Article:Ati Drivers Put To The Test

The_Neon_Cowboy

As you were posting I was testing... allready!!!

Chaos

great stuff Neon....do they have a modified ati2mtag file for the soft mod ?

ZzzombieBunny

Would be interesting to see if theres any differnce between the Omega 3.8 and 3.8.5

Judas

...wow...that is fairly hot...

Vehicon

WOW, so grear info here!Im staying away from the 3.8's for now.

Mikmouse

Hi, my first post

I use a Sapphire 9600 Pro Ultimate (Heat pipe) with the latest Omega drivers.
Should I downgrade to Catalyst 3.8. ?
I am just a girl wihout a good knowlegde of hardware and wont destroy my new graphic card.!!
I don't overclock the VPU nor memory and have no way to know the Temperature on the card.
Any suggestion welcome
(Please excuse my bad English)

D0T-C0M

Nice job The_Neon_cowboy .

While I was looking over your pics at the temps and whatnot my girlfriend walks in and says jeez he should put his dirty clothes in the hamper before displaying that to the whole world. hahahaha I had to laugh cause it would take a woman to notice that. Her response "typical men" hehehe . Then she wonder why I'm always on my computer.

curiophile

I agree with this guy - I wonder if the VPU on or off would make a difference with the temps.

WaltC

Zero game bugs here with the 3.8's. To me they seem a cut above the 3.7's. No temperature problems of any kind--and 3dMK03, including the Nature Test, runs flawlessly, every time.

I can't see what there might possibly be that would be confusing about this situation. What TM from ATi actually said, if you read the post literally and absorb it properly, was that they detected a *slight* (tiny, miniscule, microscopic, etc.) temp rise for the 3.8's running the Nature Test--but that because it was such a *small* increase, it was "well within the normal operating specs for the product." What that means is that ATi has officially said that it has been unable to find any temperature problems relating to using the 3.8's with any of its products. That is what they said, so there shouldn't be any confusion about that. Detecting a tiny temp increase running the Nature Test in 3dMK03, which doesn't come close to exceeding design parameters, let alone cause any crashes, does *not* constitute a problem with temps and the 3.8's. You are reading much more into ATi's statement on the matter than ATi actually said, if you interpret ATi's remarks to mean in any way, shape, or form that people shouldn't use the 3.8's because of temperature spikes. ATi was clear in saying it could not find any substance in such claims in its own testing.

The rest of this post is not addressed to your comments, but to many in general in this thread...

I did find it interesting that both the Omegas and the "leaked" Asus drivers cause hikes in temps--hikes which the standard 3.8 CATs did not cause in NC's tests.

I really wish people could grasp the notion that a "modded" driver is not automatically a "better" driver. In truth, there is at least an equal chance that modifications done to manufacturer drivers will *degrade* the performance and/or image quality of the manufacturer's drivers, as well as introducing bugs not found in the reference drivers or actually eliminating bug fixes that are found in the standard set of drivers as released from the manufacturer.

People should take a look at the list of bug fixes in the read.me's which accompany a standard Cat driver release. It's surprising that more people don't realize that when you mix & match older driver components with the new ones, and make changes to the .inf setup files that come with the standard drivers, as Omega drivers typically do, that many of the bug fixes addressed in the new set actually *remain* in the modded drivers, and because the driver files themselves are from *mixed* releases, *new bugs* may actually be introduced that do not exist in *any* of the released manufacturer driver sets.

Definitely, "modded" does not mean "better," and the more refined a manufacturer's drivers become with each release the greater the probability that driver mods will result in *worse* drivers than what the manufacturer provides. The amount of pre-release testing that ATi does on each driver set is easily hundreds of times in excess of what Omega does prior to releasing his modded drivers--I think Omega uses his audience as his "testing" ground and relies on their feedback to "fix" things... Oh--and you can throw the WHQL certification right out of the window when running modded drivers--you no longer have it.

Also, I chuckle every time I hear someone say, "I'm running the 3.8 Omegas!" Huh? The only entity which produces the 3.8 CATs is ATi. What Omega does is to release a *bastardized* set of Catalysts, including driver files from *older driver sets,* which he says is *based on* the official ATi CAT release. Be assured that running the "3.8 Omegas" is absolutely not the same as running the 3.8 CATs as released by ATi. It's like the difference between reading books and then seeing movies "based on" those books. Often there is quite a difference, and rare it is when the movie is better than the book it's based on. The simple truth is that when you run a set of Omegas you don't know *what* you are running--regardless of the Catalyst number attached to them. The only thing you can be sure of is that you are not running a particular version of the Cats as released by ATi.

Having said the above, as long as people use these drivers with their eyes open and understand what they are--that's fine. But I really hope the idea that modded=better is put on the back shelf where it belongs because often modded=worse, instead.

Next we move to the concept of the "leaked" driver--the source of which is, of course, unknown. It may be an incomplete beta driver which has been leaked, and so therefore is buggy, or it maybe someone's own "modded set" of drivers which they want to see people use so they change file dates, mix and match driver components between versions as Omega does, etc., and upload them to a server on the Internet where they are called "leaked" in the hopes of getting people to use them. It is very naive to assume when installing a "leaked" set of drivers that you are simply getting something "better" than what you have but are getting it "early" because it's "leaked." The fact is you don't know what you are getting. I've often thought that if I was a virus writer I'd put some nice ones into a set of Detonators and upload them to some obscure site somewhere and call them a "leaked" set of "new" Dets... Spreading your virus is virtually guaranteed using that method.

Take the instance of Dell drivers. Why would anyone want Dell drivers? The products Dell orders from ATi are generally *not* the same products available on retail store shelves. They are OEM products specifically designed to Dell's specifications so that Dell can buy them cheaper than the retail versions but yet advertise them on its site as the same thing, but with slight nomenclature differences--like for instance the "9800TX" (not to be confused with the retail XT from ATi.) Often these products require different drivers because they use cores clocked lower than the retail versions, often use slower/cheaper ram, may even use pcbs using fewer layers (because such pcbs are also cheaper) because the OEM products may not need to draw as much power as their retail counterparts. In short, if you use a retail version of an ATi product, especially one BB ATi, you have a better, higher-quality product of a higher cost, for which ATi makes a *different* set of drivers than it provides Dell for its OEM products. I feel like I need to use Dell drivers about like I need a hole in the head...

Last, looking at the "Asus" drivers--again, being "leaked" there's no way to verify what state they are actually in--it's really not uncommon for board OEMs like Asus, or Guillemot, to "fudge" slightly on the default clock rates their drivers use. It's not uncommon at all for such drivers to set core and ram clocks 5MHz-20MHz higher than manufacturer defaults for the reference design, in the hopes that they will garner higher performance numbers in benchmarks than the cards built strictly to manufacturer specs and thus provide an incentive for people to purchase their cards. There may be other advanced timings set in these drivers that are not user-adjustable and are intended primarily for a specific bios on a specific card--and thus won't run as intended on another manufacturer's card. With such leaked, board OEM-specific drivers there's really no way to verify *what* the drivers are doing (again, assuming the "leaked" claim is at all valid.)

Basically, it seems to me, that with ATi releasing a WHQL-ed officially supported set of CATs every month now, there's just no reason to use anything other than the official CATs. Really, if you look at the Detonator situation, with nVidia releasing only 3-4 officially supported drivers per year, and only half of those at best being WHQL-ed (which only means that Microsoft certifies them to run properly with Windows), you can see why nVidia owners are so hot for "leaked" drivers--they are starved for driver releases. If you are an ATi customer you aren't in that position relative to official driver releases and support, not to mention WHQL support, and I would think that going the "leaked" route would cancel out many of the advantages of ATi's driver program relative to ATi's competition.

I've run nothing but official CATs since September '02, with both a 9700P and a 9800P, across two motherboards with different core logic (KT333/nf2), three different cpus, along with innumerable Windows XP patches along the way, and I see absolutely no reason to use anything other than stock ATi drivers. In fact, if I used anything less than the standard ATi CATs I'd consider I would only be cheating myself.

WildChild

there is a problem with the new drivers have experienced it myself ....3dmark 2001se locks up on me card gets to hot to touch ....this didnt happen with previous versions....

Zardon

interesting comments and certainly worth reading to get different people opinions. To be fair to omega I think over anyone due to his experience since 3DFX days he is more in a position to release modified drivers than any other of the modders on the scene atm, a point strongly supported by ATI as in fact they help him with his sets, which is very cool, showing that even they see merit in his work. Of course he wont be able to have the same kind of support when testing before releasing.

He isnt working within a billion dollar company just a guy who wants to help people with certain key issues on driver releases which by going on feedback generally he is successful. Variety is the spice of life! I also feel that most people who install omegas sets are quite sure of what they are doing with any possible risks that come with installing offically unsupported sets.

GMan[CSR]

Awesome work Cowboy, nice way to put in a lot of time and dedication to change the minds of skeptical ATI 3.8 driver users.

MaTT01

Personally I dont see much difference between the Detonator releases and the Catalyst releases. I could care less about WHQL for reasons I wont get into here. What I want is just to see drivers being turned out _fixing_ problems. If it takes constantly leaking betas as Nvidia does then so be it, that typically means you'll get a fix in your hand fairly quick. I may get flamed but in terms of stability and bugs I fail to see a difference between the constant Cat releases and the Det leaked drivers.

In regards to using stock ATI drivers over Omega. There _is_ a reason(s) a vast number of us use his modded drivers. In fact I cant think of one of his releases which hasn't improved upon any of the stock ATI releases. I mean who do you think had a much better installer, ATI or Omega? So in this case "modded" does mean better for most of us As far as the leaked drivers goes from Dell, it was a miracle for alot of people since that release fixed a lot of game problem, display panel problems, etc for many users. NV users aren't the only ones starving for leaked driver sets I assure you.

BWX

I think Omeaga does a great job, he obviously knows what he's doing- his drivers LOOK great and the performance is smooth. If they have a bug he puts out a patch or new set in about 5 minutes. (Sometimes literally). I hope people don't use this issue /w the 3.8's to trash Omega.. I'm sure he is looking at it closely.

snake

Thx a lot for lighting us in all this mess
A very good job indeed

sonycrx

Thanks for you input.

I decided to do some testing myself. using 3.7 and 3.8 drivers only at stock speeds on my setup.

AMD Athlon XP 2800+ / Thermaltake VOLCANO 11 Xaser
MSI K7N2 Delta-ILSR / 512Mb Ram (2 x 256Mb PC2700)
Powercolor ATI 9800 Pro 128Mb / Zalman Heatpipe VGA Cooler (ZM80C-HP)
Hansol 710D Monitor
1 x 80Gb Maxtor 6Y80PO (ATA133) (8Mb Cache) 1 x 20GB Quantum Fireball AS (ATA100) Drives
Sony CRX300E Combo Drive / Sony DRU500AX DVD Writer
Creative Audigy Sound Card / Panasonic RP-HT379 Headphones
Thermaltake Xaser III V2000A Super Tower Case
Thermaltake Xaser Silent Purepower™ 480W PSU

Since the heat pipe is installed I put my sensor onto the top edge of the front heatsink of the Zalman heatpipe cooler.

my results are as follows using idle (with side casing removed) and Ambient temp 18.5c


I let both idle temps settle after 15mins and looped 3DMark2003 for 15Mins and then I took the load reading.

NOTE the top T1 temp is my Alarm Temp, T2 Bottom temp is Sensor temp on Heatsink.



3.7 Drivers Idle (above)




3.7 Drivers under load (above)




3.8 Drivers Idle (above)




3.8 Drivers under load (above)


As you see:

Official Cat. 3.7’s

Idle 42.6
Full Load 50.1


Official Cat. 3.8's

Idle 41.4
Full Load 48.4

the sensor was not totally on the gpu and only on heatsink but heat dissapation should be exactly the same or similar.

Memory temps were not checked. I did however when upgrading from 3.7 to 3.8 , uninstall 3.7 drivers. the ran NFR utility to make sure all ati files were removed. then rebooted and installed 3.8s then rebooted again. and then tested 3.8s.

I think some people do have serious problems with 3.8s but however in my case it didn't seem as bad as I had thought it might be.

karlotta

3.8 is cooler then 3.7.. both are fine in the temp department.
the dell/asus.. leak has a temp increase.. but not realy all that hot ... realy not that hot!
Omegas and the leaks fix some bugs....
Omegadrive is the best modder and has no control over the heat issue of the leaked set...
SO THERE IS NO PROBLEM.. whooo
Ok maybe in a poorly vented case ran at OC. running the "test" for 3 years.24/7 you would MAYbe hurt your card...lol

WaltC

Heh... I don't mind differences of opinion. It's often what forums are for, I think.

Well, I understand your points--and as long as people understand what these driver sets are--I certainly have no objection to anyone running anything he chooses on his own machines. By the same token, if a customer calls ATi with a problem, the first thing ATi will advise him to do if he is running a set of Omegas--is to remove them and install the ATi CAT release. Right? ATi isn't going to support the Omegas, for all of the obvious reasons. I think it would be a mistake to infer that because ATi tolerates them and doesn't actively oppose them (as was not true of nVidia there for awhile, apparently) that the Omegas have ATi's official blessing and support. I don't think that is the case. Really, how could it be? Things like "softmod" scripts and the like surely aren't supported by ATi. I can't see how they'd even have been unofficially encouraged by ATi--certainly I've never seen Terry say that he thinks "softmod" scripts are "cool"...

