View Full Version : Futuremark patch 440 does not effect 5700u
bkswaney
11-15-03, 01:34 AM
Anyone know anything else about this?
It's a Club 3D card. :eek:
Futuremark patch doesn't affect Club 3D FX 5700 Ultra
In house test
By Fuad Abazovic in Wien: Friday 14 November 2003, 11:16
THE MAINSTREAM Geforce FX 5700 Ultra doesn't seem to be the slightest bit bothered about the latest Futuremark patch, it has emerged.
We are doing all we can to post a review of this card vs. Radeon 9600XT but in the meantime we took time to look deeper into this Futuremark patch 340 that is supposed to lower performance of Nvidia cards.
I haven't had time to test the FX 5900 non ultra the only high end FX part that I have, but here is what happens when you run this test on FX 5700 Ultra that was provided to us by the plucky Club3D company that is not even listed among Nvidia partners but still uses NVDA chips.
On the Club 3D Geforce FX 5700 Ultra patch 340 don't decrease performance in any way and even increases its performance by a very small margin on this 475/909 MHz clocked card.
I must admit that I had to retest this a few times to believe after I saw all this claims on the web, and was expecting that the FX 5700 Ultra would dramatically drop its performance in 3dmark03 but there it is. It's a fact.
Anyway only the game results and total score changed and in patch 330 3dmark03 scored 3961, Game 1 - 158.8, Game 2 - 25.9, Game 3 - 21.1, Game 4 - 22.0 while the other results were unaffected.
With patch 340 Club3D FX5700 Ultra scored 3965, Game 1 - 159.2, Game 2 - 26.0, Game 3 - 21.2, Game 4 - 21.9 so even better.
A full review will follow in the goodness of time, whatever that is. µ
bkswaney
11-15-03, 01:56 AM
Good ? .... They do not say. hummm :confused:
Here is the guys email address if someone wants
to try and get a reply via email. :)
fudo@theinquirer.net
GlowStick
11-15-03, 02:23 AM
Originally posted by bkswaney
Good ? .... They do not say. hummm :confused:
Here is the guys email address if someone wants
to try and get a reply via email. :)
fudo@theinquirer.net Actually i do like to email the writers at the inq alot, and they allways reply.
bkswaney
11-15-03, 03:22 AM
Originally posted by GlowStick
Actually i do like to email the writers at the inq alot, and they allways reply.
I sent him one a few min ago. :)
Not true, according to Baron: Don't know original results, but a 5700U at stock clocks gets 3347 (hey, wasn't that what my 9600 Pro got? scary).
http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=20856&perpage=20&pagenumber=3
http://service.futuremark.com/compare?2k3=1553380
Hellbinder
11-15-03, 06:40 AM
On the Club 3D Geforce FX 5700 Ultra patch 340 don't decrease performance in any way and even increases its performance by a very small margin on this 475/909 MHz clocked card.
I must admit that I had to retest this a few times to believe after I saw all this claims on the web, and was expecting that the FX 5700 Ultra would dramatically drop its performance in 3dmark03 but there it is. It's a fact.
If that is True and They have done everything correctly on their end then this indicates honest Compiler Optimizations for the 5700U. Which is a good thing.
52.16 of course, them and the 52.70 are the only drivers that support the FX5700.
The Baron
11-15-03, 11:00 AM
DaveB said Fudo screwed up.
digitalwanderer
11-15-03, 11:22 AM
Originally posted by Dazz
52.16 of course, them and the 52.70 are the only drivers that support the FX5700.
Unless nVidia gave 'em a newer beta set with a "fix" for the 3.40 patch. ;)
Besides which, I'm calling "bullspit!" on this story since no one else has been able to replicate it and in fact others have had a 13-15% score decrease on a 5700 Ultra using the 52.16 & 52.70 drivers. :cool:
I think FUDo did screw up and then got all excited and jumped the gun and posted the story a bit prematurely and is gonna be writing a tiny little retraction soon. ;)
Originally posted by digitalwanderer
Unless nVidia gave 'em a newer beta set with a "fix" for the 3.40 patch. ;)
NVIDIA talking to Fudo?
That's that'd be a first! ;)
Uttar
digitalwanderer
11-15-03, 11:48 AM
Originally posted by Uttar
NVIDIA talking to Fudo?
That's that'd be a first! ;)
Yeah, he probably just screwed up. :)
bkswaney
11-15-03, 06:09 PM
He told me he was using the 52.16's.
So, are you guys saying it's just the Club 3D card that keeps it's score with the 340 patch or it's all 5700 Ultras? If it's just the Club 3D card then I would say there is a bug in 3DMark03 that is failing to detect the card/driver and not turning off the cheats.
