PDA

View Full Version : Driver comparison: 67.65 vs 71.24 vs 71.50


boro
01-15-05, 09:15 AM
I compared some of the latest nVidia drivers in terms of performance.

All test are on a 2450MHz AMD XP, 1 GB RAM and a geforce 6800GT at stock speeds and default driver settings.


Chronicles of Riddick

67.65 : 76.2 fps
71.24 : 76.6 fps
71.50 : 77.3 fps

Settings: 1280*960, SM2, 0xAA/16xAF, custom timedemo on security research 2 map


Painkiller: Battle out of Hell

67.65 : 33.1 fps
71.24 : 33.9 fps
71.50 : 34.8 fps

Settings: 1600*1200 4xAA/16xAF, custom timedemo with heavy flamethrower action on Leningrad map


Doom 3

67.65 : 70.8 fps
71.24 : 74.8 fps
71.50 : 75.6 fps

Settings: 1024*768 Ultra quality 4xAA/8xAF, timedemo demo1


Enemy Territory

67.65 : 30.9 fps
71.24 : 31.0 fps
71.50 : 31.0 fps

Settings: 1600*1200, 8xSSAA/16xAF, custom timedemo on Oasis map with many airstrikes. This game is normally CPU limited but SSAA changes that :)


Shadermark 2.1

67.65 : 157 fps
71.24 : 160 fps
71.50 : 161 fps

Settings: default settings, SM3.0, 0xAA. Result of shader test 19 (combination effect)


3DMark05

67.65 : 4489
71.24 : 4799
71.50 : 4795

Settings: default


Aquamark

67.65 : 53,273
71.24 : 53,931
71.50 : 53,410

Settings: default


Videocard temperature (core/ambient)

67.65 : 80C / 51C
71.24 : 78C / 50C
71.50 : 80C / 52C

These are max temps which always occurred during Shadermark. Actual max temperature in games is quite a bit lower. Note: 70.24 was tested first on a cold PC, maybe that's why it's lower.


Conclusion

Out of these 3 drivers,71.50 takes the performance crown, which probably also means it is the fastest driver yet. Only in 3DMark and Aquamark was it slower than 71.24. Dispite of other people's findings I couldn't notice a real difference in temperature between these drivers.

Clay
01-15-05, 09:17 AM
Thanks for posting your findings. :thumbsup:

saturnotaku
01-15-05, 09:18 AM
That's all very well and good, but the 71.40 and 71.50 drivers have the yellow water problem in HL2. That's precluding many of us from using those drivers. It's a shame too because performance and IQ from both those sets are extremely good otherwise.

Clay
01-15-05, 09:22 AM
That's all very well and good, but the 71.40 and 71.50 drivers have the yellow water problem in HL2. That's precluding many of us from using those drivers. It's a shame too because performance and IQ from both those sets are extremely good otherwise.
Based on his results above I'm assuming that he doens't play HL2 anyway. :D But you're correct of course. Hopefully we'll see a yellow water fix in FW soon.

wEEt
01-15-05, 02:14 PM
Yeah, I like the 71.50, too. But my temps... :|

ericdrum
01-15-05, 04:48 PM
That's all very well and good, but the 71.40 and 71.50 drivers have the yellow water problem in HL2. That's precluding many of us from using those drivers. It's a shame too because performance and IQ from both those sets are extremely good otherwise.

That's what is keeping me from joining the party. Does nVidia know about the yellow water?

Electronic Punk
01-15-05, 05:08 PM
Drivers compiled since the 10th January will be free of the hl2 water bug.

Sazar
01-15-05, 05:09 PM
could the yellow water just be a result of uric acid combining with the water?

if that is the case (and it may well be since there are so many sewer pipes draining into the water) is it not possible that nvidia in fact is rendering the scene correctly right now with the yellow water?

would make sense to me anyways

:angel:

Graphicmaniac
01-15-05, 06:28 PM
What about IQ? is it true that 71.50 have the best IQ ever?

about yellow water, i think that if i was a HL2 player i would not care so much about it and i'll choose the fastest drivers instead of the ones who give me white water and anyway, if i'm not wrong, it is yellow just in one particular point.

