difficulty went down

[chiropteran]

Is this summer heat shutting some miners down, or is this a result of dumb miners wasting work without the current version making invalid blocks?  A little of both, I guess.

Or is there another explanation?

[Stephen Gornick]

Well, the prior adjustment period had an increase of 8.5% so this drop of 3% doesn't even reverse half of the previous rise.
 - http://bit.ly/HCbFOp (difficulty history spreadsheet)

There definitely was a fair amount of wasted hashes on blocks that became orphaned but there's been enough time for miners to have updated and the hashing level hasn't yet turned back upward.
 - http://bitcoin.sipa.be/speed-lin-2k.png

There are miners rotating out of GPU mining and delving into FPGA mining, so during that transition there is a period where some are doing no mining, particularly if they need the proceeds of the hardware sales to fund the purchase of the FPGAs.  Even then, many end up on a waiting list for the FGPAs ordered, in the meantime.

Here's one miner that is making that switch:
 - http://bitcointalk.org/index.php?topic=75616.0

[freshzive]

I'll take it!

[Shadow383]

The large botnet that was mining zero-transaction blocks appears to have stopped, at least for the time being, but the increase won't be fully reversed due to the increase in FPGA mining and the greater hashrates of the 7xxx series ATI cards.
If BFL for example were to clear their backlog of orders we'd probably see 2000000+ difficulty in no time at all.

[rograz]

It somewhat get's mitigated by ppl dropping out tho, at least here in Europe a lot of ppl are very close to where it isn't worth the hassle when running old 58xx rigs around 2mh/w (unless you undervolt) and paying 0.15$+ per kwh.

[Raize]

Well, the prior adjustment period had an increase of 8.5% so this drop of 3% doesn't even reverse half of the previous rise.
 - http://bit.ly/HCbFOp (difficulty history spreadsheet)

There definitely was a fair amount of wasted hashes on blocks that became orphaned but there's been enough time for miners to have updated and the hashing level hasn't yet turned back upward.
 - http://bitcoin.sipa.be/speed-lin-2k.png

There are miners rotating out of GPU mining and delving into FPGA mining, so during that transition there is a period where some are doing no mining, particularly if they need the proceeds of the hardware sales to fund the purchase of the FPGAs.  Even then, many end up on a waiting list for the FGPAs ordered, in the meantime.

Here's one miner that is making that switch:
 - http://bitcointalk.org/index.php?topic=75616.0

Stephen's right. He even correctly pointed out mrb who was known to have 15 gh/s as of last year:
https://bitcointalk.org/index.php?topic=6169.msg91235#msg91235

I'm pretty sure most of the more establish miners are moving to FPGAs either now or soon. I'm sure they are looking at this summer and cooling costs and focusing more on getting a better hash/joule right now. I'm actually looking at doing the same, but I'm still (mostly anyway) accumulating coin instead of selling.

[Meni Rosenfeld]

Even if the total network hashrate remains fixed, difficulty adjustments of ~5% are to be expected simply due to randomness.

[Tittiez]

Even if the total network hashrate remains fixed, difficulty adjustments of ~5% are to be expected simply due to randomness.

At the moment the next re-target is expecting -13.73%, doesn't look much like randomness.

[the joint]

Even if the total network hashrate remains fixed, difficulty adjustments of ~5% are to be expected simply due to randomness.

At the moment the next re-target is expecting -13.73%, doesn't look much like randomness.

That's likely because we're only ~19% through this difficulty period.

Imagine if the first block after a difficulty change took 1 minute to solve -- then, you might 'expect' a -90% difficulty adjustment the next time.  But, if the 2nd block takes 59 minutes, then you'd be looking at a +600% difficulty adjustment.

Yes, there is going to be some variance in mining power over a 2 week period.  But, there is also going to be variance in luck.  13.73% variance over a period of a 2-3 days can be expected.

[antares]

Imagine if the first block after a difficulty change took 1 minute to solve -- then, you might 'expect' a -90% difficulty adjustment the next time.  But, if the 2nd block takes 59 minutes, then you'd be looking at a +600% difficulty adjustment.

max readjustment is 400% up and down.
however, we are currently seeing a drop in hashing power of about 20-30%. this isnt normal fluctuation. I'd be rather interested in what is the real cause for this.

