5970 Waterchilling REV 2 Final

[Manicdan]

dude i dont think VRMs are suppose to reach those temps. the gpu should have throttled i hope

[Scaniris]

Crazy temps! You'll kill your card @ these temps!!!

Phase 1 temp - 149C
Phase 2 temp - 155C
Phase 3 temp - 146C

Don't you think this is maybe a bug with GPU-Z? Because the ''1'' before the 49, 55 and 46 seems to be a non wanted guest. And your GPU temps are normal...Hope this is not bad contact with the block. Did you have some lockups? Because @these temps... check asap the contacts on your card.

What is the block you're using?

Here are my temps after 1h30 of Furmark with standard watercooling and EKWB block:

[Dude_Person]

Definately a bug or total lack of contact. My cut up P2 heatsink cools better than that... If those are the real numbers your lucky your card isn't cooked.

[mechman]

I have a Koolance AR-597 on my 5970... GPU temps are good, under 45+C with 1.25V/1ghz running Furmark, but the VRMs getting really hot... up to 130C.
With no overclocking/overvolt they get up to 75C, idling around the same as the GPUs, 35C.

I need to tear it apart and check for bad contact I think.

It would be hard to believe those temps! Check contact for sure or a a different version of GPU-Z ASAP!!!

[nusihb]

I'm using the latest version if GPU-Z... 0.3.8.
And yes, it throttled after a few minutes at that speed.

Temps when not overvolting are more decent though, around 75-80 when running at 800/1200.

Anyways, it's coming apart tomorrow...

[Dude_Person]

75-80 is what my chopped P2 sink gets at stock, so your block should do much better.

[nusihb]

I removed the Koolance wb to see if the VRMs had bad contact or something, and looking at the pads it seems like that's what it is - but you can't tighten the wb anymore so it doesn't matter.
Tried some other pads as well, no improvement.
This block is heading for the junkpile... on to the DD.

[Dude_Person]

hmm. You could try sticking a copper plate with some solder or TIM in there... Not sure if it'd work, or if you are willing to try it. Maybe koolance has some insight.

[wez]

hmm. You could try sticking a copper plate with some solder or TIM in there... Not sure if it'd work, or if you are willing to try it. Maybe koolance has some insight.

A thin copper plate + adhesive TIM is great for these sort of things, not optimal but works very well Done this on motherboard pwn coolers before, where the mosfets where soldered uneven.

But imo you shouldn't have to do it on a card-specific block like this...

[nusihb]

Evilclocker is going to test the same Koolance block later today (or tomorrow), it will be interesting to see if he has the same problem.
The VRMs looks pretty unevenly mounted on my card though, so maybe its more my card than the wb itself.
I tried with different TIM pads, some with copper embedded but they were just too thin to work on this block - the mounting mechanism prevents the block from being 'over tightened'.

[Wolf132]

Just cut a thin aluminum plate to bridge the gap between the vrm and the heatsink. Im sure you could make it work.

[EvilClocker]

I got my Koolance Mounted Last Nite.. I post some pics tonight.. The mount is PERFECT! I was really Impressed with the block so far... I do some testing this weekend.. When I looked at my TEMPS on VRM with GPUZ I was at 25C on the cores and 35C on the VRM.

MAKE SURE YOU US THE PINK/WHITE PADS AND REMOVE THE FILM FROM THE BACK.


More soon

[EvilClocker]

nush.. do a test for me.. Put Some Cheap o white paste on top of the pink pad where the VRM are.. and do a mount then remove the block. and check contact.. take a pic and post it on here

[nusihb]

I have the same temps... until I start to load with Furmark (and overvolt).
It's cold in my 'computer room' (around 60F/16C), so my watercooling system is indirectly 'chilled'

As I said above, I tried already with different TIM pads - no difference.

[EvilClocker]

What Brand 5970 do you have.. I have found they are NOT all the same componet wise..

[nusihb]

Powercolor... with a really low serial number.

[EvilClocker]

Send me some pics of your card... so I can compare to mine

[nusihb]

I'll do that when I rip it apart again tonight.
The problem (I think) with the Koolance block I have, is that the 'nuts' on the wb itself has a flange(?) on it, helping when you mount it but also limiting how much you can tighten it.
I guess if I grind them down a bit, I could tighten it a little bit more and increase the pressure over the VRMs...

[jaredpace]

Here are my temps after 1h30 of Furmark with standard watercooling and EKWB block:

Got a pic of load temps @ 1.3v?

[Romir]

Got a pic of load temps @ 1.3v?

Thats my XFX 5970 @ 1000/1225 1.3 with an EK block. MSAA is on like the previous furmark screenshots. Looks good right? Now lets see what happens with MSAA turned off.



32c cores and 140c VRMs. That screams bad contact but the pads have very strong imprints and been checked over half a dozen times now. The side of the block by the vrms also measures extremely high temperatures with an ir temperature gun. What really stands out to me is the VDDC current, 90 and 94 amps? I haven't found any other results to compare that against so I'm not sure whats up.

Furmark power results:
313w idle (209w no cpu oc)
346w 1 prime95 thread (simulating furmarks 100% one thread load)
590w stock load - 48a typical -244w gpu draw
700w 5970 spd - 68a typical - 354w gpu draw
848w 1000 1.3v - 88a typical - 502w gpu draw

[X2beach]

I will not be doing long periods of Furmark Testing on my 5970.. It Puts Way to much load on them....

[nusihb]

So I'm not the only one with a space heater then

Romir, could you try the following settings:
Core: 1.25v/900mhz
Mem - no overvolting - 1200mhz

..and see what VRM temps you get?

[jaredpace]

weird, wonder why 0x MSAA is making vrms goto 120c and only 75c with 4x MSAA @ 1.3v. Couple questions:

What happens to temps @ 1.3v 4x MSAA if it runs longer than 100 seconds?

What temps do you get at stock voltage 0x MSAA?

And finally, are those 120-140c temperatures real (verified with your Temp IR gun on back of card or block)? Just making sure it isn't a sensor glitching.

[Romir]

nusihb and jaredpace, I'll get back to you two later tonight when I can run the system without tripping the breaker.

I can say the stock temps and vddc draw, without MSAA, are within 10% of the 1.3v MSAA results. There's a 10~ degree rise on the gpus and gpu1s vrms go into the upper 50s and low 60s. The temperatures peak very quickly and remain flat. I currently have the system in an unheated room with two PA120.3s and a MCR320 dedicated to the GPU, hence the 20c~ idle and 30c~ load.

[EvilClocker]

Romir.. Be carefull running furmark and 1.3 Vcore on you 5970.... The VRM Get Hot very hot... Doesnt matter how well u cool them.. I had my Water Temps down to 15C the block was COLD.. adn the VRM still get very close to 120C Just so much heat coming out of such a small spot..

I Currently have (3) 5970 in my hands Sapphire, XFX, HIS and they are all little different.. componets different heights ect ect