• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD ATI Catalyst Driver Performance Comparison @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
2FA
Joined
May 18, 1997
Messages
58,433
AMD ATI Catalyst Driver Performance Comparison - Ever wonder just how much performance has improved with drivers since launch date with AMD's ATI Radeon HD 5800 series? We did after NVIDIA's recent driver launch, so we aimed to find out just how much AMD's drivers have improved. We compare the launch driver, the first WHQL driver, and the current WHQL driver.
 
Last edited:
They may have just tapped all the performance they can out of this revision of the 5000 series. If they want to surpass Fermi they will need to step it up on the next revision and make crossfire performance scale just as well as it seems to on Fermi cards.
 
Interesting. I would have expected more of an improvement, but what I really want to know is how they stack up against NVidia's driver releases - do they have the same, more, or less improvement?
 
Last edited by a moderator:
i think that would be too hard to tell right now due to the 6 month delay from nvidia
 
Like the article. Quick. To the point. Cut and dry. That's the way to do it!

And yep. I'm ready for next gen (taps wrist watch).
 
I liked it, it was pretty much what I expected but it proves that you aren't going to get a giant performance gain later just because of drivers, you'll get some more but nothing drastic.
 
I am a bit surprised by the results but I can’t say I am disappointed or anything.

The results seem to indicate the release drivers were very good quality and I hope they can do the same with the next generation of cards.

The other thing it seems to indicate is that the old days of degrading IQ further and further with every new driver as a way to increase performance are really gone. Or at least I can hope they are.
 
Check out 10.6 with crossfire though, they're awfull!! From BSODs, to Stuttering, poor H.264 performance, to somehow locking i7 multipliers to the 12-14 range. Absolutely attrocious!! Every forum I look on has dozens of posts complaining about them. They lasted 10 minutes on my system before I had to uninstall them.
 
The article was very good. I am sure it was alot of work. Thanks.

But.......while ATI has done a ton of work on the Wow factor implementation.....EyeFinity....it's pretty evident from the article that the performance aspects have been pretty stagnant.

CrossFire still remains behind the nvidia SLI solutions by a good margin, as noted by the latest nvidia beta release in Surround setups.

I keep coming back to ATI like an abused child, but after seeing these results, I think it's time to move back to nvidia.:eek:
 
These new comparisons are awesome!

Thank you SO much for taking the time to do these. I think they are very valuable. There are very rarely firm answers on whether to upgrade to the latest drivers or not and the community is always split.

Now if we could only get these every time a new release comes out... :D
 
I'm impressed that ATI has kept to thier promise to release new drivers once a month. I generally prefer nvidia, since I've had some horrible experiances with ATI cards and drivers.

However, nowadays it seems like its swinging the other way. Nvidia releasing card frying drivers with fan screwed up, and ATI consistantly releasing new and improved drivers.
 
Devistater said:
Nvidia releasing card frying drivers with fan screwed up, and ATI consistantly releasing new and improved drivers.
Click to expand...


That happened with 1 driver (IIRC), and not everybody experienced it.

You are basing your opinion regarding Nvidia on that? Wow.
 
Great article. Running duel 5870's myself. I know this testing takes a bit of time, but it would have been great to see a comparison between Crossfire setups with the different drivers. See if there's been any improvements in that department.
 
Ruachta said:
Great article. Running duel 5870's myself. I know this testing takes a bit of time, but it would have been great to see a comparison between Crossfire setups with the different drivers. See if there's been any improvements in that department.
Click to expand...

Reading around its pretty much accepted that ATI has abandoned trying to get crossfire to work and scale as well as SLI has.
 
Splinter Cell Conviction

Here we really aren’t seeing any improvements at all through the drivers. There isn’t anything else to say here, it seems to perform the same now as it would have nine months ago.
Click to expand...
I'd call a >30% boost in minimum framerate a significant improvement!
 
I wish you guys would throw one or two 4000 series cards comparisons. I know that by today's standards they're dinosaurs but it could interesting, maybe..
 
It makes sense that the 8.66 RC6 and 9.10 drivers performed about the same - the D3D driver is identical in version (8.14.10.0700). The only version change was with the OpenGL driver (6.14.10.9016 vs 6.14.10.9026).
 
I forgot to mention in the article, the Bad Company 2 load times were very noticeable, the older drivers were very slow, but Cat 10.6 was fast to load.
 
Pretty good article and I'm glad we finally have some kind of published article shedding some positive light on ATI's drivers. Everyone still seems to think that they're horrible and everyone's so quick to dismiss them because of it.

I've been running a 5870 since launch and the driver experience has been nothing short of stellar. I've been installing drivers over the old one's and upgrading with ease. I'm glad to see a trend of minimum frame rate improvements over the driver releases because it really helps keep things smooth. I plan on adding a second card unless some details on the "southern-islands" or a refresh come up. That might change my experience but I somehow doubt that. I've run 4850's in crossfire before and even back then the experience was really good.

ATi's been doing a stellar job of releasing drivers and hotfixes. There has been a hotfix for every major bug these past few months and I don't really mind waiting a few extra days for it but I guess some really do. The introduction of eyefinity features and application profile's is also a nice thing and people still bitch about it, but at least we have options now.

I would like to see some crossfire testing done @ [H] with the latest app profile's and driver set. The PC Perspective review kind of has my mind boggled. A majority of the big performance improvements in catalyst lately have been crossfire specific, that bundled with the app profile's should be making for a pleasant gaming experience. I find it odd that the crossfire setups were only a small fraction faster than the single card setups and sometime's even slower.

I don't doubt that Nvidia has improved their scaling with their latest drivers but I find it hard to believe that crossfire is now all of a sudden greatly underperforming when all of these people and reviews have shown extremely positive results since the 5800's launch.
 
Its nice to see that drivers for a video card are being decently written. So much programing seems so sloppy now a days.
 
i wonder how much performance is affected when the driver team takes their attention away from an older card. say how does an x1950 perform from drive to drive from release to end of driver support.
 
fausto412 said:
i wonder how much performance is affected when the driver team takes their attention away from an older card. say how does an x1950 perform from drive to drive from release to end of driver support.
Click to expand...

They aren't supported in the new drivers. It's considered a legacy card and they don't recieve updates regularly and if they do it's only if something is severely broken. I think they said they'd release driver updates for legacy cards every 3 months or something but don't quote me on that.
 
ATi makes great cards, but they burned me too many times with my old 4850CF setup. Once I went through 4 bad releases and a fifth that SOMEHOW wiped my MBR, they lost me as a customer. I'll sooner get one of those SiS Chrome cards (or whatever) before I go ATi again. It's just not worth the hassle.
 
Modred189 said:
ATi makes great cards, but they burned me too many times with my old 4850CF setup. Once I went through 4 bad releases and a fifth that SOMEHOW wiped my MBR, they lost me as a customer. I'll sooner get one of those SiS Chrome cards (or whatever) before I go ATi again. It's just not worth the hassle.
Click to expand...

Sounds like user error to me? My 4850's ran beautifully, and I believe that was right around the cat 9.4-9.10 era. I understand a bonked driver release, sometime's it happens to both amd/nvidia but 4 bad installs and claiming it fucked up your mbr? That sounds like other issues and/or user error. You must of fucked something in your registry up when trying to get rid of old driver installs.
 
The only weakness in this study is clearly stated at several points in the article:
Keep in mind that several game patches have also been released since this game has been released which have also improved the games [sic] performance since the game launch.
Click to expand...
Unless Kyle and company go through the trouble of testing every iteration of game patch and driver version, improvements will appear flatter than they probably were. That improvements exist at all should be our take home here.

Something else that barely showed up in the testing is very boolean: Does the game (or certain features) run at all? FPS means very little (and may be entirely unmeasurable) if your game keeps crashing. It's a testament to the quality of both AMD's and nVidia's recent releases that this kind of test can be done at all.

bazylik said:
I wish you guys would throw one or two 4000 series cards comparisons. I know that by today's standards they're dinosaurs but it could interesting, maybe..
Click to expand...
fausto412 said:
i wonder how much performance is affected when the driver team takes their attention away from an older card. say how does an x1950 perform from drive to drive from release to end of driver support.
Click to expand...
It would be useful to run this test on older cards that use the same driver package, and to continue testing driver updates on the 5870 well after AMD's next card refresh. The specific cards are not important. If performance consistently plateaus (or degrades) for older cards on newer drivers, that trend could tell when, in general, we should stop upgrading drivers or consider stepping up to the next GPU version.

niconx said:
I would like to see 5770 Crossfire added to these results.
Click to expand...
I don't see much value in tackling multiple GPU flavours in the same generation, though, particularly if they are released after some of the driver versions being tested. (I'm not sure where the 5770 fits in the timeline here.) We all have our favourite cards, but it's the performance trend that's most interesting here, not where our card sits above or below the trend line.
 
Battlefield: Bad Company 2 - Adaptive Super Sampling AA
Click to expand...

What? In CCC you get options for Adaptive Multi Sampling AA or Super Sampling AA, but no combination of the both. Perhaps you could expand on what settings you used?
 
srangara said:
I don't see much value in tackling multiple GPU flavours in the same generation, though, particularly if they are released after some of the driver versions being tested. (I'm not sure where the 5770 fits in the timeline here.) We all have our favourite cards, but it's the performance trend that's most interesting here, not where our card sits above or below the trend line.
Click to expand...

From what I have seen crossfire 5770 cards can often match and even best a single 5870, not to mention that two 5770 cost significantly less than a single 5870. So yea I would like to see how 5770 crossfire matches up here.
 
niconx said:
From what I have seen crossfire 5770 cards can often match and even best a single 5870, not to mention that two 5770 cost significantly less than a single 5870. So yea I would like to see how 5770 crossfire matches up here.
Click to expand...
I see that as the subject of a review for comparing cards, not a review of driver performance from one revision to the next. Plenty of sites already have compared how 5770 in crossfire performs next to a 5870; I'm pretty sure this site has as well.

If ]H[ardOCP were going to do an additional comparison, it would make more sense to use a single 5770 to say "We compared how Cypress performs across driver revisions, now here's a comparison of Juniper". It would only make sense to do 5770 in crossfire if they also did a 5870 in crossfire to say "Here's how Crossfire Cypress/Juniper setups perform across driver revisions".
 
LoneWolf said:
I see that as the subject of a review for comparing cards, not a review of driver performance from one revision to the next. Plenty of sites already have compared how 5770 in crossfire performs next to a 5870; I'm pretty sure this site has as well.

If ]H[ardOCP were going to do an additional comparison, it would make more sense to use a single 5770 to say "We compared how Cypress performs across driver revisions, now here's a comparison of Juniper". It would only make sense to do 5770 in crossfire if they also did a 5870 in crossfire to say "Here's how Crossfire Cypress/Juniper setups perform across driver revisions".
Click to expand...

My point was I would like to see how the new drivers affect 5770 crossfire. Maybe add in 5870 crossfire to be thorough?
 
cesium666 said:
What? In CCC you get options for Adaptive Multi Sampling AA or Super Sampling AA, but no combination of the both. Perhaps you could expand on what settings you used?
Click to expand...

I used exactly what is stated, Adptive Super Sampling, not Adaptive Muli Sampling, there is a visual difference between the two modes with Adaptive Super Sampling looking much better, and harder on the hardware
 
magoo said:
But.......while ATI has done a ton of work on the Wow factor implementation.....EyeFinity....it's pretty evident from the article that the performance aspects have been pretty stagnant.

CrossFire still remains behind the nvidia SLI solutions by a good margin, as noted by the latest nvidia beta release in Surround setups.

I keep coming back to ATI like an abused child, but after seeing these results, I think it's time to move back to nvidia.:eek:
Click to expand...

you're getting 4 extra FPS in BFBC2 and you're complaining? This kind've driver release is right on target with what both green and red have been doing with there cards for generations: weeks after launch, push a driver that gives you 5-10% extra performance, and those are the exact figures were seeing here.

Devistater said:
I'm impressed that ATI has kept to thier promise to release new drivers once a month. I generally prefer nvidia, since I've had some horrible experiances with ATI cards and drivers.

However, nowadays it seems like its swinging the other way. Nvidia releasing card frying drivers with fan screwed up, and ATI consistently releasing new and improved drivers.
Click to expand...

you generally prefer Nvidia. Where is all this goddamn brand loyalty coming from!?

Guys this market functions on a product-by-product basis. What one company does one release has very very minimal impact on what they do the next release. eg R600, flop, R700, very successful. X800 series, driver problems, X1800 series, no driver problems, HD2900 series, driver problems... Nvidia's G71 was a lean-mean frame pumping machine. Its successor, the G80, was the biggest GPU ever made (at the time, since surpassed by GT200 and GF100 :\)

To generally prefer one brand is just going to cost you money.

BoogerBomb said:
Reading around its pretty much accepted that ATI has abandoned trying to get crossfire to work and scale as well as SLI has.
Click to expand...

Where does that come from?

In almost every instance where the single card ATI solution drops below 40FPS minimum, the crossfire config scales 70%+, more often hitting the mid 80's. Nvidia's SLI'd 480's were keeping pace for the most part, but Its best to reserve judgement since that was using a beta 257 release --meaning SLI 480 performance has probably gone up since that review.

Modred189 said:
ATi makes great cards, but they burned me too many times with my old 4850CF setup. Once I went through 4 bad releases and a fifth that SOMEHOW wiped my MBR, they lost me as a customer. I'll sooner get one of those SiS Chrome cards (or whatever) before I go ATi again. It's just not worth the hassle.
Click to expand...

MBR as in master boot record?