[the joint]

Imagine if the first block after a difficulty change took 1 minute to solve -- then, you might 'expect' a -90% difficulty adjustment the next time.  But, if the 2nd block takes 59 minutes, then you'd be looking at a +600% difficulty adjustment.

max readjustment is 400% up and down.
however, we are currently seeing a drop in hashing power of about 20-30%. this isnt normal fluctuation. I'd be rather interested in what is the real cause for this.

Ah, thanks for reminding me about max adjustment.

One possibility though is summer heat.

I'm getting about 10-15% fewer m/hash come to think of it because I've had to downclock.

[Littleshop]

Imagine if the first block after a difficulty change took 1 minute to solve -- then, you might 'expect' a -90% difficulty adjustment the next time.  But, if the 2nd block takes 59 minutes, then you'd be looking at a +600% difficulty adjustment.

max readjustment is 400% up and down.
however, we are currently seeing a drop in hashing power of about 20-30%. this isnt normal fluctuation. I'd be rather interested in what is the real cause for this.

Ah, thanks for reminding me about max adjustment.

One possibility though is summer heat.

I'm getting about 10-15% fewer m/hash come to think of it because I've had to downclock.

Ditto.  I switched (just in time) most of my rigs to summer mode.  I do about 20% less.  I am underclocked and undervolted now. 

[DeathAndTaxes]

Ditto.  I switched (just in time) most of my rigs to summer mode.  I do about 20% less.  I am underclocked and undervolted now. 

Same here although the watercooled ones are running at 15% higher clock.  Just need to get all the rigs converted.

[TheHarbinger]

I had to downclock about a month and a half ago.      I fucking hate Florida.

[disclaimer201]

I had to downclock about a month and a half ago.      I fucking hate Florida.

If you're in Florida, you definetely need FPGAs. How do you even deal with all those power outages over there? This must be killing rigs, too.

Greetings from a much too cold spring in Central Europe! Temps are still higher 40s during the day and 36 F right now (midnight). 

[the joint]

Speaking of heat and cooling, my current cooling setup is....odd.

Apparently, my rig is cooler with my desktop-case closed rather than leaving the sides of the case open (thanks to the CoolerMaster fans I'm guessing).  Interestingly, an open case with a desk fan blowing directly on the cards is hotter.

Even more interestingly, taking a desk fan and blowing it through one of the CoolerMaster case fans (fan squared) is the coolest.

Go figure.

[disclaimer201]

Speaking of heat and cooling, my current cooling setup is....odd.

Apparently, my rig is cooler with my desktop-case closed rather than leaving the sides of the case open (thanks to the CoolerMaster fans I'm guessing).  Interestingly, an open case with a desk fan blowing directly on the cards is hotter.

Even more interestingly, taking a desk fan and blowing it through one of the CoolerMaster case fans (fan squared) is the coolest.

Go figure.

Not so odd at all. It's all about airflow and effective heat dissipation. If you leave out the sides, you ruin the airflow inside your case. Adding more fans, even a deskfan, won't help the heat to leave the case. Worst case you will just mix the hot air and a lot of it'll stay in the case. This, and the fact that open cases tend to accumulate much more dust and will build up additional heat made me go back to closed cases and air filters.

[the joint]

Speaking of heat and cooling, my current cooling setup is....odd.

Apparently, my rig is cooler with my desktop-case closed rather than leaving the sides of the case open (thanks to the CoolerMaster fans I'm guessing).  Interestingly, an open case with a desk fan blowing directly on the cards is hotter.

Even more interestingly, taking a desk fan and blowing it through one of the CoolerMaster case fans (fan squared) is the coolest.

Go figure.

Not so odd at all. It's all about airflow and effective heat dissipation. If you leave out the sides, you ruin the airflow inside your case. Adding more fans, even a deskfan, won't help the heat to leave the case. Worst case you will just mix the hot air and a lot of it'll stay in the case. This, and the fact that open cases tend to accumulate much more dust and will build up additional heat made me go back to closed cases and air filters.

Word.  CoolerMaster rocks.

[Raize]

It's also worth noting that Microsoft and some antivirus vendors took down a botnet that was supposedly mining on the Bitcoin network recently too, at one point it was getting between 15-20% of the mined coin by some estimations. I don't think it was recent enough to account for the present drop that looks to be a pretty steep one, but like others have said, give it another thousand blocks before trying to really estimate it.