Attention HD5970 owners

[Scaniris]

Mmmm ... I showed the photo to my wife and she's not sure that the VRM is higher than the other.... I don't have better photo of my card.

[Eddy_EK]

Your des look higher, but you should not worry. AMD does not even have a active cooling on these chips.

[kiwi]

Does that CPL chip gets really that hot that it caused hardlocks?

[jcool]

In my experience, they do. As Scaniris said, with the stock cooling design, the chips get constant airflow from the fan. If that airflow is missing, well...

[Scaniris]

To be honnest most (all) of the lockups issues disapeared after driver update with version 9.12. I did not reproduce lockups... but I've tested only one day with new driver. And what's reallystrange ist that some lockups occured after block installation, not before... but maybe before I didn't use the card with same apps or during same 3D load. I don't know.

Something else: I NEVER had any problem with Furmark... after many hours of stability test. Seems to be higher on my PCB but I prefer to make sure and ask him to double check.

And last thing: if this component is critical, why don't we have any monitoring of it? I mean if this component was critical and subject to high temps, ATI would have probably put a sensor on it? I don't know ... maybe?

But anyway this is really important that people know that it MAY happen to them. And check component high during block installation. And what's the workaround to fix it.

You know when I received my GTX285 PCB Rev 2 waterblock... I first had a look on forum to have feed back about the mounting. And on the XS forum I saw a thread talking about bad contacts at the end of the waterblock between VRM and block, because the PCB was bended even if using the spacers. As I saw this BEFORE mounting the waterblock, I applied the workaround proposed by people on XS and it was perfectly working, I didn't have to do it later. Great to have people sharing this kind of information. Very helpful for beginners like me.

[Diverge]

And last thing: if this component is critical, why don't we have any monitoring of it? I mean if this component was critical and subject to high temps, ATI would have probably put a sensor on it? I don't know ... maybe?

It's an inductor, a passive component. Basically, it's a coil of wire, that stores energy in switching circuits. It's rated up to 125 deg C per the datasheet.

[jcool]

Something else: I NEVER had any problem with Furmark... after many hours of stability test.

Did you run the renamed exe, not the FUR.exe? Because if you run FUR.exe the card will downclock to 550 core and NOT crash, at least for me

[Romir]

Mmmm ... I showed the photo to my wife and she's not sure that the VRM is higher than the other.... I don't have better photo of my card.

It is a little higher on mine, like pictures in yours, but the thermal pad showed no contact marks like the one the OP posted.



I don't recall many hard screen freezes except when trying to run Vantage above 1025 core. Even furmark only throttled when the VRM temps became obscene at 1.3v.

[sargatanas]

yea I think it's due to the drivers, cause lookups only appeared at 2d for me, you should try to change the 2d clocks, helped few ppl.

[lowdog]

I think all the problems are caused by the fan control in the drivers via cards bios. All my locking/freezing problems started when I put the EK block on my 5970, before that it was fine and rock stable. After days of windows locking up in 2D desktop with artifacting and then getting to the point of not even being able to run any 3D game without system hardlocking I decided to try something.

Did a fresh os install with EK block still on the card, installed latest cats and tried to run some 3D benchmarks - 3DM06, stalker COP etc.....all hardlocked yet if I disabled Catalyst A.I and ran single core there was no lockup

Took off the EK block and put stock cooler back on with THE FAN CONNECTED and bingo, NO MORE PROBLEMS!!!!

Looks like the cards bios and drivers gets confused when there is no fan connected to the fan pins when watercooling the 5970 and this seems to result in all sorts of problems.


EDIT; problems have started again so it's not the fan at all it's the EK block. It's not fitting the card correctly in some way and it's causing the card to become unstable after some use.

[Lord_Farken]

I think all the problems are caused by the fan control in the drivers via cards bios. All my locking/freezing problems started when I put the EK block on my 5970, before that it was fine and rock stable. After days of windows locking up in 2D desktop with artifacting and then getting to the point of not even being able to run any 3D game without system hardlocking I decided to try something.

Did a fresh os install with EK block still on the card, installed latest cats and tried to run some 3D benchmarks - 3DM06, stalker COP etc.....all hardlocked yet if I disabled Catalyst A.I and ran single core there was no lockup

Took off the EK block and put stock cooler back on with THE FAN CONNECTED and bingo, NO MORE PROBLEMS!!!!

Looks like the cards bios and drivers gets confused when there is no fan connected to the fan pins when watercooling the 5970 and this seems to result in all sorts of problems.

So what can be done to keep using the water block
get micro fan and plug it
I got the same water block in the way and I'm gonna get to work no-mater what

[Scaniris]

Did you run the renamed exe, not the FUR.exe? Because if you run FUR.exe the card will downclock to 550 core and NOT crash, at least for me

No, as shown in screen shots, no downclock during furmark stress test.

Hello,

After EKWB FC-HD5970 installation yesterday, here are the temp results after 1h30 of Furmark @ 1920x1200 with MSAA @ x8.