Electronic Punk
01-15-05, 07:34 PM
could the yellow water just be a result of uric acid combining with the water?

if that is the case (and it may well be since there are so many sewer pipes draining into the water) is it not possible that nvidia in fact is rendering the scene correctly right now with the yellow water?

would make sense to me anyways

:angel:

If it appeared with ATI cards I would guess that would be right (as ATI got alot closer to Valve than NVIDIA did :))

ericdrum
01-15-05, 08:36 PM
could the yellow water just be a result of uric acid combining with the water?

if that is the case (and it may well be since there are so many sewer pipes draining into the water) is it not possible that nvidia in fact is rendering the scene correctly right now with the yellow water?

would make sense to me anyways

:angel:


See here:

http://ericgchristensen.com/coastBug.jpg



http://ericgchristensen.com/coastNoBug.jpg

Sazar
01-15-05, 08:38 PM
dang... thats looks like fugleh algae :eek:

rewt
01-15-05, 08:51 PM
Well, with 67.03 you get water banding. 71.24 you get piss water. Looks like a no win situation to me.

Graphicmaniac
01-15-05, 09:32 PM
but with 71.50 isn't the yellow water completely gone?

SquireSCA
01-15-05, 09:59 PM
Out of these 3 drivers,71.50 takes the performance crown, which probably also means it is the fastest driver yet. Only in 3DMark and Aquamark was it slower than 71.24. Dispite of other people's findings I couldn't notice a real difference in temperature between these drivers.

Performance crown?

Dude, the differences are all less than 1% which is way within the margin of error...

OWA
01-15-05, 10:06 PM
but with 71.50 isn't the yellow water completely gone?
The problem is still there with the 71.50s (at least for me).

mikedep333
01-15-05, 10:07 PM
Drivers compiled since the 10th January will be free of the hl2 water bug.
EP, why must you torture us with your inside information without releasing the drivers?
Well, ok, I do actually like to know this info, thanks for at least releasing this tidbit.

ericdrum
01-15-05, 10:29 PM
I just installed the the 71.20's and I DON'T get yellow water. Weird stuff.

superklye
01-15-05, 10:50 PM
Well, with 67.03 you get water banding. 71.24 you get piss water. Looks like a no win situation to me.
67.02g work great. :)

TierMann
01-16-05, 01:24 AM
Performance crown?

Dude, the differences are all less than 1% which is way within the margin of error...

I agree with that. A 1.1fps difference isn't much. You could run the same bench 3 times with the same drivers and come up with the same gap.

It does show 67.65 falling short on a few of those benches/demos though. And for me, Doom 3 performance does seem to drop a bit with that set.

I'll be sticking with 71.20b in windows for now.

boro
01-16-05, 04:35 AM
Performance crown?

Dude, the differences are all less than 1% which is way within the margin of error...

You are ofcourse free to make your own conclusions. I did repeat some benches to see if they were consistent, and they were.

K007
01-16-05, 05:08 AM
ure core is 80c on idle?!?! Holy ****.

SquireSCA
01-16-05, 11:16 AM
You are ofcourse free to make your own conclusions. I did repeat some benches to see if they were consistent, and they were.

Yes, but is there really a difference betwwen 71.4fps and 72.3fps?

I mean seriously, is it even worth the trouble to download and install a new beta driver just for an extra fraction of a frame rate?

I just don't get it, I guess... If new drivers gave you a 5-10% increase in performance without sacrificing IQ, then I am there. But it is almost comical to see people scrabbling in a feeding frenzy every time a new beta drive comes out and gives a 0.2% performance gain...

If that kind of gain is enough to make a difference, then perhaps a hardware upgrade is in order... hehe

boro
01-16-05, 03:06 PM
Yes, but is there really a difference betwwen 71.4fps and 72.3fps?

I mean seriously, is it even worth the trouble to download and install a new beta driver just for an extra fraction of a frame rate?

I just don't get it, I guess... If new drivers gave you a 5-10% increase in performance without sacrificing IQ, then I am there. But it is almost comical to see people scrabbling in a feeding frenzy every time a new beta drive comes out and gives a 0.2% performance gain...

If that kind of gain is enough to make a difference, then perhaps a hardware upgrade is in order... hehe

Well there IS a 5%+ difference in doom3, painkiller and 3dmark between 67.65 and 71.50.

I don't install nor benchmark every new driver btw, I just wanted to know what the difference was between the latest stable 6x.xx driver (12/31/04) and the newest 7x.xx(12/22/04). I thought I post my findings on a nvidia driver board. Without any frenzy. OK ? :rolleyes: