Ok so I have the temps back to where they should be. What a relief!

It seems that the heatsinks do not sit flush on the dies - and I had to use twice the normal amount of ceramique to get proper thermal conduction. It's still not where I want it to be, but maybe the MX4 will change that.

Thanks for all the help, anyone recommend lapping the HS?

I just finished re-doing the thermal paste with ceramique - this time using the tint and bead method. Surprisingly, the temps now idle at the same 51C - so it is clear to me that this is definitely a thermal conductivity issue.

The problem is that now both cards are above 80C at full load. I'll wait 24 hours to see if this improves, but I went ahead and ordered the MX4 to see if that helps. Likewise, I will keep this thread updated for anyone down the road experiencing similar issues.

I'm considering going down the watercooling route - as this might be cost-effective in the long run (if I end up with 4 6990s, so I don't need a box fan 24/7).

I was looking at the mcw60 block, I can get two of them for around $30 each. Does anyone know of a cheaper alternative to these blocks? Or a resource for watercooling GPUs?

Thanks for all the help guys, I was freaking out that card #1 was somehow damaged - and it looks like that's not the case.


Also, is mineral oil submersion an option?

I've taking note of these new practices - and will try "tinting" and just applying a pea. Thank you for the reminder.

Although, I wouldn't call it "the wrong way" if it's what worked 15 years ago, they have just improved upon an old practice... and I was unaware of this improvement.

What do you mean when you say that a GPU is not the same as a CPU? Are there according methods for application of the thermal paste?

On graphics cards? Which card? 6950's? This is so frustrating =(

I have to tone back my hash rate on the hot card from 362 to 165. 

I tried reseating 3-4 times with no luck. If it doesn't improve by this time tomorrow I may consider getting that better thermal grease that Jack is talking about.

I still don't see how the paste could be the problem though - maybe I should try sanding the heatsink to get it to conduct better? Anyone know of any other mods to get the heatsink to sit better on the die?

That sounds reasonable - but doesn't explain why one card is running so much hotter than the other - their idle temps are 10C apart...

I have two Sapphire 1GB 6950s, I have been running them for about a month - and getting the following consistent stats:

Card #1
Core Voltage: 1120mV
Core Clock: 920mhz
Memory Clock: 520mhz
Fan: 100%
Temperature: 78-79C

Card #2
Core Voltage: 1120mV
Core Clock: 920mhz
Memory Clock: 520mhz
Fan: 100%
Temperature: 76-77C

and getting 725 megahash total.

So I thought it would be a good idea to open up the cards and replace the thermal paste. In my experience, this normally lowers temperatures. Much to my dismay, card #1 is running 84C and card #2 I have been shutting off (stop mining) after it hits 86C - which takes 10 seconds to get there.

All of my stats and the specs are the same (except the temps/Mhash), I have checked them in both GPU-Z and MSI afterburner (what I use to OC).

I was suggested in IRC to try using zip ties because the heatsink does not sit well in these cards - I was shown this photo: http://imgur.com/TWF7Z

I tried that, but it hasn't changed anything.

The thermal paste I used is Artic Silver, and I have many years experience applying it evenly and in a thin coating. I have reapplied the paste 2 times (cleaning with alcohol wipes) to card #2 with no avail.

Any help would be really appreciated - this makes no sense to me!

Thanks!

Can you cite that economist?

What happens when your prediction is wrong? I bet you will stop replying to this thread...

The -f flags do nothing - but disabling hardware acceleration on the flash menu seems to have worked. I also disabled all of the options on the video section in catalyst - maybe that helped too.

Thanks guys! Now I can watch my youtubes

Why don't you look up your transaction, so that it can be confirmed that you actually sent the btc?

I have this problem where my computer freezes when I watch a video. It only happens when I am mining on both of my Sapphire 6950s. I have tried re-installing all of the drivers / catalyst - and I used driver sweeper.

I have tried resetting my cards to their default clock settings - doesn't fix it.

If I am mining on both cards, and click on a video (on youtube or whatever) - the computer instantly freezes, and I have to restart it.

Anyone else have this problem, or know how to fix it?

Hi, I would like to post on the mining forums to ask questions about a mining problem I am having. As for my cred, I have been running 980mh/s for the past week and a half. I split it between slush's pool and double c's namecoin pool.

I use guiminer with poclbm, but have tried phoenix. I overclock my cards, and pull 750mh/s out of my two 6950s with locked shaders.

I'm a regular in IRC - under the same name as my forum account.

Just trying to ask a few questions about the 6950s, and catalyst 11.6 in the appropriate forums.

Hi!

that did it! thanks, lol

I have tried downloading Afterburner 2.2.0, editing the conf file - and still I cannot decrease my memory clock to below 450 mhz on windows 7.

Does anyone else know how I could decrease the clock past 450mhz - aka remove the under limit?

Any help would be appreciated, I have spent over 2 hours trying to figure it out!

I'm considering it, but have no idea how to run a bitcoin miner and namcoin simultaneously on windows. I have a separate machine I could mine namecoins on, but i don't want to waste time on it if lamecoins don't have a viable future...

I keep getting the error: Problems communicating with bitcoin RPC

My poclbm.ini reads correctly when I view it:

{
    "profiles": [
        {
            "username": "myusernamel",
            "balance_auth_token": "",
            "name": "thurjun72011",
            "hostname": "localhost",
            "external_path": "",
            "flags": "-w 256 -f 1",
            "autostart": false,
            "device": 0,
            "password": "mypass",
            "port": "8332"
        }
    ],
    "bitcoin_executable": "E:\\coinage\\Bitcoin\\bitcoin.exe",
    "show_opencl_warning": true,
    "show_summary": true,
    "show_console": true
}

Anyone know of a solution to this? I realize that it is not connecting to my bitcoin.exe correctly, but I don't understand how to fix it!

Thanks for the awesome GUI, and solid community.