The GPU Z screen shots show maximum temps recorded during the benchmark. The GPUZ windows up shows GPU1 info, the second window down displays GPU2 info.

BEFORE the stability test, here are the monitoring informations:


And DURING the benchmark, a screen shot of monitoring information (the Furmark was running in front of all the rest during stability testing, here it's running behind only for the screen shot, I used the Alt + tab function to switch windows)

[Scaniris]

It is a little higher on mine, like pictures in yours, but the thermal pad showed no contact marks like the one the OP posted.



I don't recall many hard screen freezes except when trying to run Vantage above 1025 core. Even furmark only throttled when the VRM temps became obscene at 1.3v.

I can't check this anymore... but the new owner of my card is aware of this and if he has some issues, he knows what to do. I never tried to overclock my card and lockup happened probably because of the drivers.

[lowdog]

New conclusion it has nothing to do with the fan being connected.

Have interchanged between remounting the EK block and stock cooler 4 times now and each time with the stock cooler the card preforms in a stable manner in both 2D and 3D gaming.

Each time with the EK block within an hour of running problems start, games locking and freezing, artifacts in 2D desktop followed by driver recovering issues and system locks.

Looks like the EK block for 5970 isn't designed correctly and needs to be revised. I'm pretty gutted to have paid so much for this block and have it cause nothing but problems when mounted on my card.

[jcool]

Did you try what I suggested in post #1?

I received both the EK and and Aquacomputer block for testing today. If I get around to it I'll post definite results tonight..

No, as shown in screen shots, no downclock during furmark stress test.

You do realise your screen shows 550 core clock for GPU 1?
Just run the renamed exe and it won't downclock.

[lowdog]

Did you try what I suggested in post #1?

.

Yes I tried the so called "fix" but it made NO difference to the hardlocking issues when the block was installed. Tried 4 remounts of the EK block and 4 remounts of stock cooler over the last few days.

Stock cooler works fine, EK block doesn't!!!!!. The block needs a re design simple as that because the card becomes flakey when it's installed on the card.

[Scaniris]

You do realise your screen shows 550 core clock for GPU 1?
Just run the renamed exe and it won't downclock.

you're right I missed it! Sorry for my mistake... in fact it's running @550.

But I can't test it because I swapped this HD5970 for 2x GTX295... So... the new owner of my HD5970 didn't noticed any problem for the moment, I warned him before the deal that I had some issues so he's aware about it but until now didn't report any issue. After driver update during 1 day I didn't had any issue too. But you're right, my Furmark test was not good, and thank you for the information I'll take care with my future Furmark stability test to run the renamed.exe. I don't know what it is but I'll find it. Thanks

[jooney]

My Asus 5970 did not have the step so i took a second pad and everything works fine with EK cooler.
Eddy please put a 1mm and a 1,5mm pad in the box for those who dont have the step,thats it.



Furmark 8xMSAA 1000/1200@1,2V

[Kurz]

Curious why doesnt my Card go up to 3d clocks in windowed mode while running Furmark?

[jcool]

Ok guys,

sorry for the delay. Gonna set up a test rig with the sample cooler Eddy sent me, and the original Diamond brand HD5970 I was having the issues with. I'll report back in an hour or so...

[jcool]

Darn.. same freezing issues with the Diamond card.
I just tried pluggin in the stock fan while the card is still watercooled, makes no difference. So definitely not a fan bug.

Could somebody with a working 5970 + waterblock please send me their VGA bios? I would really appreciate it. Unfortunately I forgot to copy the Sapphire's bios because I was in such a hurry
Starting to think this Diamond bios is fishy.. or something else is wrong with the card.

[jcool]

Sorry for triple-posting () but trust me, this one deserves it. You think you've seen everything after more than a decade of system building and then hardware still surprises you.

My Diamond HD5970 card has a COLDBUG at 35C P O S I T I V E (!!!)





That's why it works fine on air cooling, but freezes on water. Don't believe me? Hold on.



Weird FUR graph right? Let me explain.

I am running a card-only test loop here, that means card, triple rad, and pump with res. I disconnected the pump so the card could heat up a little. THEN I started up the renamed FUR.exe and let it run (with the pump unplugged, so no flow).
It began heating up quickly, but didn't freeze.

At around 80C core I plugged the pump in for a few seconds so the card would cool down again. I did this a few times, represented by the Arcs you see in the picture.

Then, after a few cycles, I reconnected the pump for good. Still running fine, core temp went down to about 40 (25C ambient in here).

Next, I opened the window. Temps dropped quickly and voila.. screen froze right there where you see it (that's why I fotographed it).

And yes it's reproducable.

So this is NO issue with EK Waterblocks (or any other waterblocks, for that matter), even though it can't hurt to have the little CPL chip make contact properly

Jeesus. I just hope flashing anotehr bios will solve this crap, otherwise I got a perfectly useless card (useless for me anyway).

Or maybe I'll just run it at a constant 1,3V

[Scaniris]

Hey.... could this be really possible? Crash because of low temps?

[sargatanas]

well my 2 powercolor do not crash at lower temps

[zalbard]

Coldbug at +35C?!