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

[Tetedeiench]

I wanted to bring these topics back to your intention, in case you either missed or intentionally ignored me.



Also, it doesn't really matter if the card crashes on a 1500W or 550W.

If the OCP is, as you hypothesize, set to 82A maximum, then are you not exceeding design specifications by running this test? In this case, is it not the same as overclocking - do so at your own risk?

I didn't ignore your post. It's hard to answer everybody

I actually asked for more information :

• Did you use Shader complexity 3 ?
• Can you take a picture of the VRM of your cards ? here is an example : http://mangafranceworld.free.fr/Divers/DSC00550.JPG
• Can you redo the same test and monitor the VRMA just as i did using RivaTuner, and post the screenshot ? The idea is to see if your card stays below, or higher than 82A on the VrmA. Another user did it on Page 10

If your card withstand the test, either the VRM stage is not the reference one (i.e. 4-stage, and not 3-stage), or you forgot to check the fullscreen checkbox Or it's something new !

[Tetedeiench]

those are very simple protections that have been added to recent voltage regulator designs. if an output stage mosfet blows (overheat) it is possible that it pulls the whole output stage to 12V = bad(tm). ovp protects against that

ocp is for cases where a short circuit occurs somewhere in the consumer circuit. short circuit = lots of current = bad(tm).

in a proper configuration ocp is set in a way that it is never triggered during "normal" use. apparently ati didnt do enough testing and set ocp too low.

in addition to those two, volterra has added overtemperature protection as well, which is triggered when an output stage gets too hot. there are some more fault triggers as well which are not important for us

Many thanks for your explanations Things are much clearer for me now We may have reached one of those limits, indeed.

[kiwi]

For reference, 260GTX (216) 55nm easily passes 30 minutes Temps only 70C, 1680x1050, FPS ~43 (i7 @ 3.9ghz). Give me more juice

I could test reference 4890 after a week or so if it will be still actual

[fellix_bg]

Looks like this GPU test doesn't like overvolted boards!
I've tested my Sapphire 4890 with the default o.c. of 950/1150 MHz and 1.35v for the GPU (in the BIOS) and it just shuts down the monitor immediately, and I had to reset. Going down to 700MHz core and default memory clock rate didn't help. I had to reduce the vGPU to get it running, but still hadn't conducted a thoughtful testing on that matter. FurMark is fine, though!

By the way, the 4890 reference design have 3+2+2 DC phases -- that is 3 for the GPU core, 2 for the memory interface I/O and 2 for the GDDR5 array, AFAIK. The 4870 is 3+2.

[Shintai]

Hehe, this kinda also explains the monthly catalyst updates

AMD gave their view on it...

Your limited testing is flawed. If the VDDC's overheat they will shut down the card to protect it. If you dont want to have a crash when testing artificial means such as this stability benchmark then increase the fan speed to cool the VDDC's. Obviously underclocking acomplishes a similar goal as they dont get as hot. No games are affected just artificial stability tests

http://forums.amd.com/game/messagevi...&enterthread=y

[Warboy]

http://mangafranceworld.free.fr/Divers/DSC00551.JPG

Vtec just failed yo!

Oh this is classic, Especially what the n00b AMD mod said.

[largon]

For comparison:
During OCCT GPU test GPU-Z reports an average vDDC current draw of 82.72A.
During normal gameplay in Crysis GPU-Z reports an average current draw of 42.18A.

OCCT GPU test causes 96% higher current draw than Crysis.
2×

Isn't the 4890 4 phase by default?

Afaik the asus cards (top) have 5 pahses + 2 for memory. ( 4890)

edit: seems like all 4890 have 5 phases + 2 memory (at least the one i have checked: xfx, asus, powercolor, HIS, Club3d)

which brings up another point, why is the 4890 affected by this, it even has a stronger vrm circut then the 4vrm 4870.

HD4890 has 3 core vDD(C) phases just like reference HD4870. In addition, HD4890 also has 2 phases that are yet to be officially identified which are listed as "???" in GPU-Z sensors tab. This unknown phase could be dealing with the GDDR memory controller perhaps, as it's current depends on GDDR frequency yet the voltage these phases generate are not fed into the GDDR5 chips...
And ofcourse, there's 1+1 (mvDD+mvDDQ) memory phases too.

http://mangafranceworld.free.fr/Divers/DSC00551.JPG

Vtec just failed yo!

Oh this is classic, Especially what the n00b AMD mod said.

It's "Vitec" not "Vtec" and that's just an inductor.
A passive component. It's a coil of copper wire. It cannot fail in any to contribute to this issue.

For reference, 260GTX (216) 55nm easily passes 30 minutes Temps only 70C, 1680x1050, FPS ~43 (i7 @ 3.9ghz). Give me more juice

Hehe, my HD4890 @ 850/850 (with underclocked memory and a 720BE @ 3.2GHz.) floats between 80-86FPS depending on the angle of the object.

