OCCT 3.1.0 shows HD4870/4890 design flaw - they can't handle the new GPU test !

[Tetedeiench]

since i'm one of the ones who is experiencing 0 isues i figured i'd post this for comparison. testing length for this was 2 min i can go longer if anyone is interested. my earlier test is here.

(please note i have the max displayed for vddc temps and amps)

Did you run the test in FullScreen mode ?

Your VDDC max stayed below 82A, that's why you're not experiencing the issue.

The 82A limit seems to hold true so far...

PLEASE NOTE THAT IN THE STABLE VERSION, SHADER COMPLEXITY 0 IS SELECTED AS DEFAULT IN OCCT. YOU HAVE TO SELECT COMPLEXITY 3 MANUALLY, AS OPPOSED TO THE RC1, WHICH AUTO-SELECTED "OPTIMAL"

[ryan92084]

DAMN thats hot, I think you just got lucky, those vrm are cooking and I think you just got cherry picked vrm that can handle the heat.

they've been quite a bit hotter for longer : P

Did you run the test in FullScreen mode ?

Your VDDC max stayed below 82A, that's why you're not experiencing the issue.

The 82A limit seems to hold true so far...

PLEASE NOTE THAT IN THE STABLE VERSION, SHADER COMPLEXITY 0 IS SELECTED AS DEFAULT IN OCCT. YOU HAVE TO SELECT COMPLEXITY 3 MANUALLY, AS OPPOSED TO THE RC1, WHICH AUTO-SELECTED "OPTIMAL"

yes it is full screen and yes 3 is selected.
after playing around i am able to recreate the black screen issue. by going into the driver settings and taking the anistropic filtering off of forced 16x and on to "use application setting" it will produce a black screen/freeze everytime for me. changing any other setting to application setting causes no issues.

i'm certainly no expert but this would seem to imply that it is an application issue would it not?

[Tetedeiench]

they've been quite a bit hotter for longer : P



yes it is full screen and yes 3 is selected.
after playing around i am able to recreate the black screen issue. by going into the driver settings and taking the anistropic filtering off of forced 16x and on to "use application setting" it will produce a black screen/freeze everytime for me. changing any other setting to application setting causes no issues.

i'm certainly no expert but this would seem to imply that it is an application issue would it not?

No, it's just because forcing anisotropic filtering lower the GPU load When you unforce it, the load goes up the roof, and there you have the bug. That's why, before, the test ran fine : you forced the load to stay under 82A with this setting.

Great finding, that's something new i haven't think of !

[ryan92084]

No, it's just because forcing anisotropic filtering lower the GPU load When you unforce it, the load goes up the roof, and there you have the bug. That's why, before, the test ran fine : you forced the load to stay under 82A with this setting.

Great finding, that's something new i haven't think of !

glad to be of assistance
just for clarification (and forgive me for being slow) are you saying that the act of forcing 16x reduces load when compared to application 16x? Or, are you saying the application is running the filtering at a higher rate than is available for forcing?

[Tetedeiench]

glad to be of assistance
just for clarification (and forgive me for being slow) are you saying that the act of forcing 16x reduces load when compared to application 16x? Or, are you saying the application is running the filtering at a higher rate than is available for forcing?

My app is running at Anisotropic 0. Anistropic filtering is a process that reduces the load on the GPU during the test. I disable it.

Your setting forced it during the test, and as it is set in the drivers, i cannot have any control on it, unfortunatly. That's why the GPU load went down

[ryan92084]

My app is running at Anisotropic 0. Anistropic filtering is a process that reduces the load on the GPU during the test. I disable it.

Your setting forced it during the test, and as it is set in the drivers, i cannot have any control on it, unfortunatly. That's why the GPU load went down

i see, i had always assumed the filtering added load and not the other way around ( see i said i was slow).

by forcing the anisotropic filtering down to 2x the program will still run. FPS increases by roughly 10 and amps go up to exactly 82 so this is inline with the thinking that beyond 82 amps is the killer

[Tetedeiench]

i see, i had always assumed the filtering added load and not the other way around ( see i said i was slow).

by forcing the anisotropic filtering down to 2x the program will still run. FPS increases by roughly 10 and amps go up to exactly 82 so this is inline with the thinking that beyond 82 amps is the killer

Hey, no problem, ,naturally we think it increases the load

Thanks for your tests, and thanks alot for your discovery

[SNiiPE_DoGG]

My app is running at Anisotropic 0. Anistropic filtering is a process that reduces the load on the GPU during the test. I disable it.

Your setting forced it during the test, and as it is set in the drivers, i cannot have any control on it, unfortunatly. That's why the GPU load went down

I still dont get why this program crashing cards is a problem, so if applying AF stops the card from crashing then your set - just enable AF in your games and all woes are gone.

I'm not calling this thing a power virus or anything like that, but it is clear to me that this program is loading the card in certain way that no game would ever do - even if it had the best graphics in the world - because games aren't rendering a mostly static image with little geometry and no AF setting

[Tetedeiench]

I still dont get why this program crashing cards is a problem, so if applying AF stops the card from crashing then your set - just enable AF in your games and all woes are gone.

I'm not calling this thing a power virus or anything like that, but it is clear to me that this program is loading the card in certain way that no game would ever do - even if it had the best graphics in the world - because games aren't rendering a mostly static image with little geometry and no AF setting

Well, there are still a few problems :

• Take an old game, play it with your brand new card. Not alot of geometry. Simple shaders. Wow, black screen ? That could happen. It's worth a try. Any idea of a game or bench ? I'd say 3dMark2003 or 06... or any game of that era that uses simple shaders. Can't think of one at the moment.
• GPGPU applications. They may reach this value also. They don't have to wait for geometry there. ATI will have to lower the values also. Even if i don't think they'll reach this value, my limited experience in the field cannot answer that question for sure.
• You buy a card that boast it can do X, yet when you have it, it can't. I'd say it's a problem.

Now, right now, what ARE the implications of this problem ?

• Limited overclocking margin. You'll reach it quickly.
• You cannot run any 3d app you want on your card. The list is not that huge, i'd agree, it's not that bad. But morally, ethicaly, this comes into question. Especially since this list can grow in the future.

I have checked the logs, AMD downloaded OCCT yesterday. I guess they must not be taing the problem lightly, or at least, they're taking a look at it

It's true it is not a bug that will make everyone ditch their card, or make ATI/AMD recall their card, IMHO. But still, it's a flaw, a quirk in their design, and worth notice.