(Ocaholic) Asus HD 7970 Matrix reviewed

[RussianSensation]

That review is awful. FurMark for power consumption, some outdated games, no overclocking, no noise levels. Ok, next.

 

[sze5003]

Yeh that's odd its like they rushed everything. Ha you know what, checked status of my card and its in Ontario Canada. Dunno why newegg sends it there from California first.

 

[RussianSensation]

Special delivery. Maybe you'll get it with a side of maple syrup and Tim Hortons flavoured water (I mean coffee).

 

[Accord99]

Yeh that's odd its like they rushed everything. Ha you know what, checked status of my card and its in Ontario Canada. Dunno why newegg sends it there from California first.

It could be Ontario, California.

 

[sze5003]

Yeh your right its possible it is still in cali. In the meantime I'm looking for a CPU cooler that fits my case finally going to overclock this thing for fun.

 

[Jeff007245]

Yeh that's odd its like they rushed everything. Ha you know what, checked status of my card and its in Ontario Canada. Dunno why newegg sends it there from California first.

Did they get it in stock already?

 

[3DVagabond]

That review is awful. FurMark for power consumption, some outdated games, no overclocking, no noise levels. Ok, next.

It's really strange. Ocaholic were the ones who first showed of the O/C'ing prowess of the 7850 when you gave it some voltage. They were also the ones who showed us the performance improvements with each Cat release. Now they are running old drivers and they've stopped O/C'ing. WTF?

 

[zaydq]

It's really strange. Ocaholic were the ones who first showed of the O/C'ing prowess of the 7850 when you gave it some voltage. They were also the ones who showed us the performance improvements with each Cat release. Now they are running old drivers and they've stopped O/C'ing. WTF?

Just my thought... why are they using 12.4? Its well known 12.7, 12.8, 12.9 have given the 79xx series a pretty nice boost. Also, whats with the lack of temperatures, overclocking and noise levels... a big part of this card is the fan/heatsink Asus stuck on it.

We'll wait for a more sophisticated review...

 

[Gunslinger08]

I'm in the market for a 7970 or 680. I wish ASUS would give concrete pricing and dates, since the DCII cooler is supposed to be one of the bests and I like that this is at 1100 mhz from the factory, since I don't really feel like overclocking myself, regardless of how easy it is now.

 

[The Alias]

I'm in the market for a 7970 or 680. I wish ASUS would give concrete pricing and dates, since the DCII cooler is supposed to be one of the bests and I like that this is at 1100 mhz from the factory, since I don't really feel like overclocking myself, regardless of how easy it is now.

http://www.newegg.com/Product/Produc...82E16814131471
quiet and powerful

 

[-Slacker-]

Catalyst 12.4?....

 

[Christian Ney]

Hey guys,
(I am the reviewer )

So to clear things up:

1. Catalyst 12.4 use.
We did use them to keep results comparable with previous cards we benched with. As they are more than onre or two, spending ~2.5 hours per card to bench them all again with latest drivers is not worth it.
Furthermore it's aobout to show the performance difference between the regular 7970 and the Matrix. Whenever you use the 12.4 or the 12.8, performance difference will be the same (not against nvidia cards though as some performance bugs have been fixed in new drivers). That's one reason we have driver performance tests.

2. No noise level/No Temperature.
Remember we published the pics of the card ? Well ASUS moved the NDA without notify us so they asked to remove our pics.
Now they forgot to notify us that they moved the NDA for reviews again from 27th september to 16th October, as they did a mistake twice they said we can keep the review published but we had to remove some parts (like the noise level, temperature, overclocking and so).

3. Overclocking.
First reason of no overclocking: ASUS asked us to remove that part.
Second reason is that we didn't manage to raise the voltages of the card.
First of all both GPU Tweak It version we had (one provided by ASUS and the one in the bundle didn't work) adjusting voltages wise.

Then this week end we tried during two days to raise the voltages of the card without sucess. And it's not like we are noobs as in our team we have some world top overclockers (Hiwa, Splmann, myself ). We tried, buttons on the cards, asus hotwire, vmods, oc key, gpu tweakit, gpu tweakheaven, afterburner, asus powercontrol, well we did everything, no way...


4. Games outdated.
Do you have any suggestions ?



btw, thanks for the feedbacks guys, really appreciated.

Cheers
Chris

 

[HurleyBird]

1. Catalyst 12.4 use.
We did use them to keep results comparable with previous cards we benched with

Yet the Nvidia cards used 4 different drivers.

As they are more than onre or two, spending ~2.5 hours per card to bench them all again with latest drivers is not worth it.

I disagree. You weren't even testing a lot of cards. Being one driver version behind, two at the most, is understandable. Being 4-5 versions behind is unacceptable.

Furthermore it's aobout to show the performance difference between the regular 7970 and the Matrix.

May as well clock any 7970 to 1100/1650 then. It's not like people don't already know how the 7970 performs at that level.

Whenever you use the 12.4 or the 12.8, performance difference will be the same (not against nvidia cards though as some performance bugs have been fixed in new drivers).

But you included Nvidia cards in your reviews.

That's one reason we have driver performance tests.

Do you expect casual readers to factor that in?

2. No noise level/No Temperature.
Remember we published the pics of the card ? Well ASUS moved the NDA without notify us so they asked to remove our pics.
Now they forgot to notify us that they moved the NDA for reviews again from 27th september to 16th October, as they did a mistake twice they said we can keep the review published but we had to remove some parts (like the noise level, temperature, overclocking and so).

You should have waited and done a proper review.

3. Overclocking.
First reason of no overclocking: ASUS asked us to remove that part.
Second reason is that we didn't manage to raise the voltages of the card nor overclock it.
First of all both GPU Tweak It version we had (one provided by ASUS and the one in the bundle didn't work), both frequency and voltages.
We managed to overclock it via MSI Afterburner Xtreme Edition (frequencies not voltages) but well ASUS doesn't want to show that we are using their competitor to overclock their card.

Then this week end we tried during two days to raise the voltages of the card without sucess. And it's not like we are noobs as in our team we have some world top overclockers (Hiwa, Splmann, myself ). We tried, buttons on the cards, asus hotwire, vmods, oc key, gpu tweakit, gpu tweakheaven, afterburner, asus powercontrol, well we did everything, no way...

You couldn't raise the voltage, most overclocking utilities don't work, you're an overclocking centered website, and you gave it a 5/5 rating?

4. Games outdated.
Do you have any suggestions ?

I didn't really have an issue with game selection, except that DiRT 2 is pretty old. I'd change it out for DiRT 3 or Showdown

btw, thanks for the feedbacks guys, really appreciated.

Apologies for the evisceration, but I hope the feedback helps.

 

[Christian Ney]

Yet the Nvidia cards used 4 different drivers.

Yes because for exemple you can't use the drivers used for the 660 Ti with the 660 as they aren't supported in those drivers. When cards are supported with the previous drivers I use them.
nVidia told me there were no performanvce improvements made between those two drivers in the exemple above, only added 660 support.

I disagree. You weren't even testing a lot of cards.

AMD Radeon HD 7970 CrossFire
ASUS Radeon HD 7970 Matrix Platinum
ASUS Radeon HD 7970 DirectCU II Top
AMD Radeon HD 7970 (XFX Double Dissipation)
AMD Radeon HD 7950 CrossFire
ASUS Radeon HD 7950 DirectCU II Top
AMD Radeon HD 7950
ASUS Radeon HD 7870 DirectCU II Top CrossFire
ASUS Radeon HD 7870 DirectCU II Top
AMD Radeon HD 7870
ASUS Radeon HD 7850 DirectCU II Top
Club3D Radeon HD 7750 Royal Queen

That's 12 cards belonging to the HD 7xxx series, so that would be 40 hours of additional testing.

Being one driver version behind, two at the most, is understandable. Being 4-5 versions behind is unacceptable

In fact we are at two drivers behind, AMD didn't release any 12.5 or 12.7 WHQL. So newer ones are only 12.6 and 12.8

We try to use WHQL version usually, we only use Beta version when no WHQL version are supporting the hardware.

May as well clock any 7970 to 1100/1650 then. It's not like people don't already know how the 7970 performs at that level.

I agree on that one, that's why there will be another article following

But you included Nvidia cards in your reviews.

Yes, you are right

Do you expect casual readers to factor that in?

True point, I should mention it,I did it last time, forgot this time...

You should have waited and done a proper review.

Wrong, ASUS should have done their job right twice.

You couldn't raise the voltage, most overclocking utilities don't work, you're an overclocking centered website, and you gave it a 5/5 rating?

Pretty sure it will turn out we have a defective sample or something like this as I know some of my overclocking friends who managed to raise voltages and overclock the card.