[Tetedeiench]

largon : impressive

Oh, the best shader complexity vaue for GTX260 is 8 (auto-selected). This lower DRASTICALLY the fps. The GTX260 might not have selected the Shader complexity value of 3... thus the HUGE difference in FPS.

[Particle]

As a developer of both general purpose and DirectX applications, I've got to say that a lot of you guys are very frustrating with your assumptions about engineering and 3D programming. =/

As for my bit:
If the card can be made to draw current in excess of either its supply capability or its current protection threshold using nothing other than normal API commands, there's a design problem. There's nothing wrong with the chip, so maybe the title of the thread is misleading, but the reference cards' power supplies need some work. That being said, this was probably an intentional action. It makes sense to aim for above the middle as it'll provide compatibility with real life games. Making a design that is good for the upper 1% of applications that aren't widely used at an extra cost doesn't make a whole lot of business sense.

[Sparky]

I don't care honestly, my 4870 handles all games, handles furmark, and handles this current virus test too, so I'm happy enough!

[trinibwoy]

Making a design that is good for the upper 1% of applications that aren't widely used at an extra cost doesn't make a whole lot of business sense.

Well hopefully this doesn't extend into the future. Because as hardware becomes more generalized the idea of limiting the scope of "normal applications" will go away. There's no such notion on CPUs and soon won't be on GPUs either.

[Nullack]

As a developer of both general purpose and DirectX applications, I've got to say that a lot of you guys are very frustrating with your assumptions about engineering and 3D programming. =/

As for my bit:
If the card can be made to draw current in excess of either its supply capability or its current protection threshold using nothing other than normal API commands, there's a design problem..

+1000000000000

[MsB]

Ok so we have some that work and some that don't here. I would submit this. I have taken apart many cards to re- tim them with silver or diamond based tims. I have found many times that the factory had installed heatsinks where there was little and in some cases, NO contact in some areas between heatsink and surfaces to be cooled. So could some guys maybe take apart cards to see just how well they have thier thermal compounds or pads installed. Could this be a case of some cards having enough contact or compound to work under normal conditions(meeting factory specs minimum) but not enough to remove heat quick enough for this test? Just sayin.

[lyserg]

i Dont know if it helps u but here :

This was done with a newer Version of Powercolor HD 4870 1GB PCS
800/925 (stock on new revision is 780/925)
BIOS 02/24/09 21:07
ATOMBIOSBK-ATI VER011.013.000.002.000000

Like u said before ,no Blackscreen

But i have the same Powercolor with old Revision,i´ll test it in the next half an hour.

[Manicdan]

its fun watching people bash the card that practically saved ATI/AMD from utter destruction.

[HiJon89]

Ok so we have some that work and some that don't here. I would submit this. I have taken apart many cards to re- tim them with silver or diamond based tims. I have found many times that the factory had installed heatsinks where there was little and in some cases, NO contact in some areas between heatsink and surfaces to be cooled. So could some guys maybe take apart cards to see just how well they have thier thermal compounds or pads installed. Could this be a case of some cards having enough contact or compound to work under normal conditions(meeting factory specs minimum) but not enough to remove heat quick enough for this test? Just sayin.

This thread would be a lot shorter if people would read before posting The OP has stated multiple times that in many cases the crash is instantaneous, the VRMs don't even have time to get hot.

[bluedevil]

ever since i both my new system I've been having random restarts in games like crysis unreal tournament 3 even with no overclock on anything .tested cpu ram with orthos intelburningtest memtest all good , tested the video card with furmark and it was stable but 10 sec into ooct gpu test using the setings provided in first post and boom restart!
what i don't understand is why is it unstable in unreal 3 or crysis after 30 min to 1 hour (no artifacts no wird stuff), it never uses the gpu 99% what the hell is wrong with my card ?
sys spec
Gainward reference 4870 512 mb 3 vddc
E7200
2x1gb kingmax ddr2 1066
asus P5k/epu (not using the epu driver and can't find a epu seting in bios so i guess it's disabled )
thermaltake 600 w (this is my second psu i rma the first one because i thought it was the culprit )
I'm a normal user i really don't care if my card is stable or not in any soft besides games. I've been having this problem for 6 months now with no luck spoting the culprit so if anyone can help me make sure the card is the problem or has any other suggestions for me to try pls pm me

[MsB]

This thread would be a lot shorter if people would read before posting The OP has stated multiple times that in many cases the crash is instantaneous, the VRMs don't even have time to get hot.

I have read the thread bud. I have ran the test several times as well. Have you? It causes heat to increase instantly so my comment is not out of the question. This thread would be a lot better if guys like you would keep your insulting little quips to yourself. and btw some have taken a few seconds to black out like the guy above.

[HiJon89]

I have read the thread bud. I have ran the test several times as well. Have you? It causes heat to increase instantly so my comment is not out of the question. This thread would be a lot better if guys like you would keep your insulting little quips to yourself. and btw some have taken a few seconds to black out like the guy above.

Tetedeiench said that in some cases the donut never even has a chance to appear, the card can't heat up that quickly. Yes in some cases it takes a few seconds or minutes to crash, but since heat can't be ruled out in those cases it's more useful and worthwhile to investigate the cases where the crash is instantaneous because there are fewer variables and the only thing that seems possible of causing an instant crash is the power draw. And yes, I have run the test without problem on HD3870 and an 8800GT.

[MrHydes]

that's a shame for Ati RV770 reference based cards

[BababooeyHTJ]

What exactly is the upper safe temps for these vrms? I ran this with my GTX280 (great program, btw) since from my understanding it uses the same vrms as the 4870 but at stock clocks was a hair short of hitting 82A at 1920x1080 fullscreen but didn't have any problems. I stoped the test since I got a little nervous about vrm temps.

[AKM]

its fun watching people bash the card that practically saved ATI/AMD from utter destruction.

It would be even funnier and sad if this will result in deterring people from buying Radeons.

[Heinz68]

Same like some others I believe the heading is little sensational and even misleading, since OCCT is sure not the only GPU benchmark whether new or not.

Now we have basically two arguments here going on and on.

First even if this can't be duplicated in ANY game, some people including Tetedejench argue MAYBE one day this problem will show in future games. We all know sooner or later all Graphic cards are going to be obsolite. Since the game developers need to sell many games it usually doesn't happen too fast.

The other argument is, nobody plays OCCT so who cares and I kind off agree with that since I don't have any problem with the card at all. Looks like mainly Nvidia guys have problem with the card even if they don't have any.

Now I did run the OCCT shader 3 bench with NO error check and it crashed immediately so it could not have anything to do with overheating.
It crashed so badly that i could not get the Vista 64 Appearance Option till second reboot, so I was telling myself this OCCT really sucks.

Anyway I tried too more settings.

First with Shader 1, NO error check and it run OK. Here is the graph, 1 minute idle, 1 minute test and 4 minute idle.


Than I tried to run the Shader 3 with the error check ON, same times as above and it did not crush at all. The graph looks the same as the above one except the Error Check. The test went in full screen but the "furry wheel" was not moving yet the test looks the same including FPS reported. This Graph reports "Test Completed without errors" which is not right.
So the simple application also has errors. Here is the graph.


The test was run with 4870 X2 but I also tried without the crossfire with same results.

PS English is not my first language so don't get too critical about that.

[Tetedeiench]

Same like some others I believe the heading is little sensational and even misleading, since OCCT is sure not the only GPU benchmark whether new or not.

Now we have basically two arguments here going on and on.

First even if this can't be duplicated in ANY game, some people including Tetedejench argue MAYBE one day this problem will show in future games. We all know sooner or later all Graphic cards are going to be obsolite. Since the game developers need to sell many games it usually doesn't happen too fast.

The other argument is, nobody plays OCCT so who cares and I kind off agree with that since I don't have any problem with the card at all. Looks like mainly Nvidia guys have problem with the card even if they don't have any.

Now I did run the OCCT shader 3 bench with NO error check and it crashed immediately so it could not have anything to do with overheating.
It crashed so badly that i could not get the Vista 64 Appearance Option till second reboot, so I was telling myself this OCCT really sucks.

Anyway I tried too more settings.

First with Shader 1, NO error check and it run OK. Here is the graph, 1 minute idle, 1 minute test and 4 minute idle.


Than I tried to run the Shader 3 with the error check ON, same times as above and it did not crush at all. The graph looks the same as the above one except the Error Check. The test went in full screen but the "furry wheel" was not moving yet the test looks the same including FPS reported. This Graph reports "Test Completed without errors" which is not right.
So the simple application also has errors. Here is the graph.


The test was run with 4870 X2 but I also tried without the crossfire with same results.

PS English is not my first language so don't get too critical about that.

Posting from my iphone, so i will keep it short.

Error check mode = still, unmoving donut. That is wanted. As stated in the first post, this mode lowers the gpu load quite a bit. It is not surprising the problem does not appear.

On the graphs, an error detected =a white dot counted by the gpu test. The counter is updated every 3 sec for efficiency purposes. If the counter is incremented in the test and not in occt, it is a bug indeed. I doubt it impacts the blackscreen bug you are encountering.

[Sparky]

It would be helpful if people would list the make and revision of their card.

The Asus non-reference 4870 1GB has no issues.
My Visiontek reference design v1 (there's a later reference design as well IIRC) 4870 512MB has no issues. Which strikes me as odd that any others based off the reference design would, what makes mine different?

Any others?