Catalyst 5.7 available

[Killrose]

Start here for 5.7 Cats

Big gains supposedly.

 

[Beiruty]

Linky, review?

 

[nRollo]

Originally posted by: Killrose
Big gains supposedly

What the? If this is on games that have been in existence a while, this could be good or bad.

My reasoning: ATI architecture is old as dirt. They've made some "breakthrough" that raises performance significantly without reducing IQ?

If this is true, great, but what have they been doing the last three years that they didn't have this?

 

[Munky]

Yeah, usually big gains only come from a driver if the previous driver was buggy, or if the architecture is flawed and they include a bunch of replacement shaders and optimizations/cheats. Since neither of these is true for a 3 year old architecture, I'm kinda skeptical about the big gains.

But it wouldnt hurt to try anyway, and even a 1% gain is better than nothing.

 

[Killrose]

Originally posted by: munky
Yeah, usually big gains only come from a driver if the previous driver was buggy, or if the architecture is flawed and they include a bunch of replacement shaders and optimizations/cheats. Since neither of these is true for a 3 year old architecture, I'm kinda skeptical about the big gains.

But it wouldnt hurt to try anyway, and even a 1% gain is better than nothing.

Maybe they are enabling extra pipes that have been hidden away :shocked:

 

[Elcs]

Originally posted by: Rollo

Originally posted by: Killrose
Big gains supposedly

What the? If this is on games that have been in existence a while, this could be good or bad.

My reasoning: ATI architecture is old as dirt. They've made some "breakthrough" that raises performance significantly without reducing IQ?

If this is true, great, but what have they been doing the last three years that they didn't have this?

ATI's last driver set was 'supposed' to give a 0-20% gain on a plethora of games. What did we get? Not even 5% accross the board I dont think.

As Rollo said, if its good... where do improvements come from? They've had quite a while to sort out major bugs and problems. Theres only so much a driver can do for performance until another bottleneck is reached.

 

[Ackmed]

It did give the increase they said it would, under certain circumstances. Which anyone with a brain should have been able to figure out themselves.

 

[JBT]

Release Notes Looks like all the performance gains are for cards with 64-128 MB of memory, NOT 256 MB.

AKA cards that are crap at high res + AA and AF are twice as fast now as they were. Thing is twice as fast as crap is still ussually crap. YMMV of course.

 

[jasonja]

Dang you guys are impossible to please. You slam ATI for getting kicked by nVidia this round and when ATI does release driver improvements you assume something is suspicious becuase the architecture is "old". Contrary to what you believe a driver changes all the time and engineers do come up with new approaches and improvements to existing code all the time.

 

[Unkno]

well a update in the driver is still good if it offers better/new features, fixed bugs, and other stuff...as long as performance does NOT decrease.

 

[Munky]

Originally posted by: JBT
Release Notes Looks like all the performance gains are for cards with 64-128 MB of memory, NOT 256 MB.

AKA cards that are crap at high res + AA and AF are twice as fast now as they were. Thing is twice as fast as crap is still ussually crap. YYMV of course.

The thing is, if their crap is now twice as fast as the competitions's crap, then it makes it look good in benches, regardless of the crappiness. So, if at 16x12 4x8x a 9800p gets 20fps and a 6600gt gets 10fps, then the 6600gt has less bragging rights, even if it's faster at lower resolutions. But in real games, 10 or 20 fps is pretty much unplayable, so it won't make much difference.

 

[Noid]

OK ... since no one has tried these and posted results ...

I will ... gimme about an hour ...

 

[Killrose]

I have an 9500np 128mb/256 bus in my second comp., i'll do some benchies and compare to the 5.6 Omega vers.

Maybe

 

[Noid]

All I can say is wow ... and wtf ...

Just when I thought I was maxed...

My first bench is 3dM05 ...

over 100 point gain ...

from 2948 to 3061 ...

more benches to follow

 

[Noema]

One week ago I'd be downloading this.

But my 9600XT finally fried and my 6800GT is in traveling across the continent, heading to my abode as we speak.

 

[Noid]

odd results in 3dM01

I lost about 800 points from the 4.12's old score...

The 4.12s still rock ...

 

[JBT]

Originally posted by: Noid
odd results in 3dM01

I lost about 800 points from the 4.12's old score...

The 4.12s still rock ...

probably because they are not optimizing for 3dmark01 anymore? As its a pretty bad benchmark for actual GPU performance.

 

[Noid]

Lost around 50 in 3dM03
(from the 4.12's)

3dM01 uses DX8 and is a better CPU with GPU bench.
(not just GPU)

I'm stickn to the 4.12's

 

[Viper96720]

you base your decision on just 3dmark? Where's some actual game bencmarks.

 

[Acanthus]

i love 3dmark 2001! but the car game always crashes after i get by the robot.

 

[Noid]

hmmmm ... I experiencing lockup in 3dM03 now ...

... and just got a BSOD with a 'driver-not-equal' error IE6 when enter this forum ...

 

[Pete]

The gains are due to HyperMemory, no? Isn't that what we learned back when ATI's Catalyst timeline was published by The Inq and Anandtech and everyone else?

 

[nRollo]

Originally posted by: Ackmed
It did give the increase they said it would, under certain circumstances. Which anyone with a brain should have been able to figure out themselves.

Big deal. If they replace some shaders on Doom3 so black looks brown, you want me to cheer when they call that "big increases with new drivers"?

Like I said, on an old as dirt gpu, you don't get "big increases" across the board unless a. they weren't doing their job for the years before b.they're lowering IQ and hoping you don't notice.

I'd be just as skeptical if nVidia released drivers and said "Big fps increases for the Ti4600!" and then someone posted "It only in certain circumstances".
:roll:

 

[Sylvanas]

Why the negativity? ATI release new drivers every month without fail, and they provide various performance enhancements and improvements. When they state they do include performance gains, you all immediately claim that they 'cheated' and assume they are filthy liars. I have not encountered any problems wth the CCC and this driver is fine as most are. People having problems should correctly clean their old drivers before installing, use Drivercleaner availiable at Driverheaven Every new release Rollo is always first to tell you how crap they are and point out anything negative. If ATI have such 'Old GPU tech' why do they more than keep up with Nv and often best them in game benchmarks? Keep an open mind.

 

[Drayvn]

From what ive heard from another forum, they arent that bad, better than the 5.6 cats.