I didn't really have an issue with game selection, except that DiRT 2 is pretty old. I'd change it out for DiRT 3 or Showdown

True, I wanted to do so but I didn't have DIRT3 and Showdown back in the day, will be done for next GPU charts update

Apologies for the evisceration, but I hope the feedback helps.

You are welcome, it's very appreciated, it's the only way for me to improve reviews, if I don't have any feedbacks then I can think it's a prefect review , but I know I have to rework the GPU reviews and a few other ones, I joined ocaholic not long ago and I am currently working on old templates even if I already improved them

 

[HurleyBird]

nVidia told me there were no performanvce improvements made between those two drivers in the exemple above, only added 660 support.

It isn't just those two drivers. You go all the way from 296.10 to 306.02.

AMD Radeon HD 7970 CrossFire
ASUS Radeon HD 7970 Matrix Platinum
ASUS Radeon HD 7970 DirectCU II Top
AMD Radeon HD 7970 (XFX Double Dissipation)
AMD Radeon HD 7950 CrossFire
ASUS Radeon HD 7950 DirectCU II Top
AMD Radeon HD 7950
ASUS Radeon HD 7870 DirectCU II Top CrossFire
ASUS Radeon HD 7870 DirectCU II Top
AMD Radeon HD 7870
ASUS Radeon HD 7850 DirectCU II Top
Club3D Radeon HD 7750 Royal Queen

That's 12 cards belonging to the HD 7xxx series, so that would be 40 hours of additional testing.

Two things:

One: There were a total of ten cards in the review, five of which were Radeons. You might need to retest other cards for different reviews, but you don't need to do them all at once.

Two: Many of those custom cards might not need to be retested, ever. You usually only need the stock cards for comparison.

In fact we are at two drivers behind, AMD didn't release any 12.5 or 12.7 WHQL. So newer ones are only 12.6 and 12.8

While that's true, it's a technicality. Even if AMD had bothered to release WHQL versions of Cat 12.5 and 12.7 the current WHQLs would be just as far ahead.

Wrong, ASUS should have done their job right twice.

Sure Asus screwed up, but that's not what readers are thinking when they read your review. At the very least you could have framed things better, either by explaining to the reader the situation, or just calling the review "Part 1" and withholding grading until part 2 is finished.

Pretty sure it will turn out we have a defective sample or something like this as I know some of my overclocking friends who managed to raise voltages and overclock the card.

A defective sample could mean quality control issues. As a reviewer, you should try to get to the bottom of it and inform your readers as best you can. If I were in your place, I'd wouldn't be so quick to hand out a 5/5 to a company that gave me a defective sample.

True, I wanted to do so but I didn't have DIRT3 and Showdown back in the day, will be done for next GPU charts update

Understandable.

You are welcome, it's very appreciated, it's the only way for me to improve reviews, if I don't have any feedbacks then I can think it's a prefect review , but I know I have to rework the GPU reviews and a few other ones, I joined ocaholic not long ago and I am currently working on old templates even if I already improved them

Well, seems like you have an open mind, and I'm glad I didn't offend you.

 

[Christian Ney]

It isn't just those two drivers. You go all the way from 296.10 to 306.02.

Yes 296.10 for everything before GTX 6xx series as they aren't compatible with any GTX 6xx we had to use 30x.x ones.
301.10 were the first drivers compatible wit the GTX 680 (used for only one card then I used the 301.24).
301.25 were the first ones for the GTX 670. I did also use them for some GTX 680 cards clocked at clocked different than stock (was for a dedicated article). nVidia told me there were no performance improvement made from one to another, just newer cards support added.
But you right I should mention it in the Test Configuration page.
303.53 were the first drivers with 660 Ti support.
306.02 were the first drivers with 660 support.
But yeah I guess if we take a look from 301.24 to 306.06 there maybe are dperformance differences.
As for the HD 7xxx cards from AMD, they were all compatible with the Catalyst 12.4 so that's why we can use always the same driver here and not with nVidia.

Two things:

One: There were a total of ten cards in the review, five of which were Radeons. You might need to retest other cards for different reviews, but you don't need to do them all at once.

Two: Many of those custom cards might not need to be retested, ever. You usually only need the stock cards for comparison.

You have a very good point, I had this idea in mind but finally removed:
First because I am a maniac (or stupid, depend from the point of view) so when I have results I put them in tables
Second, because those days I didn't have time to do so.

But yes it's my fault, you have a very good point here and I will have to change this asap.

While that's true, it's a technicality. Even if AMD had bothered to release WHQL versions of Cat 12.5 and 12.7 the current WHQLs would be just as far ahead.

:whiste:

Sure Asus screwed up, but that's not what readers are thinking when they read your review. At the very least you could have framed things better, either by explaining to the reader the situation, or just calling the review "Part 1" and withholding grading until part 2 is finished.

We were caught off guard, I will check with my colleague to edit the review and add into the introduction and the conclusion as I am out of office for 3 days.
I posted a comment under the review anyway already.

A defective sample could mean quality control issues. As a reviewer, you should try to get to the bottom of it and inform your readers as best you can. If I were in your place, I'd wouldn't be so quick to hand out a 5/5 to a company that gave me a defective sample.

Though this one is very likely a retail version of the card as we got it very early some stuff like this happens. May be true, may be wrong for the retail version but as we can't state on this we didn't mention it before having more informations on the matter. ASUS just sent a new BIOS to us and other review websites that should fix the issue. Hopefully the BIOS is updated on retail product, I will have to ask.

But generally you right, it would have been more intelligent to remove the review and wait before having the final version of it. ASUS screwedup, we were caught off guard and we had to ask very fast and didn't do it the best way.

I will work on it as soon as I am back from the event I am attending to provide readers a better review.
I hope I can do the drivers stuff as well and also replace DIRT 2 with Showdown now that I bought it

Well, seems like you have an open mind, and I'm glad I didn't offend you.

You are welcome

 

[3DVagabond]

@Chris,
Thanks for showing your concern. I've really enjoyed your reviews, and yes, you get very good O/C's on your test samples. It's apparent you know what you are doing.

I understand the difficulty with retesting multiple cards every time a new driver comes out. Even if you only do WHQL releases. I still think it would be a good idea to try and include latest drivers for the review card, somehow. Especially on a new(ish) arch like GCN where improvements are often appreciable. I don't have to tell you how much drivers have improved for GCN. You've done all of the performance reviews, yourself.

For games;

I would recommend maybe picking up Showdown instead of Dirt 2. It's being included with most new AMD cards. So it's a game the purchasers would be interested in. You just have to be careful not to skew overall results too much towards AMD though. Possibly add BL2. It's bundled with some nVidia cards and should help balance out things a bit.

Tessmark? Does AMD even optimize their drivers for that bench? It doesn't seem so. Heaven should cover tessellation fine and is well optimized for both brands.

CoD MW3. Is it really needed to test games that are over 100fps? It seems like performance is all over the place anyway. 660 beats 670, 660ti beats 680? Doesn't seem to give any relevant performance figures with results like that.

Starcraft 2 is another one that gives flaky results. It doesn't really come out in any logical order. Maybe because it's so CPU dependent?

Skyrim is a game worth keeping, but I don't think you are getting accurate figures with Cat 12.4's. IIRC it was a later driver that fixed that game for GCN.

Get Alan Wake or Sniper Elite V2 pus Max Payne in there = newer games. Metro 2033, because it's so demanding. You need a game that even the best card gets beaten and bloodied trying to run.

Expand your horizons;

Maybe Luxmark, or some other OpenCL bench? At least for the AMD reviews. Since it's such a strong point for GCN it seems like it's not really giving an accurate performance picture for the architecture by not including some numbers for that.

Likewise, for nVidia reviews show benches with PhysX. I understand that you can't do comparable benches with AMD cards using PhysX, but if people are supposed to buy nVidia cards because of PhysX, don't you think they should see how the cards perform when using it?

Maybe PS6, if both companies drivers are ready to support it. We get quite a few queries about card recommendations for Photoshop here. I believe that Gimp (open source) also supports GPU acceleration and is free if you don't have Photoshop in your programs suite.

I know that's a lot of changes and changing them all would be very difficult and a lot of work. They are just some recommendations and they would have to be play tested to make sure that what you change won't unfairly effect the overall balance and give you results that don't accurately reflect the overall performance between the cards. There is no way I'm expecting you to do everything (or anything) that I'm recommending. Maybe just see the overall effect I'm going for with these recommendations and change things your own ways to update and expand the scope of your reviews.

Thanks for asking. :thumbsup:

 

[Jaydip]

Good ideas 3d

 

[Christian Ney]

Yes wouldn't be a bad idea to drop CoD and StarCraft (already thought about it).
Tessmark has been optimized by AMD but they did it wrong, I contacted them about it because I saw the issue but they never contacted me back on this one. Will drop it to.

Wanted to add BL2 now that I bought it and Showdown was already planned to replace DIRT2.

As for Skyrim that's true, Catalyst 12.4 aren't performing well that's why the values are in red, I show this in our driver performance tests.

I don't have alain wake nor sniper elite v2 nor max payne, are they really demanding game ? Sorry I didn't follow them

As for Metro 2033, It's demanding but was very bugged at first, lots of reviewers complaned about it, so I removed from my mind. Maybe the issues have been fixed now...

As for OpenCL, do you think a lot of people will be interested to see results ?

PhysX, yeah on a separated article, but a lot of websites do it too, maybe if something interesting come out that no one did then I will give it a try.

What is GCN btw ?

 

[HurleyBird]

Graphics Core Next, the architecture underlying the Radeon HD 7000 series.

 

[3DVagabond]

Yes wouldn't be a bad idea to drop CoD and StarCraft (already thought about it).
Tessmark has been optimized by AMD but they did it wrong, I contacted them about it because I saw the issue but they never contacted me back on this one. Will drop it to.

Wanted to add BL2 now that I bought it and Showdown was already planned to replace DIRT2.

As for Skyrim that's true, Catalyst 12.4 aren't performing well that's why the values are in red, I show this in our driver performance tests.

Thanks for considering everything. I knew there were performance issues early on w/Skyrim, but the game is performing better now. The problem with that is a lot of people don't follow video cards. They merely read reviews when they are actually shopping for a card. A lot of times it could be the first time they've ever even looked at a tech site. The info you guys give them is what they are basing there buying decision on today. They need to know current performance levels to make this decision. Having old or broken drivers or games that skew the results can mislead them into believing one card is better than another when it's not the case.

I don't have alain wake nor sniper elite v2 nor max payne, are they really demanding game ? Sorry I didn't follow them

Sniper Elite V2, 2012 release, Dx11 and can be very demanding depending on in game settings used. Uses Super Sampling, D3D accelerated AA and DoF, Shader Model 5 contact hardening shadows.

Max Payne 3, 2012 release, Dx11 with lots of modern rendering features, generally runs "where you would expect it to" with both brands.

Alan Wake, Probably not as important, feature wise, but is a popular 2012 release.

While I think it's important to maintain a base with some older games for reference purposes, overall the benchmark suite should be fresh and up to date. For one thing you'll be testing games that people are likely going to be buying new cards to play. Plus you'll be driving advancement by showcasing new features as well as performance.

As for Metro 2033, It's demanding but was very bugged at first, lots of reviewers complaned about it, so I removed from my mind. Maybe the issues have been fixed now...

Not sure about bugginess (sic). just mentioned it because it can be very demanding and is a game that is widely used as a benchmark. Can be used to compare relative to other reviews because of that.

As for OpenCL, do you think a lot of people will be interested to see results ?

PhysX, yeah on a separated article, but a lot of websites do it too, maybe if something interesting come out that no one did then I will give it a try.

What is GCN btw ?

OpenCL is an emerging feature. It's not immensely popular at the moment. It's another way to implement GPU acceleration to dramatically increase performance in tasks that have generally been done using the CPU. I suppose to do it right someone would need to do a performance article featuring OpenCL. See what useful applications run it and show people the benefits. Then some benchmarks could be included in reviews with a reference back to the original article for a better explanation of it's uses.

As it is now sites just tend to throw a Luxmark bench in their reviews with no real explanation. Not only don't most people use Luxrender, but most people don't even know what it is. Adobe leveraging OpenCL though is going to make it an important feature going forward. That's a lot of corporate weight behind something. I guess it just depends on how proactive you want to be and the resources you have to throw into it.

I think PhysX should be included in every nVidia review, period. If someone is going to buy an nVidia card because it has PhysX (and we are constantly told that it's one of the main advantages that nVidia has) then they need to know how it will run, at the very least.

If someone is going to go out and buy an nVidia card to play BL2 or Batman and use PhysX. What card are they going to need to use PhysX and have acceptable game play? GTX-650? 660? 680? Are people better off with one powerful card? Is it better to offload the PhysX to a dedicated PPU? People need to know these things. As it is now, lots of people just buy nVidia and think they're golden.