The Baron
11-15-03, 07:16 PM
Originally posted by ChrisW
So, are you guys saying it's just the Club 3D card that keeps it's score with the 340 patch or it's all 5700 Ultras? If it's just the Club 3D card then I would say there is a bug in 3DMark03 that is failing to detect the card/driver and not turning off the cheats.
no, we're saying that Fudo screwed up and it doesn't actually keep its score. mine is 600 points lower than Fudo's score, and all 5700Ultras are created equal (well, chip wise anyway, maybe not in terms of cooling).
I'm just saying perhaps the Club 3D card uses a different vendor id in their bios and 3DMark03 is just not detecting that (assuming Futuremark is detecting the id and turning off these cheats).
Chris you should have been paying more attention what you have suggested makes absolutely no sense, however since it seems clear it was all a mistake it is irrelevant
(as to why it makes no sense, Futuremark was not changed to detect nvidia cards and screw them, it was changed so that Nvidia cards (the driver) could not detect it, hopefully that makes sense to you)
Originally posted by sxotty
Chris you should have been paying more attention what you have suggested makes absolutely no sense, however since it seems clear it was all a mistake it is irrelevant
(as to why it makes no sense, Futuremark was not changed to detect nvidia cards and screw them, it was changed so that Nvidia cards (the driver) could not detect it, hopefully that makes sense to you)
Yeah, and just how do you know this? I didn't know you were part of the development of Futuremark and had the inside scoop on what methods they are using. :rolleyes:
If you would bother to click on the "Details" button in 3DMark03, you would know that they are indeed detecting the vendor id of the graphics card you are using. It is a common practice to use this id in certain programs. I should know as I have written several programs that does just that.
As for your confirmation of this, show me the link where it was proven that this is in fact not the case.
The Baron
11-15-03, 08:29 PM
are device IDs vendor specific if they're following the reference design?
cthellis
11-15-03, 08:42 PM
FM already announced they "turned off the cheats" so to say by reordering (not mathmatically altering) the shaders and renaming registers so there could be no shortcuts taken with shader detecting and knowing what would come next before the fact. Vendor ID has nothing to do with it, and is pretty much exactly against what they want. (They don't want to know what a card is, and they don't want a card to know what IT is, they just want the operations run to their specifications the best way the card can.)
I'm not saying this is what they are doing. I'm only suggesting the possibility that they are using this as part of whatever process they are using to detect these drivers. I don't even know what Club 3D's vendor id is. The way the story if written suggests that Club 3D cards are not losing points with the 340 patch. Maybe the guy screwed up...maybe not. Maybe it is something Futuremark did that is causing the cheats not to be blocked? Maybe, just maybe, Futuremark really does have some kind of vendetta against nVidia? Futuremark is not going to tell us what they are doing because they don't want to make it easier to defeat their protection mechanisms. They can't just detect the driver since that can be easily hacked to reflect another driver version. I'm just trying to keep an open mind here.
Originally posted by cthellis
FM already announced they "turned off the cheats" so to say by reordering (not mathmatically altering) the shaders and renaming registers so there could be no shortcuts taken with shader detecting and knowing what would come next before the fact. Vendor ID has nothing to do with it, and is pretty much exactly against what they want. (They don't want to know what a card is, and they don't want a card to know what IT is, they just want the operations run to their specifications the best way the card can.)
That was the method they announced they used the first time they patched 3DMark03. The very next drivers from nVidia released just a few days later just changed the order right back. It was not very wise for Futuremark to tell everyone exactly what they did as that made it much easier re-enable the cheats.
The Baron
11-15-03, 09:16 PM
Originally posted by ChrisW
That was the method they announced they used the first time they patched 3DMark03. The very next drivers from nVidia released just a few days later just changed the order right back. It was not very wise for Futuremark to tell everyone exactly what they did as that made it much easier re-enable the cheats.
um... you're missing the point. there's no active cheat protection. the whole point is to disable the cheats by changing the CRC of the .exe file, defeating the driver's ability to detect the application and then preventing shader detection and ultimately, replacement.
StealthHawk
11-15-03, 09:27 PM
Originally posted by The Baron
DaveB said Fudo screwed up.
The Inquirer making mistakes? Say it ain't so :rofl
cthellis
11-16-03, 03:48 AM
Originally posted by ChrisW
That was the method they announced they used the first time they patched 3DMark03. The very next drivers from nVidia released just a few days later just changed the order right back. It was not very wise for Futuremark to tell everyone exactly what they did as that made it much easier re-enable the cheats.
Also, now that they actually have rules and regulations in place, it will be up to them to STICK to them and not certify any patches or new drivers that bring in illegal optimizations. They have a "blank slate" state right now, but it's tenuous. If nVidia starts up again and FM starts waffling again... That will pretty much drive in the last nail.
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.