the R600 in a more favorable light. It's the truth.

In fact, my 2900XT is just slightly faster w/ Rainbow Six Vegas than my 8800GTX was on the 158.19 and earlier drivers. If it weren't for these latest drivers that Nvidia has released, these benchmarks between the two cards would be a lot closer on the whole.

When you look underneath the surface it makes perfect sense that Nvidia would stall performance increases in the drivers until the competition released their new flagship. It's a good surprise tactic to use in a business sense.

[sammy sung]

You need to test it out with actual games, though. Go bench Oblivion on both drivers and see how much of a difference there is. Also bench Rainbow Six Vegas if you have it, because I know for a fact that the game used to slow down to 45FPS on a single GTX any time that you put the camera on your ally soldiers. It doesn't do that anymore, as SLINROB showed earlier. It flies along at 70+FPS like it doesn't even phase it.

Trust me, later drivers (158.22 and later) have made some very significant performance increases in games across the board. This isn't some lame BS excuse that I'm making up in order to put the R600 in a more favorable light. It's the truth.

In fact, my 2900XT is just slightly faster w/ Rainbow Six Vegas than my 8800GTX was on the 158.19 and earlier drivers. If it weren't for these latest drivers that Nvidia has released, these benchmarks between the two cards would be a lot closer on the whole.

When you look underneath the surface it makes perfect sense that Nvidia would stall performance increases in the drivers until the competition released their new flagship.

That CAN also be because of them wanting to improve stability firsthand and improving performance second.As i mentioned this was in 3dmark only and since i have NO intention to install 97.92 again i wont test them with the games you mentioned,returning to BSOD's and random reboots isnt anything i would like to do just to prove a point:) If someone else wants to thats fine with me :p

[Xion X2]

That CAN also be because of them wanting to improve stability firsthand and improving performance second.

Well that's another view, but in order to believe that you must believe that everything just fell into place perfectly. Because G80 drivers just suddenly started yielding performance increases upon R600's release.

So, great timing on their part, in other words. Quite the coincidence, especially when they went 4 months without a single driver release in XP between January-April. Remember that? Where were all the drivers during that period that would've brought better stability? Looks to me like they were stalling.

[sammy sung]

Well that's another view, but in order to believe that you must believe that everything just fell into place perfectly. Because G80 drivers just suddenly started yielding performance increases upon R600's release.

So, great timing on their part, in other words. Quite the coincidence, especially when they went 4 months without a single driver release in XP between January-April. Remember that? Where were all the drivers during that period that would've brought better stability?

To be honest i dont buy that last bit about nvidia deliberatly holding off on drivers to check what the competition had to offer in terms of performance,i mean most people here are still using betas that's been out weeks before the first 2900 benches started to surface.Its just their official drivers thats been dragging and some of the betas are buggy,the 160.xx shows weird lighting bugs in oblivion etc and they werent really in shape to officially released when they were available as betas.And i dont know about these either,how perfected they are since i havent gamed with them yet but MANY of the betas had the slowdown bugs like in the flagroom on the citadel map ut2004.

[Xion X2]

To be honest i dont buy that last bit about nvidia deliberatly holding off on drivers to check what the competition had to offer in terms of performance,i mean most people here are still using betas that's been out weeks before the first 2900 benches started to surface.Its just their official drivers thats been dragging and some of the betas are buggy,the 160.xx shows weird lighting bugs in oblivion etc and they werent really in shape to officially released when they were available as betas.
And how many review sites that compared the two cards on R600 launch day were using those betas from long before the 2900XT surfaced for G80? Most of them were using new driver releases. The key was not necessarily the user-base but publicizing the difference in performance. :)

This would be the perfect move if Nvidia wanted to surprise ATI a bit and rub some unexpected performance in the face of the competition to make them look inferior for all to see.

And how do you explain why they went 4 whole months inbetween driver releases on XP? I have never, ever seen anything like that happen before. I mean, there weren't even any betas out during that time. Nada. Zilch. Nothing. Not a peep from Nvidia until 4 months later--around the time that R600 was expected to release.

If over that time they were focused on improving stability as you say, then where were the new drivers inbetween January-->April? Surely they could've improved on a few things during that time period.