PDA

View Full Version : 3dmark03 Pro and GeForceFX Cards

StealthHawk
06-26-03, 06:26 AM
I think some of you guys have 3dmark03 Pro.

For those that do, and who also own a GeForceFX card, could you please run the test normally, and also enable the option that forces cards to use PS1.1 instead of PS1.4 and post the results?

Thanks.
Sazar
06-26-03, 12:00 PM
Originally posted by StealthHawk
I think some of you guys have 3dmark03 Pro.

For those that do, and who also own a GeForceFX card, could you please run the test normally, and also enable the option that forces cards to use PS1.1 instead of PS1.4 and post the results?

Thanks.

why ?

the scores will automatically be lower ?

experiment ?
Uttar
06-26-03, 12:18 PM
My bet: With cheating drivers, they'll be lower. With non-cheating drivers, they'll be higher.
Simply because with cheating drivers, nV is *already* using FX12 even though it isn't normally permitted by PS1.4. ( but it is by PS1.1. ) - and then, all you get is an additional performance hit of having to do multiple passes...


Uttar
Sazar
06-26-03, 12:20 PM
uttar... concerning the new 44.6x driver set... per the grapevine @ b3d... the scores are back @ pre patch 330 levels with the 44.03...

comments ?
Nutty
06-26-03, 12:32 PM
if FX12 isn't allowed in PS1.4, then what format did the R8500 use?
Uttar
06-26-03, 12:36 PM
Nutty: The R(v)2xx supports FX16. Yeah, small difference, but MS cares I guess :)

Sazar: My guess is that the 44.65 do the same thing as the 44.03 did. But there's another upcoming driver release which will retrieve a few 3DMark "optimizations", but keep most in.


Uttar
Sazar
06-26-03, 12:38 PM
Originally posted by Uttar
Nutty: The R(v)2xx supports FX16. Yeah, small difference, but MS cares I guess :)

Sazar: My guess is that the 44.65 do the same thing as the 44.03 did. But there's another upcoming driver release which will retrieve a few 3DMark "optimizations", but keep most in.


Uttar

seemed interesting.. since a lad with a 5900U on b3d posted a thread about perceived 'optimizations'

ah well..
SurfMonkey
06-26-03, 12:55 PM
Originally posted by Uttar
Nutty: The R(v)2xx supports FX16. Yeah, small difference, but MS cares I guess :)

Sazar: My guess is that the 44.65 do the same thing as the 44.03 did. But there's another upcoming driver release which will retrieve a few 3DMark "optimizations", but keep most in.


Uttar

The 44.65's don't seem to have the same problems as the 44.03's had in 3DMark, renaming it has no effect and there are no glitches when you go 'off the rail'. So either they moved really quickly in making sure these optimisations aren't as easily detected or they already had the code in place ready for such a situation.

Which means that if it was the former, then they just don't care about peoples perceptions of them or, if it was the latter, they were working with malice afore thought. They always knew they'd get caught eventually and were already set up to combat it. Either way it's quite sad. Now if these peformance boosts were legitimate then all well and good... but I don't think they are somehow.
Uttar
06-26-03, 01:39 PM
Hmm, likely the 44.65 includes the same changes as those upcoming drivers ( official release likely to finish by "7" ;) )

From my understanding, they managed to agree with 3DMark that some "optimizations" were "alright" ( that is, thanks to their legal department ) , but still some had to go.
95% of the "optimizations" are still there though I believe.

Just what part of "Everyone is being paranoid about cheating now and wants to trap you everywhere they can for the fun" can't nVidia understand? :(

As I said, they're footing themselves in the shoot.


Uttar
Nutty
06-26-03, 02:03 PM
Yup, even apple fudged the benchmarking of their new G5.

http://www.haxial.com/spls-soapbox/apple-powermac-G5/
Sazar
06-26-03, 02:47 PM
Originally posted by Nutty
Yup, even apple fudged the benchmarking of their new G5.

http://www.haxial.com/spls-soapbox/apple-powermac-G5/

they took it a step further... :)

apple == teh suck...

teh very GOOD LOOKING suck... but teh suck nevertheless :)
StealthHawk
06-26-03, 08:28 PM
Originally posted by Sazar
why ?

the scores will automatically be lower ?

experiment ?

Well, I would assume that running PS1.4 would give higher scores than running PS1.1. But until someone tests, there really is no clear cut evidence of that....and it is still an assumption.
Sazar
06-26-03, 11:04 PM
Originally posted by StealthHawk
Well, I would assume that running PS1.4 would give higher scores than running PS1.1. But until someone tests, there really is no clear cut evidence of that....and it is still an assumption.

I think it would especially count in gt2/gt3... due to fewer passes required to render same thing...

but should not matter for the dx9 cards... I don't think...
StealthHawk
06-27-03, 01:00 AM
Originally posted by Sazar
I think it would especially count in gt2/gt3... due to fewer passes required to render same thing...

but should not matter for the dx9 cards... I don't think...

Why wouldn't it matter for DX9 cards? I thought that the R3xx cards showed more of a performance difference between PS1.1 and PS1.4 than the R2xx did.
Sazar
06-27-03, 01:20 AM
Originally posted by StealthHawk
Why wouldn't it matter for DX9 cards? I thought that the R3xx cards showed more of a performance difference between PS1.1 and PS1.4 than the R2xx did.

k... I was not sure what you were looking @ in respect to the results...

I thought you were asking for something completely different...
StealthHawk
06-27-03, 07:55 AM
Originally posted by Sazar
k... I was not sure what you were looking @ in respect to the results...

I thought you were asking for something completely different...

What I want to investigate is whether gfFX cards use PS1.4 in GT2 and GT3, and whether there is any difference between PS1.4 and PS1.1 with those cards :)
tertsi
06-27-03, 08:23 AM
5800 Ultra and 44.65 drivers...

3DMark03 (330):

GT2 + PS1.4 - 37.1 fps
GT2 + PS1.1 - 29.5 fps

GT3 + PS1.4 - 31.5 fps
GT3 + PS1.1 - 25.4 fps

Have fun!
StealthHawk
06-27-03, 07:50 PM
Thank you very much for that Tertsi :thumbsup: