Hello,
i had my rigs (2 7970 on Win7 Pro 32bit with catalyst 12.8 and cgm 2.8.7) mining btc for months now (with an automatic restart script every day at 4am just to be sure), steady income, no hangers.
I switched to 2.11.3 because after the fork rumours and pools ceasing merged mining, forcing btcd0.8.1 and the increasing btc/usd ratio i felt like it was the time to upgrade and service my rigs.
What an error i did make there....

I have a quite unstable DSL with Dyn IP, but that was never a problem since 2.8.7 came out.
Now, the machines will just mine until "something" happens, then stop, giving windows a finger. It then shows a window "cgminer has stopped working".
I did not change any settings, just updated the exe.
As the error window states some vga related dll (forgot the name, but was AMD related) has ceased working i thought updating catalyst to 13.8 may help, but it doesn't.

CGMiners window shows no hardware errors, all looks like usual, besides it is not moving anymore.
I did not overclock the cards, although i can't recall the actual settings, but they did work since 2.8.7 came out, and temperature is not a problem, we had colder and hotter days there, both were no problem.
Also, i did not change protocols, i was mining stratum with BTCGuild and ozco.in before and after the incident, using getwork mtred as backup, but that almost got no shares.

What did change from 2.8.7 to 2.11.3 that causes this, and what can i do to stop it?

@FreshJr:
I had a similar problem.
My simple solution was to add a planned task in windows, executed 1h after the automatic winupdate, having the shutdown command.
So my machine updates every night, then reboots.
Since then (almost 8 weeks now) no problems with hanging CGM for more than a few hours, without any touching of the machines.
I had a hang sometimes, needed to plug monitor and keyboard, try to restart, notice restarting CGM did nothing (maybe opencl hung), restart the machine.
On weekends i often had one (problematic) machine often stop shortly after i was on holiday, missed almost 2 1/2 days of work.
Now, the h, and no intervention needed.

The place where my machines (4x 2x7970) work has a unstable DSL, it sometimes reconnects 12 times a day, sometimes only 1.
I have 4 pools in failsafe mode, 98% of all shares go to pool 1, but sometimes even pool 4 gets some shares (0,25 BTC since Dec. 1  )

i can not tell about seti or folding, but a server usually does almost nothing (file or email), rendering does use the components, but not as heavy as mining, at least what i have tried.

@tiberiandusk:
ask Google how short lived usual Server HDDs are.
They have lots of thousands of them, and they saw what the datasheet numbers really meant.

Also, your computers may be powered 24/7 but i doubt they have to do work full load like mining is.
Since few years the hardware mostly sleeps and turns a lot of unneeded parts off, and wakes them up in milliseconds if needed.
That induces stress in the capacitors and the power supply, but on the other hand reduces heat and overall current consumption, which makes the stress wear less.

Maybe.
Capacitors age, that makes the voltages wobble a little bit more.
Then, Silicon migrates over time with temperature and that may shift the tolerances down.
And, last but not least, your sensor readings may not show what really happens there, because there are more points of heat than just the one shown.

Don't forget, your system uses a lot of power and generates a lot of heat 24/7, while almost all components are designed for a max. of 8h/day and 20 days/month, just to live past warranty period (in average that is).
So your hardware will live it's average live in ~6 months, but as you do not only have one unit, you will see the variance in that.
Some units will die after a few months or even weeks, some will work 'forever'.
I personally had made the same experience with 4 6950s, the longest living individual did almost 6 times of the shortest time.
Also, mine had 2 fans each, 1 has 2 blocked fans, 2 have one (both the same), the last has 2 working fans, but the chip gave up and produces blocks on the desktop (no mining!).
I have modified the remaining 3 cards, removed the fans and plugged a 12cm one in, now they continue to work for almost 3 months now.
I got them killed in 4 beforehand.
But i slowly replace them with 2 7970 in my rigs. 340Mh @ 200W -> 1300Mh @ 500W at the outlet is a good deal.
also, no cost for hardware (borrowed).

As far as the pool supports it, yes, and without local nmcd.
For example, BTCGuild is Stratum and Merged NMC/BTC, so i am going for 2.8.7 after the positive reports here.
Slush0's Proxy stopped working for me, so my 2.7.6 miners have switched to another pool instead, because stratum was buggy under win7.

I have experienced similar, that's why my rigs have stratum proxy 1.1.1 running with 2.7.6, flawless for days without interruption, on a shaky DSL that disconnects ten times a day.
Also, higher hashrates than on 2.8.x.
I have that proxy as pool one, the pool the proxy points to as pool two and two other non-stratum pools as pool 3 and 4.
To my surprise, 98% of the work go to the proxy or the direct pool, 1% each to my reserve pools.
I have seen that after a DSL reconnect the proxy sometimes is a little picky, then after a few minutes all work resumes to it.

I have the same probs, ckolivas told me it's from stratum use.
You will have to use non-stratum pools (or disable stratum with the option, see help) or reduce engine clock.

Do you have higher difficulty?
If you use stratum protocol and your workers give too much hashes, the pool increases difficulty for you, so you have less shares but get paid better per share.
CGM shows that in the shares line.

Yes, that is right, but it is hard for me to recognize if 2857rpms are ok or not. All my 6 units have slightly different rpms for the given percent.
But: the regulation uses temperature to calculate the needed percentage.
So as long as temperature is in ok ranges, fan percentage is lower then the set max, and when i see 52% i know all is well, at 55% i know something gets too warm, either i have to check fans or switch my ac on.
But fan rpm is needed also, so my request was to display fan percentage additionally.

That is quite interesting, i must admit.
My GPUs are mostly at 99% load, i have the CCC open all the time, but must admit i don't look all the time.
Is it that the load drops sometimes for a short period and that makes the GPU cool or recover so it can look like lasting longer, and on stratum it is on 99% load 24h a day?

Also, i have a request:
Is it possible to show the fan percentage with the rpm's? I use CCC to monitor fan percentage, because that is my measure when and if  my computers need to be air condititoned.
As long as fan percentage is below my set max of 55% i declare them sufficient and do not reduce engine clock. (But on the other hand it seems i need to to have stratum work?)

I was using 2.7.6 on my dual 7970 (Club 3D and Sapphire with AMD cooling) rigs for as long as it was out now, with 1125MHz core & 1000 mem, no voltage change, power setting at -20%, all Win7 32, 1250Mh per rig.
I have noticed all 2.8 builds have the problem of declaring the GPUs sick with that clock rate.
They work with the stock 925 and a little bit more, but above 1000 i get sick fast.
I have not modded the cards in any way, just plugged them in, installed latest AMD driver and started CGM.
It (or the driver?) itself set the clock up to 1125, the maximum, more is not allowed by the bios.
I have to manually reduce engine clock to below 1000 to have it run a few hours, but then it inevitably dies with GPU declared sick, and a windows pops up "CGMiner.exe has stopped working".
What is going on here?

I better stick to 2.7.6 and use the proxy, that does work for days now.

Can anyone tell me if there are any certificates with the gold?
Not that it prevents the manufacturer/seller from sending tungsten bars with gold coat.
But it makes selling the gold here easier, afaik.

I did not mean to offend you, and also i did not mean "you" are anon, please accept my apologies.
BTC itself is anon and can not be tracked to any individual.
Also, you are the first to respond.
Did you have problems on the website lately?

My 7970s run at 78°C with 50-55% fan, -20% Power Setting, 640Mh per piece with 1125 (C3D) or 1025 (Sap) MHz.
My 6950 has a new 120mm fan and has 72°C with 100% fan (1500rpm), 320Mh with 840MHz at -20% Power.
All C3D and Sapphire with AMD design.

All in standard cases, but i have no sidewall mounted. Also, one 120mm fan blows cold air on the cards for my twin 7 machines.

Just FYI:

At the moment i am trading ~2kW electric Power for ~4GHash/s = 1,3 BTC/Day.
I have:
1 Rig with 2 Club3D 7970 = 1350Mh for 520W
2 Rigs with 2 Sapphire 7970 = 1220Mh for 520W/each
1 Rig with 1 Sapphire 6950 = 320Mh for 215W

Just calculate if your electricity costs are below the ~15USD/10Eur i earn per day. For 20ct per kWh it does never pay.

I also tried to get info from coinabul, bitcoincommodities and btcgold.
None of them answered my question.
I personally believe they just keep the BTC, because they are anon and can't be refunded.
I would really like to switch to real hard investments, but it's not easy to get money from BTC here in Germany...

Alternatives, please?

@Rogue:
- You need powerfuls GPU, AMD HD6 or HD7 preferred.
- You need a mining software, i personally use CGMiner, but there are others, also dependent of which OS you use.
- You should use a pool, BTCGuild is my favorite. But depending on your location, other pools may be better because of lower load, better speed. No need to push all the data around the globe...
- You need time, electricity and a lot of GPUs.

One single GPU can just mine for fun. It will never ever be profitable, at least if you didn't steal them and (!) have to pay for electricity.

That were my thoughts also.
The reward halves in a few weeks, and also all the BFLs will go online, pushing the difficulty.
Mining with GPUs will get unprofitable by then, at least if you don't steal the electricity and hardware.
But BTCs earned and kept up to that time: What will happen?
- Reward halving will drop the price for a given work, so BTC -> USD will likely also drop, your coins will lose value.
- Diff rise will increase the work for a coin, so it gets harder to get some. BTC -> USD will likely rise.

The question is: Which effect is greater?
I have seen "early place in shipment of some BFL" offers for weird amounts of BTC.
If you had the luck to get one a few days earlier then others, the effect will be noticeable, i think.
But the network quickly adapts and keeps the block gen time at 10mins by rising diff.
So when all pools are flooded by BFL, we will have to see what happens...

I have made some 400 USD not thinking about the electricity and hardware, some by speculation on exchanges.
Now i for myself split my earnings, half are sold/used, the rest is kept to sell after the rewad drop & diff increase.

Hello,
i just registered to participate on the discussion on cgminer.
I have some miners with dual 7970s and some single 6950s running, earned a bit with them. 
I now want to use stratum, doesn't work as expected.

I hope to be a non noob soon, so i can be a valued member of this community!