Catalyst drivers v02.2 up to 50% faster in Serious Sam...

[Guest]

This PDF from ATI says Catalyst drivers are giving up to 50% boost in performance.

http://mirror2.ati.com/drivers/Catalyst-02.2-Web-Posting-Release-Notes.pdf

Up to 25% improvement in Quake III (1600x1200, Max Quality)
Up to 35% improvement in Return to Castle Wolfenstein (1600x1200, High Quality)
Up to 50% improvement in Serious Sam: The Second Encounter 1600x1200, Extreme Quality)

Correct me if I am wrong but this is the largest performance increase in ANY driver set that I have ever heard of...

 

[stevem]

So - dare I say it - appart from "pom-pom waving"TM, has anyone tried these yet? Results? DaveB?

 

[Fuz]

It would be nice to see some numbers. Any one?

If there really is a huge increase in performance, well then you got to hand it to ATI for doing a great job.

If they keep this up for a year or so, maybe they will finally shake of their reputation for bad drivers. Hmmm...... imagine that!

 

[T2k]

I've tried: gained +100-120 3DMarks (yes, it's just D3D )

PS: Actually, currently I use my Parhelia and my tenant/friend borrowed my 8500 - he got this score on a 1800+/A7V266 config...

 

[Brent]

i didn't really have time tonight to test them in detail, but i did a few very quick benchies with (i know i know ) 3dmark2001se, quake3 and jedi knight 2

ever so slightly faster in 3DMark2001SE

not a lick of difference in Quake3

JDK2 same thing

3DMark2001SE

Catalyst 2.1

1024x768 - 8993
1280x1024 - 7422
1600x1200 - 5927

Catalyst 2.2

1024x768 - 9018
1280x1024 - 7459
1600x1200 - 5942

Quake3

Catalyst 2.1

1024x768 - 211.3
1280x1024 - 165.4
1600x1200 - 121.9

Catalyst 2.2

1024x768 - 210.9
1280x1024 - 165.4
1600x1200 - 122.2

Quake 3 4X AA

Catalyst 2.1

1024x768 - 43.5

Catalyst 2.2

1024x768 - 43.4

JDK2

1024x768 - 93.5
1280x1024 - 93.7
1600x1200 - 83.3

1024x768 - 93.1
1280x1024 - 93.3
1600x1200 - 83.1

Also, I had no problems with these drivers at all in the benchmarks i used above, no texture problems, no bugs, very stable. I installed these drivers clean though, with no previous driver having been installed. I have a ghost image i use that has all the benchmarks installed except it doesn't have chipset and video card drivers instlalled, so for each driver set I ghosted my machine for a nice clean driver install.

Sorry, I didn't have the time to test any further on it tonight

Test system is as usual

MSI K7T266 Pro 2
AXP 1800+
256MB PC2100 DDR
AIW 8500 128MB
Latest 4n1's

3DMark2001SE run with default settings in the 3 resolutions

Quake3 1.30 run with Q3Bench at Max settings in the 3 rez's

JDK2 run with max settings in the 3 rez's with the default demo run

 

[Dave Baumann]

So - dare I say it - appart from "pom-pom waving"TM, has anyone tried these yet? Results? DaveB?

I'm not going to start 'driver comparison' articles around here. However, ATi contacted me last night and said they had some stock of Radeon 9000 Pro's and (assuming their shipping dept doesn't mind shipping to the UK) I should be in line to recieve one some time next week. Naturally I'll be comparting it against the same ATi Radeon 8500 as was used in the recent Hercules 3D Prophet FDX 8500LE review so direct comparisons between the driver sets can be made from those.

I'll wager some other site will do a comparison before I get the Radeon 9000 Pro review up though.

 

[Reverend]

Thread title is wrong... it should be "...up to 50% faster in one game".

Sorry, couldn't resist

 

[Guest]

Thread title is wrong... it should be "...up to 50% faster in one game".

Sorry, couldn't resist

Hrm, to slim it down even further ..... in Open GL at high res. I really hadn't had much time to mull over actuall performance as there weren't any to spot. I just thought that it was news worthy, thats all. Appologies to anyone offended by ATI driver release.

EDIT: I changed the thread title.

 

[jvd]

it seems

it seems that you see the most gain in xp with sp1. the mozilla bug is still in this driver set . speed ups well i can see a little speed increase not much though a frame or two at most . Warcraft 3 and nwn speed up though and no longer crash for any reason .

 

[LittlePenny]

Well, I don't necessarily give two poops about this post, some poor bloke got yelled at not too long ago for having a same kind of thread about NV dets(although I can't find it, maybe I am nuts). So even though we have had driver threads before, for the sake of equality, I think this thread should drift off into oblivion.

 

[RussSchultz]

I'm all for deleting posts about driver comparisons from all vendors.

Every forum can't be everything to everybody, this one excels at things other than driver comparisons. Hardocp or Anand's probably has enough people who are interested enough to test the drivers on their rigs to make the discussion worthwhile, whereas here....

 

[jb]

Brent

Thanks for the work. I did not see any increase in those numbers but maybe on other systems they did eek out some sort of increase?

 

[Brent]

Brent

Thanks for the work. I did not see any increase in those numbers but maybe on other systems they did eek out some sort of increase?

who knows...

This evenning when I get home I will throw Serious Sam 2 on there and test out their claims of "50%" faster in SS2.... so we'll see, i'll test it in extreme quality mode with and without Aniso and AA...

that's about all i'll have time for tonight

 

[Gollum]

I can't seem to find the details, but wasn't there an article or post about how the performance of the R8500 based boards mysteriously sunk under specific circumstances at 1600x1200? IIRC the result was that the 2nd TMU apparently did not work at the specific resolution and/or graphic settings used (dunno, Aniso or FSAA might have been involved too), and thus in certain conditions and games, performance was almost halved to what it should be? Sounds to me like this "bug" was finally fixed in drivers and as such, under very specific circumstances, people will find a high performance increase (which is pretty insignificant, because even going from 8 to 12FPS doesn't make the game any more playable, as high as the "percentage" increase may be). I wish I could find the discusion and benchmarks that showed this, but even so I am pretty sure that my memory doesn't betray me. If it does, feel free to correct me...

 

[stevem]

I'm not going to start 'driver comparison' articles around here... ...I'll be comparting it against the same ATi Radeon 8500 as was used in the recent Hercules 3D Prophet FDX 8500LE review...

I suspected you may have tried them out as prescribed above. I don't have any ATI boards right now, so a non-trivial performance increase @ high res would have been interesting.

 

[WaltC]

Re: Catalyst drivers v02.2 up to 50% faster in Serious Sam.

This PDF from ATI says Catalyst drivers are giving up to 50% boost in performance.

Correct me if I am wrong but this is the largest performance increase in ANY driver set that I have ever heard of...

OK, I'll correct you....

Actually, a few years ago a company did boast of a "50% performance improvement" in a product brought about as a result of a driver release.

That company was......ATI (you probably guessed that already)...

It released a driver set called (it's been awhile so things are fuzzy) their "Turbo" drivers, and the speed increases concerned a 50% across the board improvement in a single, widely used benchmark at the time (it's on the tip of my tounge--but I can't remember what the bench was called.) Sure enough, it was proven that ATI's frame rate numbers did increase some 50% across the board for this particular benchmark.

However, ATI at the time advertised as though the drivers made its products run 50% faster across the the board, but the truth as it turned out was that ONLY the benchmark ATI cited actually produced numbers 50% faster. All the rest of the gaming software--and I mean all of it--showed performance numbers right in line with the previous driver release--and a few of them were even a tad slower.

The bottom line was that ATI produced a set of drivers deliberately rigged to throw back spurious performance results when the drivers encountered this specific benchmark--the drivers were deliberately written to know when the benchmark was being run and to behave in a predictable pattern as far as the results the benchmark expected to see supplied by the driver feedback. The polite term for what ATI did at the time was called "optimizing" the drivers for a benchmark--I think an accurate term might be "cheating".... There was a big stink at the time when this came out and its shadow hovered over ATI for quite sometime. The company ate humble pie, acknowledged the misrepresentation and apologized. Up to now, I thought ATI had learned its lesson.

But if what you say is true then it looks like ATI has found a way to spoof the SS benchmark and the company is up to its old tricks again. Generally, when things seem too good to be true they are.... 8)

To be perfectly fair, though, I believe that when nVidia first released its Detonator series of drivers they were advertising even higher percentages of improvements--but I can't really recall. I do remember though that those numbers were not repeatable, either.

 

[RussSchultz]

Didn't they (ATI) do that with Winbench also, with the MACH64? Or was that another company (DiamondMM?)

But certainly, benchmark cheating isn't something new at all, or that only a few vendors do.

 

[OpenGL guy]

Re: Catalyst drivers v02.2 up to 50% faster in Serious Sam.

But if what you say is true then it looks like ATI has found a way to spoof the SS benchmark and the company is up to its old tricks again. Generally, when things seem too good to be true they are.... 8)

How do you know that there wasn't some driver problem that was hurting performance? If the quality of the rendering isn't harmed, then what is your problem? Did ATi state a 50% improvement across-the-board? No. Maybe some hardware feature wasn't being used properly. There are a lot of possible explanations.

Bugs happen. It's a fact that software contains bugs. Even your simpliest "Hello World" program can have bugs because you are linking in libraries to make it link and the libraries may be buggy even if your simple program is not.

Not everything is a conspiracy.

 

[WaltC]

Re: Catalyst drivers v02.2 up to 50% faster in Serious Sam.

How do you know that there wasn't some driver problem that was hurting performance? If the quality of the rendering isn't harmed, then what is your problem? Did ATi state a 50% improvement across-the-board? No. Maybe some hardware feature wasn't being used properly. There are a lot of possible explanations.

Bugs happen. It's a fact that software contains bugs. Even your simpliest "Hello World" program can have bugs because you are linking in libraries to make it link and the libraries may be buggy even if your simple program is not.

Not everything is a conspiracy.

Generally, I tend to agree with you that "not everything is a conspiracy." But as this was ATI and I recalled a previous case in which the company admitted to such a conspiracy, and the poster asked if any other company had ever claimed a 50% performance improvement through a driver upgrade before, I felt it was prudent to answer him as I did.

Most of the time bug fixes are labeled as bug fixes--not performance enhancements. Indeed, as reported here by others who've tested the drivers and seen no improvements in the other games--certainly not the "up to 35%" and so forth the driver propaganda relates, it seems to sort of rule out any bug fixing because if that was the case we'd expect to see actual performance improvements we aren't seeing. Right now, it would appear that the performance numbers ATI is boasting for these drivers are only repeatable and obtainable by ATI. I don't know about you, but that kind of sets off my warning circuits....

But to answer your question more directly: the sort of behavior I relate isn't new to ATI. If it was I'd be inclined to give them the benefit of the doubt, certainly. As it is, it only seems like more of the "same old, same old" when they release drivers they trumpet and ascribe all sorts of fantastic performance increases to them that almost no one else can independently verify.

 

[OpenGL guy]

Re: Catalyst drivers v02.2 up to 50% faster in Serious Sam.

Generally, I tend to agree with you that "not everything is a conspiracy." But as this was ATI and I recalled a previous case in which the company admitted to such a conspiracy, and the poster asked if any other company had ever claimed a 50% performance improvement through a driver upgrade before, I felt it was prudent to answer him as I did.

Interesting. And I guess you forgot to mention nvidia's "miraculous" anisotropic filtering performance improvements? You want me to go into detail about that? I mean, I'm fully armed.

Most of the time bug fixes are labeled as bug fixes--not performance enhancements. Indeed, as reported here by others who've tested the drivers and seen no improvements in the other games--certainly not the "up to 35%" and so forth the driver propaganda relates, it seems to sort of rule out any bug fixing because if that was the case we'd expect to see actual performance improvements we aren't seeing. Right now, it would appear that the performance numbers ATI is boasting for these drivers are only repeatable and obtainable by ATI. I don't know about you, but that kind of sets off my warning circuits....

I haven't tested for myself, as I haven't downloaded the drivers, but I also don't have SS to check it. I have heard some people on Rage 3D saying that SS and other apps were working better.

But just because something isn't across the board, doesn't mean it's not working as intended. For example, one application might use a feature another does not.

And, I would call it a performance improvement if I found that something that was hurting performance but not visual quality. There are lots of ways that could happen.

But to answer your question more directly: the sort of behavior I relate isn't new to ATI. If it was I'd be inclined to give them the benefit of the doubt, certainly. As it is, it only seems like more of the "same old, same old" when they release drivers they trumpet and ascribe all sorts of fantastic performance increases to them that almost no one else can independently verify.

I see. So you're going to hold the "Turbo" driver thing against ATi for eternity? Dude, that was the Rage PRO. And have you tried to verify the performance improvements? Many people on Rage 3D are getting better results.