Attention HD5970 owners

[coltsneck]

WOW! OK, so now what should I do? I just purchased the EK block for my MSI 5970. My CPL chips are the same height. Do I do the double thermal pad? Do I return the EK block? Can't return the 5970 obviously. Where do we stand on getting a solution?

I wrote EK-Eddy about the even height chips (not about this cooling bug) and received this reply:

I have already explained and the other users also that this does not change anything.

What is really causing the problem is not determinated and what I have found for now is that the issue are some of the cards.

One user found tha card has a problem that the block cools to good. I'ts the user that first reported that problem.

We have allot of very happy users with card like you and with those two chips at the same height.

And like explained in updated manuals that can be found here:

http://81.90.180.2/ekwaterblocks/sho...1109835128.pdf

That you just use additional layer of 0,5mm thermal pad and the cooling will be 100% effective.

I would also like to express the fact that those chips does not need much cooling as you probably say the stock cooler does not even cool them.



So I am very confidend that the block you have will cool the block as effective as it could do if there were no step.



Happy New Year and best regards,



Eddy

[BlueAqua]

I'm using a 5970 with an Ek block installed and it's fine. Just give it a try.

[ckool]

Has anyone had any problems with water cooling an XFX branded 5970? i am in the process of purchasing 2 of them.Gonna be using Ek wb if that matters.

[mark84]

Signed up just to post in this thread

I've got similar problems with my Powercolor HD5970 and the EK water block.

With the stock cooler it's fine, but for me, with the water block attached about 10-15mins into a gaming session the PC just locks up/freezes without warning. No artifacts or driver resets or anything just sudden lockup.

I've tried several driver versions but that doesn't change anything.

After some 6 odd remounts of the block I've put double the thermal pads in areas I noticed there wasn't much/any contact with the block, so everything is contacting correctly.

Temps are good, about 30C idle on GPUs and 50C (off the top of my head) for the VRMs. Under load GPUs are in the low 40's and VRMs in the 70's I think it was.

It's weird though as it take about 15mins before it'll lockup. Two games I've been able to play without locking it up. Ghostbusters with the 30fps limit set in the game menu (60fps option locked it up after a bit). The other is Star Wars: Unleashed, played that last night for a few hours without problems, though that was after a remount so might try Dirt2 or something again tonight to see if the lockups still happen.


Just thinking, if the CPL chips apparently don't need cooling, what would happen if we don't apply the thermal pad on those chips?

[coltsneck]

This is really pretty disturbing. Considering the price of the 5970s why were they not built exactly to reference? Surely ATI had to know that this kind of card will primarily be sold to enthusiasts who will 9 times out of 10 water cool. Then there is the issue of the "too cool" CPL chip shutting off system. Maybe I should have stuck with Nvidia!!!!

Likewise here I find fault with EK. Given the premium price for their 5970 WB they should offer to exchange for one without the step for those of use with even CPL chips. Sticking an extra piece of thermal is a patch-work solution and frankly I think possibly prone to error or future failure.

When I pay first class prices I want first class service.

Geez. If I were to fly first class I want a first class seat..........I don't want the stewardess/steward to tell me they have no 1st class seats available, but they're gonna give me 2 seats in coach instead !!!!!!!

[r1ch]

This is really pretty disturbing. Considering the price of the 5970s why were they not built exactly to reference? Surely ATI had to know that this kind of card will primarily be sold to enthusiasts who will 9 times out of 10 water cool. Then there is the issue of the "too cool" CPL chip shutting off system. Maybe I should have stuck with Nvidia!!!!

You're misunderstanding.

9 out of 10 don't watercool. :p

The CPL chip will benefit from any additional cooling, but it's not "too cool" - re-read jcool's 1st post. It's a specific issue with his 1 card.

You sound like a WUM. :p

[lowdog]

You're misunderstanding.

9 out of 10 don't watercool. :p

The CPL chip will benefit from any additional cooling, but it's not "too cool" - re-read jcool's 1st post. It's a specific issue with his 1 card.

You sound like a WUM. :p

It's NOT just a specific issue with his 1 card!....it's much more than that. Read around the net and see all the people having problems with watercooled 5970's.