We need you guys, who test and sample all of these products, to tell us what we need (not necessarily what to buy). That indirectly leads to sales of newer products and gives these companies a reason to supply your industry.

 

[RussianSensation]

3DVagabond gave some great suggestions. The more demanding games this year ironically all favour AMD 7970 series: Sleeping Dogs, Dirt Showdown, Sniper Elite V2, Alan Wake & its expansion: American Nightmare.

I'd consider dropping SC2, STALKER:COP and Aliens vs. Predator as well. Christian, I think you tested STALKER with DX10 instead of DX11? Did you have god rays on? The game runs 4 separate scenes, you should take the most demanding scene with DX11.

There are going to be a ton of new games coming out shortly so you don't necessarily have to drop these at once. Max Payne 3, Alan Wake, Sniper Elite or Sleeping Dogs are good candidates.

Dirt 2 --> Dirt Showdown or F1 2012 (Dirt Showdown has a global illumination model and contact hardening shadows which run poorly on NV cards - the same reason AMD will have a performance edge in Sniper Elite V2). You could test F1 2012 instead but it may not be demanding enough at all.

Starcraft 2 --> If you want to test a more demanding widely popular strategy game, consider Shogun 2, or maybe go for an MMO like Guild Wars 2

Aliens vs. Predator --> Outdated game by now and could be replaced with Borderlands 2, Dishonored, Medal of Honor: Warfighter, Far Cry 3. Dishonored and Borderlands 2 may not be demanding at 1080P (without PhysX on High for BL2). Maybe consider testing such games with Super-Sampling enabled You could also take a less GPU demanding game and see how modern GPUs handle SSAA --> Portal 2, HL2 Black Mesa mod (this should be free if you own any source engine game on Steam), Skyrim or the beautiful Trine 2.

Metro 2033 is demanding but you could just revisit that series next year with Metro Last Light if you don't feel like testing it now. Although Metro 2033 is pretty cheap to buy now and the built-in benchmark in Steam folder is very easy to use.

Crank MSAA in Batman AC to 8xAA since without it the game is not very demanding on GPUs.

Dragon Age 2 ---> SKYRIM + ENB mods because DA2 was not a very liked game and Skyrim + ENB mods and 8xMSAA stresses the memory sub-system of a GPU, the high resolution textures stress the VRAM and it's a far better and more popular title.

Battlefield 3 --> See if you can instead test the multi-player component in Armored Kill expansion. This game is pretty difficult to test consistently overall.

Some other game idea could be Project Cars with 4xSSAA, Arma II Reinforcements / Day-Z mod (or wait for Arma III to release).

Next year should be a lot more games that you could use to update your suite, besides Metro Last Light, Crysis 3, Tomb Raider, Bioshock Infinite.

 

[Christian Ney]

@Russian,

Thank you very much for the feedback, I will consider that for next GPU charts update (planned for next GPU gen).
And STALKER was run in DX11.

http://www.ocaholic.ch/xoops/html/modules/smartsection/item.php?itemid=901&sel_lang=english

Review back online:
Changes:
Added: Temperatures, noise levels, RPMs, VRM temps.
Rebenched all HD 7xx0 series using Catalyst 12.9Beta.
No overclocking: the MCU of our card died (was already not working in the first place).

 

[wand3r3r]

@Russian,

Thank you very much for the feedback, I will consider that.
And STALKER was run in DX11.

http://www.ocaholic.ch/xoops/html/modules/smartsection/item.php?itemid=901&sel_lang=english

Review back online:
Changes: Temperatures, noise levels, RPMs, VRM temps.
Rebenched all cards using Catalyst 12.9Beta.
No overclocking: the MCU of our card died (was already not working in the first place).

Thanks for updating the AMD driver! That alone perked my curiosity as I know they've increased performance lately. It's also great you're taking suggestions to heart and providing good rational answers whether or not you will do them.
Now to go read.

 

[Christian Ney]

Yes, Performance from a Catalyst driver to another can change quite a lot:

AMD Catalyst 12.9 Beta - HD 7970 CrossFire Gaming Performance

AMD Radeon HD 7900/7800 Series - Catalyst 12.9 Beta vs Catalyst 12.4 WHQL Gaming Performance

AMD Catalyst 12.9 Beta - HD 7970 Gaming Performance