Same problem. Fix most appreciated!

Sending tip

Increasing core voltage helps to increase overclocking stability at the cost of increased heat production. Even with the same model, each GPU chip is different and has its own clock speed stability profile, so the fact that one out of your three is more touchy than the others is not to be unexpected. Since you can handle a few more degrees of thermal load before cooking your chips, I'd recommend boosting your voltage a bit and see if your temp stays below 85C. Let us know if your system likes is.

Checked the Gox graph lately? The last time BTC prices crashed this hard, Dwolla had just gotten snuffed out. Any idea what we're looking at this time? My guess is it has something to do with their policy change about account verification for fiat currency withdraws. Suppose it's going to jump back again? If so, it's a great time to swap your Gox dollars for BTC and sell them elsewhere if you don't just use them up instead. If it stabilizes at around $100, no big deal, still worth it for me. If it keeps dropping... Party's over! Everyone out of the pool. ASIC owners will enjoy their fancy bricks while GPU miners play games at max video settings.

Just had a thought about the bitcoin network's dynamics. The increasing diffuculty to match the total hash rate offers the bitcoin end user greater security through harder levels of encryption within the blockchain. But not really, since a 51% attack is still a growing possibility. Meanwhile, the fact that blocks come out at the same average rate means transaction won't reach confirmations any faster. That doesn't seem fair enough to the users of a growing network. If block reward quantities were frequently adjusted for total network rate, while the difficulty increased at a more steady and easy rate, we could still be cutting from the same size of pie, and block solution rate could actually represent our total hashing power. Perhaps that would be a problem with all the <10s blocks we'd be getting, though. Thoughts on the matter?

Not really a joke, but indeed laughable. $860 with tax and shipping. Nothing close to that is ever coming back to my pockets as a result. Was hoping to be able to satisfy a few of my friends' bitcoin appetite since sending our money to MtGox is not really an available option where I'm from. As was mentioned, it's an interesting hobby and it's been a fun ride. Besides, it's not like the GPUs are now useless. They can still be enjoyed for their designed purpose.

And it's predicted to happen in just 10 days. I got into BTC mining mid April of this year and bought new GPUs early May, hoping it wasn't too late to catch some profit. Absolutely horrific timing. With every passing day I am further assured of my stupidity. Soon, the electricity cost will outweigh the gain and I will have had enough of this shafting. At which point, gaming may resume.

Any other chumps in the same boat here?

I have 2 x 7970s with bridge connected. Since it's behind a cooling fan I added, I never pulled it out to try running without the bridge. With xfire enabled, card 2 never exceeds 89% GPU load and the hash rate suffers. With it disabled, both run independently at 98% GPU load.

I tweak my GPU voltage from 1175mV stock to just 1180mV to keep my overclock stable. The GPU clock speed I step up from 950MHz to 1125MHz. This boosts my hash rate from 560MH/s to 660MH/s. It could do more, but given my cooling system, I am getting GPU temps between 81C and 86C. Furthermore, the gain in hash rate per MHz drops at higher clock speeds. Can get 685MH/s at 1200MHz, but the extra 25MH/s is not worth the heat, so I don't push it past that. Using Afterburner, I turn my mem clock and mem voltage to minimum. This does not affect hash rate or stability, but reduces the heat output of the card, which allows to push the core a little further.

Hope you found this information relevant.
- Psyrick

10Mb per hour integrated over 24 hours. 100Gb per month is my limit.

Thank you for your reply. I was measuring total traffic over 24 hours. But it's a shared home network, so I could have had a number of interferences. I know it uses only kilobytes per share, but if you go through a thousand or so per hour, that becomes megabytes. Accumulate that over a month and it adds up to enough that it gets in the way of other uses I'd like to share my bandwidth to... Educational purposes, you understand.

I know this topic has been discussed before, but I would like some further clarification in discussion. I recently upgraded my hashing power from 160Mhash/s to 1.3Ghash/s. I found my network traffic was about 10Mb/h before, and since I will be hashing much faster now and I don't want to be consuming almost 10 times the bandwidth. Is it usefull to turn up my difficulty? Should I expect loss from turning my difficulty to something like 5? Is it right to assume that the result will then be that I submit shares 1/5 as fast but they are worth 5x as much while the work submitted still occupies the same number of bits as diff 1? I have tried running this difficulty for a few days now and noticed I still catch the >1 minute blocks. Haven't been carefully analyzing the bandwidth though.

I love your accent

Double quickies! Hope you got your shares in time.

I must have gone through a dozen of these. Different brands, different models. My conclusion: They're alwarys dead by the time you need them. Apparently we just haven't gotten that far along with battery technology to make a reliable backup unit. The only advancement they do is to make the alarm more annoying and more difficult to deactivate. Either that or the guys at the supply depot recognise my adress and they're like:
 "This guys again! Hey Joe, lets go find him another reject unit."

Weather cut the power on my machine while I'm at work. Happened yesterday aswell. Will have to wait until day's end to restart   

That's a relief. What a crying shame it would be to waste all that hash.

Rewards for the past 3 rounds are being recalculated right now.

Thank you for that suggestion. I was thinking it might be good to turn down the memory clock, but I wasn't sure if I'd lose stability due to the memory falling behind pace with the core. Also, I will consider the bench.

By the way. I really do appreciate the discussion I have received over this. I am very glad to be a member of this forum. The helpfulness of lack of hatefulness is very refreshing as compared to /b/.

Tuned down the voltage. Results are looking better. System is stable if I leave it alone to hash, but GPU crashes if I move the mouse too fast.
Vcc = 1188 mV    Core Clock = 1150 MHz    Mem Clock = 1515 MHz
GPU1 Temp = 88C      GPU2 Temp = 83C
Current Hash Rate = 672 Mhash/s
Default Hash Rate = 540 Mhash/s

2 x Powercolor Radeon 7970 mounted in a CoolerMaster HAF 925 case with extra 120mm fans inside. 2 for push-pull CPU radiator and 2 blowing on the video card stack from the side. Don't know if an Accelero cooler would perform a great deal better, since the stock ones seem pretty well designed. I may remount the stock ones though. I work at a precision optical manufacturing company and I can use the lapping wheel to grind flat the contact surface to within microns. Could even polish it to within a quarter of a red photon, but I think that would be waste of time, as the thermal paste layer is never close to that thin, and the mirror surface might actually do a good job of reflecting thermal infrared back at the chip. Does anyone here have any experience with diluting silicone based heatsink compound with silicone lubricant or other liquid to reduce viscosity and drastically thin-out the layer? I once did this with GPU heatsink on my Nvidia GTX 570. I got it so well contacted, that you could lift and swing the thing from the bond with no movement. Unfortunately, I forgot to reinstall the aluminum mounting frame before the heatsink, so I had to take it back apart, and the next time around I didn't dilute the compound. So no results to compare.

I appreciate your helpfulness. My cooling system is about as windy as I can make it, but I like to overclock to within the limits of it's heat transfer capacity. I'm trying to avoid going the liquid cooling route. I often do seize the chance for a cool down at the end of a long round, but when they peak up to 98C and there's no telling how much longer the round is going to be, I feel I don't have a choice. Only takes a couple minutes to get them back down to 40. After a cool-down period of my own, I find myself not really liking the waters in the Guild's pool, and as mentioned, most other pools either have the same scoring system and/or higher fees. While invalid blocks and scored out rewards give me the blues, I'm still jumping right back into Slush's pool.