This problem is specific to 5970's and watercooling simple as that and until someone finds a fix to this problem after having it happen to them once they watercool the card then none of us will know why it happens and how to circumvent it if we want to watercool our 5970's.

[mark84]

It's NOT just a specific issue with his 1 card!....it's much more than that. Read around the net and see all the people having problems with watercooled 5970's.

For the OP it might well be. He's demonstrated his lockups are temperature related. If his gets too cool it'll lockup.

Yes there's issues with 5970's and water cooling, but for jcool it appears to be temperature related, so called "warm-bug".

My 5970 on the other hand doesn't seem to be effected in that way, mine runs for 10-15mins just fine before locking up/freezing. And all temps are well below stock cooler temps but certainly don't get cooler the longer I play hence why I'm confounded by the problem :p

[lowdog]

I have a theory, if the second gpu isn't being utilised and goes into low power mode as set by the drivers then it will cool down perhaps too much and this bug may appear. Simillarly some games only use one core so you can play them fine but then if you try to use a benchmark program that uses both cores/gpu's ie CFX then the comp will lock because that secong gpu has been inactive and now want's to be used but is too cool so temp bug happens and comp locks????

[mark84]

Possible.

I found mention of disabling the low power mode via the registry here. Will try that out when I get home.

I think though mine locks up in crossfire enabled games rather than non-crossfire enabled games. Hence why I seem to be able to play less demanding games ok.

[lowdog]

Possible.

I found mention of disabling the low power mode via the registry here. Will try that out when I get home.

I think though mine locks up in crossfire enabled games rather than non-crossfire enabled games. Hence why I seem to be able to play less demanding games ok.

Mine was doing the same thing as well, could play games that didn't use CFX but when I tried to use something like 3DMark that used both cores = lockup. When the waterblock was first installed CFX worked fine but then after time it went to crap and nothing worked, even 2D desktop started artifacting and causing driver recovery problems.

[coltsneck]

You're misunderstanding.

9 out of 10 don't watercool. :p

The CPL chip will benefit from any additional cooling, but it's not "too cool" - re-read jcool's 1st post. It's a specific issue with his 1 card.

You sound like a WUM. :p

r1ch,
When a new high end card comes out at peak (over-priced) $$$ then I think it is mostly gonna be the hardcore who will by it. Whether 9 of 10 of these folks actually water cool is beside the point (it was rhetorical)......it doesn't matter if it is 0.9 of 10. No customer who spends the premium dollars should have to put up with such mundance, elementary silly mistakes. All customers, especially the early adopters, should be treated fairly and in accordance with a company's published quality statements. Companies that fail to due so are acting like the GMs and Fords of the 80s and 90s and see where that got them. This is, IMHO, a significant black-eye for ATI.

ATI has always had, rightly or wrongly, a reputation for being a bit sloppy and slow with drivers and updates; at least as compared to Nvidia. Additionally all of Nvidia support was generally good....especially if compared with the majority of HW vendors out there. ATI can build the fastest cards all they want, but if even just a small vocal minority have issues it will hurt them in the long run. Particularly if they don't quickly and accurately address them.

What is a WUM anyway???

[coltsneck]

BlueAqua,

Which water block are you using?

[BlueAqua]

BlueAqua,

Which water block are you using?

An Ek nickel/acrylic. It's a beauty and running great. The only problem I have are the drivers.

[Lord_Farken]

Signed up just to post in this thread

I've got similar problems with my Powercolor HD5970 and the EK water block.

With the stock cooler it's fine, but for me, with the water block attached about 10-15mins into a gaming session the PC just locks up/freezes without warning. No artifacts or driver resets or anything just sudden lockup.

I've tried several driver versions but that doesn't change anything.

After some 6 odd remounts of the block I've put double the thermal pads in areas I noticed there wasn't much/any contact with the block, so everything is contacting correctly.

Temps are good, about 30C idle on GPUs and 50C (off the top of my head) for the VRMs. Under load GPUs are in the low 40's and VRMs in the 70's I think it was.

