Slush's pool, for one, will get these errors up to 20 times per hour. But this has only been going on for maybe 3-4 days at most.

My miners at BTCG experienced far less, and mostly sporadic rpc errors.

Sapphire Trixx software can natively tweak the voltage. It was just recently updated to support the xtreme cards.
Download link is below.

http://www.sapphiretech.com/ssc/TriXX/

It can also set fan profile curves.

I have 3 of them in my farm, mostly using 6xxx series cards.

I run them at a 10% overclock with no added voltage.

The VRM is significantly stressed in an exponential curve the more voltage you add.
The incurred heat will be unsustainable in the long term (6+ months of 24/7 usage) if using clocks such as 1000mhz at the maximum voltage needed for stability. They will statistically start failing faster as you add more and more mV.

Lowering memory clocks is one way to burden the card less, but is not a free pass for using massive (20-30%) overclocks.
But those doing this at their own risk and claiming there is no danger should re-think whether stability over time is more important than relatively miniscule gains in mhash/s (for example, I am not willing to risk 25% larger risk of card fatality for 40mhash/s extra).

Same problem as above. Server issues roughly every few minutes since the last couple of days.

Doesn't happen on other pools. Also recieving 'problems with bitcoin rpc' about 20 times per hour, hash calculation will drop to zero and then pick up again.

There is zero difference between the brands. They are all the same chips from AMD with reference coolers and clocks.

I would recommend XFX on the basis of lifetime warranty even if the card is overclocked or using the BIOS switch.

Original pioneers and participants do not represent the general population in any way.

I am willing to bet the forum has a large amount of the (hypothetical) 0.1% of the most honest people anywhere. They are overrepresented as the project is still quite small.

Selfishness, ignorance and hostility is the norm.

You miss the point. He doesn't lose "everything" even if the value of bitcoins crashes to a cent and the currency becomes outlawed on all 5 continents under the penalty of death. He can just suspend the electricity, internet access etc.. Until it becomes profitable again (either for himself or his clients).

Even if it doesn't, he has physical hardware that can either

1. be easily sold for around 60-80% of it's original value depending on how soon this hypothetical massive crash and outlawing would occur (yes you can argue this is a loss, but depreciation of 20% is better than 99.999% depreciation through direct purchase of BTC)

2. used for any other number crunching/cryptographic scheme that requires mass computational power.

Mining bitcoins is a much more economically sound strategy than large direct, speculative purchases.

To answer the person above: Roughly 1 hour should do it, the confirmed reward should then be transferred to your BTC wallet.

To slush: Can this suggestion (by melco on deepbit.net thread) have any merit on your pool?


Would be actually quite useful, though I'm not sure how easy it would be to implement.

Also, server is constantly complaining about "problem with bitcoin rpc" every few minutes for the last few days.

IMO 5850's are the best reference card, costing slightly over $100 dollars and yielding from 280 to 330 mhash/s.
In comparison a $500-$600 dual GPU card fetching merely ~650-700mhash/s at stock speeds seems like a bad hardware investment.

The only problem is availability, you would be hard pressed to find many in new or mint condition. However there are plenty of used ones on the open market.