What?! Your machine kills 4 perfectly functioning cards and you ditch an entire brand because you figure its the cards fault? ATI's voltage regulation typically gives cleaner power to the GPU than Nvidia's more traditional analogue solution, but it is much more sensitive to voltage ripple, meaning when a PSU goes out of spec its more likely to kill the ATI card than the Nvidia card, but again, that's when the psu goes out of spec. I had a guy, very similar situation to the one you were in, come to me having killed two X800's and an X1800XL. I measured the 12V rail at 10.4V, I can only imagine what it would've looked like under a scope.

Did you bother to bust out a volt meter or even swap the PSUs while you went on your card killing spree?
 
Before I start:
(9 out of 10 picks ati around here) this is due to nv prices in smaller countries, something odd with it, ati's prices is very very low compared to nv around where i live, so only natural for me to lean towards ATI because they've given me more for the money for years now, alltho I've used, using nvidia cards everyday. and so with ati)

Psu's kill HD57xx and up.
4850 and up.

If they're bad, a antec earthwatts(low end) comes with a cheap case( it is the same price as antec 900) sitting on NMT 3G with red signal(it only exist in nordic countries if guys wonder what NMT is) we got LTE for phones and 3G yes.
So cant be bothered to surf the web for prices, but its damn cheap, and it runs 4870 Crossfire with a quadcore Phenom.
While a psu from Qtec at 500 Watts will actually Kill Either MOBO, or videocards.
Older mobo's and cheap models survive anything, You can pair a 785G mobo from asus with any psu and it'll run solidly, even tho 12V is at 13.44 V!!!!
Try that on a highend mobo and it'll break whitin a month, and the same pwm design(parts) are used on higher end ati cards.

My mates have no problems with ati, nor nvidia Hardware wise.

People should learn:
There is virtually no diffrence in RMA rates between ATI and Nvidia.
Sapphire have the lowest RMA rate.
Dualgpu cards have the highest RMA rate, 4870X2 ->295->9800GX2 so ati is worst there.
5970 have a very low rma rate if anyone wonders.

There is virtually no diffrence in drivers and issues:
Ati is worst with 5xxx with mouse issue and BC2, no more noticable over Nvidia otherwise.
If your not struck by those two issues, the grass isnt very much greener on the green camp, even tho its easy to think that way.

Before you purchase, Features->Powerconsumtion->noise->price/performance->perf/watt-> will drivers bother you.

It's not in order of importance, because thats defined from user to user ofc.

and relax, it's basicly two brands, and the truth is that ATI really have done a heck of a job with the hardware!
 
BoogerBomb said:
They may have just tapped all the performance they can out of this revision of the 5000 series. If they want to surpass Fermi they will need to step it up on the next revision and make crossfire performance scale just as well as it seems to on Fermi cards.
Click to expand...

Well, you have to remember they are probably well on their way to having a next gen of cards. It will be a refinement on the 5800 series, of course, but still. The 5800s came out a good while ago, they've had time to work on the next gen. In fact, I suspect they probably would have already launched had ATi chosen to use Global Foundries 32nm process. Instead they seem to be doggedly sticking with TSMC, despite how bad TSMC botched the 40nm stuff.

Also you have to remember they don't have to "best" Fermi by being faster, they just have to have a card that is a better value. At the high end, cards are really faster than they need to be these days for almost everything. I've got a 5870 and it just chews through any game thrown at it. It is more than enough power (which is how I like it). So ATi doesn't have to release something that is faster than what nVidia has, all they have to do is keep their cards fast enough, while being a good value.
 
niconx said:
My point was I would like to see how the new drivers affect 5770 crossfire. Maybe add in 5870 crossfire to be thorough?
Click to expand...
The point of the article was to show the progression (if any) of performance since the first drivers were released. Unless the trend might be different between cards in the same family, there's no point in testing anything other than the best reference example at introduction.

You should be able to find any number of reviews comparing cards as a driver gets released. However, you could not use the 5770 in this particular study as it did not exist when the first driver (at least) in the test was introduced.

Furthermore, to add in specific features (Crossfire, Eyefinity, AA, etc.) would:
  1. Geometrically increase the number of tests
  2. Disallow any performance improvement comparisons with competitors that don't have a certain feature
  3. Not be meaningful on those driver versions that didn't yet support a feature
 
AtkinS said:
People should learn:
There is virtually no diffrence in RMA rates between ATI and Nvidia.
Sapphire have the lowest RMA rate.
Dualgpu cards have the highest RMA rate, 4870X2 ->295->9800GX2 so ati is worst there.
5970 have a very low rma rate if anyone wonders.
Click to expand...

Where are you getting your data from?
AFAIK, manufacturers will not disclose their RMA rates willingly.
 
You must log in or register to reply here.
Back
Top