It's weird though as it take about 15mins before it'll lockup. Two games I've been able to play without locking it up. Ghostbusters with the 30fps limit set in the game menu (60fps option locked it up after a bit). The other is Star Wars: Unleashed, played that last night for a few hours without problems, though that was after a remount so might try Dirt2 or something again tonight to see if the lockups still happen.


Just thinking, if the CPL chips apparently don't need cooling, what would happen if we don't apply the thermal pad on those chips?

that a good idea to try,if mine do it ,I will try it
the stock cooler in reality is passing the heat from the gpu,s to them so they
always warm , now come the water block keeping everything super cool
according to ek eddy that don't need cooling ,so skipping the pad should be fine
however if the problem is the too cold gpu , were are screwed
ati probably need to do a revision ,to totally fix this
but don't think they are in a hurry because they selling like hotcakes
and it work fine with stock cooler
in other though why don't try to RMA ,just don't mention the water block
say that crash stock

[individual]

Companies that fail to due so are acting like the GMs and Fords of the 80s and 90s and see where that got them. This is, IMHO, a significant black-eye for ATI.

Blame AMD, they bought ATI.

trying to sound as un-technical as possible, but maybe the waterblocks are radiating lower temperatures to components of the board that arent meant to be cooled, eg. sensitive resistors, thermistors. I would suggest maybe put a peice of insulating foam, over the main board itself, of course cutting out areas where the waterblock makes direct contact with the chips, just a thought.

I know boards i used to work with at Redarc electronics, would have to be encased in silicon, not only for water protection but because if some of the components dropped below a certain level it would "mess up" the fine currents in some places, causing the board to malfunction. That is my theory with the 5970, I'm suggesting that it causes miniscule current/voltage/mA peaks and the card can detect it as a saftey feature, thinks it has been shorted, and freezes. Sounds a bit out there, but it happens, and any electronic enthusiast will know the colder components are the better they perform (contradiction?)

[Levesque]

My 5970 with EK waterblock is now working fine with the ULPS registry changes I made 2 days ago.

I can game, Furmark all I want, surf the net, Photshop, etc for hours without a single problem/freeze/bug.

My only ''problem'' is if I ''cold'' boot, I have the message ''Display Driver has stopped responding but successfully recovered'' only once, 20-30 seconds after boot-up. If I start Furmark for 30 sec (to ''warm up'' my 5970) the second I have access to my desktop, then I don't have the error message at all. And then I'm good to go non-stop! Strange, but it's working for me.

5970 idle temp is 21 celsius at start-up and 22-23 after stabilisation. Never goes over 33-34 celsius with Furmark, and VRMs at around 55 max at full load. VRMs are at 24-26 idle.

Probably a driver problem IMHO. I just let my computer on 24/24 until better drivers solving that problem.

[individual]

Levesque, if you don't mind me asking, what are your water-loop specs? pump/fans/rad(s)/block(s)

[Zeus]

Anyone tried to hook up a fan to the fanheader? Anyone took a infrared thermometer to the pcb? There might be a small component missing some air from the original fan and just overheat? Just a thought.

[Lord_Farken]

Anyone tried to hook up a fan to the fanheader? Anyone took a infrared thermometer to the pcb? There might be a small component missing some air from the original fan and just overheat? Just a thought.

yes it ben try already and not a overheat

[mark84]

that a good idea to try,if mine do it ,I will try it
the stock cooler in reality is passing the heat from the gpu,s to them so they
always warm , now come the water block keeping everything super cool
according to ek eddy that don't need cooling ,so skipping the pad should be fine
...
in other though why don't try to RMA ,just don't mention the water block
say that crash stock

Let me know if removing the thermal pads works for you. I CBF remounting my block again for the moment
Keep in mind though that on the stock HSF there is a pad for the CPL chip located between the two GPUs. So maybe that one needs cooling?

I can't really RMA mine cause it works perfectly with the stock cooling attached.

Tried the ULPS registry thing last night, but I still got lockups in game. My issue doesn't seem to be down clocking/power saving related so I didn't expect it to help. Desktop/2D is fine for me.