I'm not saying modded drivers are automatically "bad"--I'm just saying in no case can it be automatically assumed they are *better* simply because they are modded. This, unfortunately, is what I consider to be an error of judgement that some people make.

I don't say that anyone should fail to appreciate Omega's efforts. I'm just saying that they should appreciate the *problems* with this approach to driver modding--mixing driver components from different versions of released drivers--it must inevitably cause problems somewhere. For instance, suppose that a particular driver file contains the bug fix for a game listed in the Cat's read.me, but the Omega version replaces that file with a file from an earlier driver version (ATi doesn't specifiy where in its drivers each bug is fixed.) The result is that even though the Omega set may be "based on" a newer Catalyst release some game bugs relative to older releases may be carried forward in the current Omegas. Or suppose that there is a problem introduced between the mixing of old and new Cat driver files. In that case you could well wind up with new bugs of a type not found in any official Catalyst release. Again, IMO, just because it is modded doesn't mean it's better. The main thing I found of interest in this thread is that NC found higher temps with the Omegas, but not the standard CATs. Of course, TM's testing with the CATs did not include the Omegas, either.

I've had such a good experience with the stock Cats over the last year that I've had no desire to try anything else. Also, my one experience with Omegas a couple of years ago with a GF3 wasn't anything to write home about. I got faster performance--but only because the Omegas knocked out my trilinear filtering--which was immediately apparent to me. And so I uninstalled the Omegas and reinstalled the nVidia reference drivers and got my trilinear filtering back and never tried another Omega set. That may not be entirely "fair," but that's what happened. It was enough for me to troubleshoot the nVidia reference drivers--I saw no benefit to doing more of it with the Omegas... I'm not knocking the Omegas, but I don't consider they belong on a pedestal, either.

BWX

To my understanding this was because of the drivers they were based on (Dell), not because they were Omega's drivers. I'm sure Omega's based on the 3.7's or any other driver aren't making temps rise significantly.
I heard the Dell drivers do the same thing.

WaltC

Well...I don't know what to say--because I haven't had any of the problems you allude to having with the standard Cats over the past year... So, maybe that's why I haven't felt a need for the Omegas, Dells, etc. I play a lot of 3d games, but I'm also conservative in setting up the rest of my system--I don't use "3d tweakers" simply because in the past I found they most often cause more problems than they solve, as well as being difficult to uninstall with regard to registry changes they make (which are often undocumented); and while I do overclock the cpu FSB, I use a chipset that locks AGP/PCI at 66/33MHz, respectively, even so. I just have practically no problems at all, and that's the way I like to keep things... I also run a fairly loaded system--4 hard drives, two in RAID 0, and two on standard IDE for a total of ~400gigs, a DVD and a CD-R/RW, an Audigy2, two 512mb interleaved DIMMS for system ram, along with a 9800P, etc.

As far as not seeing a "difference" between ATi's official driver releases and the various so-called "leaked" sets that pop up everywhere on the Internet as Detonators--I'm surprised you'd say this considering it isn't even possible to ascertain the source of those "leaked" drivers. nVidia certainly does not claim them, and categorically refuses to support them. I've read many a statement by Burke and Perez which instruct their end users *not* to use them. As so-called "beta" drivers they are absolutely worthless since nVidia does not accept any bug reports made by end users on the "leaked" drivers. You may not see a difference--but I see plenty of difference in the fact that ATi *fully supports* the drivers it officially releases up to and including accepting bug reports from end users on them. Very big difference, it seems to me. As I said, with so-called "leaked" nVidia drivers, which may originate anywhere other than nVidia, it certainly is naive to think that nVidia deliberately "leaks" them all as beta drivers--especially in light of their public comments on them and the fact that they don't accept bug reports concerning them.

To each his own, I guess. I much prefer ATi's driver program because it is one the company admits to and claims...

Zardon

I wouldnt put any driver on a pedestal, everything in its place, maybe a gorgeous woman but driver code There will always be problems with drivers, offical or unoffical otherwise forums like this would quickly go out of business ! We support omegas drivers so they certainly arent "unsupported".

Also I think it goes without saying ATI dont offically give omegas drivers a "blessing" we are getting into a debate not really intended for this thread to be honest and while I find alot of your viewpoints interesting and your opinions important -- your comments are based on omega driver experiences from 2 years ago, surely someone as clearly educated as yourself couldnt base facts on 2 year old driver experiences, if you look back at ATIs offical driver releases 2 years ago I think you might find them a little unsatisfactory, I know I did.

Choices are good !

karlotta

Ditto.