I tried disabling Catalyst A.I in CCC and whilst I only had enough time to run it with that setting for 30mins, it didn't crash. Might try it again tonight. It seemed to disable crossfire though as the FPS counter in Dirt2 basically halved with A.I disabled.

[Levesque]

Levesque, if you don't mind me asking, what are your water-loop specs? pump/fans/rad(s)/block(s)

I just did it again to confirm it. Power off the computer for 1 hour, then boot back in Windows, and the second I can do something on my desktop, start Furmark for 30 seconds... No error message, no problems. Surfing the net, playing game for 1 hour... No problem.

Then power off computer again 30 minutes to cool down. Power on, don't start Furmark when getting on desktop, go surf on the net. ''Display driver stopped working but has successfully recovered'' message again.

Crazy, I know, but happening in my system.

I have 2 independant loops. One for CPU and one for GPU. GPU loops: EK-5970 waterblock (the full copper + plexi one), Swiftech 655, Swiftech MCR320, 3X S-Flex "E". Asus 5970, and I had to do the "extra-pad on CP-20" technique.

I had alot of problems before changing my ULPS settings in my registry. None since then.

[JoeBar]

OMG this card seems to be a pain in the a** as of yet. Glad i didn't jump on the train and stayed with my 4870X2...

[coltsneck]

Blame AMD, they bought ATI.

trying to sound as un-technical as possible, but maybe the waterblocks are radiating lower temperatures to components of the board that arent meant to be cooled, eg. sensitive resistors, thermistors. I would suggest maybe put a peice of insulating foam, over the main board itself, of course cutting out areas where the waterblock makes direct contact with the chips, just a thought.

I know boards i used to work with at Redarc electronics, would have to be encased in silicon, not only for water protection but because if some of the components dropped below a certain level it would "mess up" the fine currents in some places, causing the board to malfunction. That is my theory with the 5970, I'm suggesting that it causes miniscule current/voltage/mA peaks and the card can detect it as a saftey feature, thinks it has been shorted, and freezes. Sounds a bit out there, but it happens, and any electronic enthusiast will know the colder components are the better they perform (contradiction?)

Ok AMD = bad

I guess your theory is plausible, but I don't really think probable. There are a couple of reasons why I say that. If the block is not in direct contact with a component the heat absorbtion is going to be minimal. Also, the water in the block is at best ambient temp, and probably after a while a few degrees above. The PCB components will all radiate some heat into the open spaces between the block and the PCB. Given the air in those tight spaces is likely stagnant I would think it would in fact be slightly warmer (not cooler) over time then with the stock blower.

Some of the problems cited sound very much like driver (or any on-board firmware) issues. For example I can imagine the card on start up checking that the fan is operational, but doesn't get a reading and goes into fault condition. It may then however recover because temps are good. Some of the other problems sound like overload or short circuit issues. SO....my theories. Is it possible the block is touching an exposed PCB "wire" when the card gets a little warm and expands just slightly? Are the thermal compounds that folks using completely non-conductive and if not has any oozed onto circuits?

And finally one last thing. When I took my stock cooler off I noticed 2 very small bits of silver metal on the table. I didn't give it much thought at the time, but could this be a more universal issue? Two, albeit extremely small, pieces of metal that don't fall on the table could certainly cause a multitude of problems if trapped on the board. Perhaps some folks should scan their boards very carefully with a magnifying glass. I dunno...just some thoughts here.

[mark84]

The PCB components will all radiate some heat into the open spaces between the block and the PCB. Given the air in those tight spaces is likely stagnant I would think it would in fact be slightly warmer (not cooler) over time then with the stock blower.

I can't see that being too much of an issue. Even a case with mediocre air flow will still move that air around enough.

Some of the other problems sound like overload or short circuit issues. SO....my theories. Is it possible the block is touching an exposed PCB "wire" when the card gets a little warm and expands just slightly? Are the thermal compounds that folks using completely non-conductive and if not has any oozed onto circuits?

Doubt it's a short circuit. The block comers with plastic stand offs that sit between the block and PCB. Gives about 4mm clearance. The thermal pads provided with the block wouldn't be conductive. If they were I'm sure my card would be fried by now