ASIC power consumption estimates - page 4.

bobitza

sr. member

Activity: 560

Merit: 256

Quote from: fizzisist on September 30, 2012, 06:19:18 PM

Quote from: Stephen Gornick on September 30, 2012, 06:01:30 PM

Quote from: jamesg on September 30, 2012, 11:23:57 AM

One of the biggest costs in running mining equipment is going to be power consumption.

So ya, the cost of electricity is becoming much less of a factor than it used to be.

I disagree. Cost of electricity is always the number one concern when mining because you are essentially turning electricity into bitcoins. Cost of equipment is number two, and only matters when you are buying equipment or considering selling it.

I agree with the disagreement

In the long-run, the cost of the electricity is THE factor that will decide who keeps his/her ASICs plugged in and who will turn them off. The race to bring the most Gh/s/W is on and the winner(s) should sell the most ASICs in the market (all other things being equal).

Once you buy the equipment, the money paid should be considered sunken costs unless you plan to resale the equipment in which case you need anticipate resale value and/or time timeframe needed to recoup the investment.

To quote Obsi:

Quote

With the reward halving and coming ASIC price wars we might see first gen ASIC miners with high electricity rates priced out of the market sometime in 2013.

squid

member

Activity: 112

Merit: 10

Quote from: SkRRJyTC on October 01, 2012, 09:13:34 AM

Quote from: Inaba on October 01, 2012, 09:07:28 AM

You (and Tom) are dreaming if you think it's going to run at 60w. The reason he doesn't want to post his power specs is because he knows they are going to be high.

Why is it attainable for your company but only dream-able for others?

My guess is avalon is the only reasonable prediction and once we get actual prototypes out we will see this

SkRRJyTC

legendary

Activity: 1008

Merit: 1000

Quote from: Inaba on October 01, 2012, 09:07:28 AM

You (and Tom) are dreaming if you think it's going to run at 60w. The reason he doesn't want to post his power specs is because he knows they are going to be high.

Why is it attainable for your company but only dream-able for others?

Inaba

legendary

Activity: 1260

Merit: 1000

You (and Tom) are dreaming if you think it's going to run at 60w. The reason he doesn't want to post his power specs is because he knows they are going to be high.

freeAgent

sr. member

Activity: 240

Merit: 250

Quote from: Mushroomized on September 30, 2012, 11:34:55 AM

Having already ordered a jalapeno, I'd say im pretty happy that the power usage will be so low. If I can run this straight from the usb ports on my raspberry pi, I will be very happy.

If these generate a considerable amount of heat, I might even use it to warm my little sisters hermit crab tank, and heaters like that use around the same amount of power.

I just wonder when the new "little SC's" will be available, I might purchase one if I can somehow use my jalapeno order towards the price.
Its cool to see all these asics popping up, I wonder what kind of miners will come up next? Huge supercooled quantum miners?

There's no chance you'll be able to run the Jalapeno directly off the USB ports on the Raspberry Pi. You'll have to use a USB hub. http://raspberrypi.stackexchange.com/questions/340/how-much-power-can-be-provided-through-usb

kano

legendary

Activity: 4592

Merit: 1851

Linux since 1997 RedHat 4

Actually - up front the power performance really doesn't matter at all - it's only in the long term if difficulty rises orders of magnitude faster than the BTC price rises.

Consider today - I can run 560W of mining hardware for about $80 a month in an expensive energy price country (Aus)
My 560W produces almost 1BTC per day - but with $12 BTC that's approximately $300 - $80 = $220 profit a month and the cost is about 27%

If I switched to a 54GH/s bASIC miner for $1069.99 from http://www.BitcoinASIC.com and it was using (as a random number guess) 60W, then that would be at the moment around $8.80 a month and produce approximately $6,931.83 profit a month - so only 0.13% energy cost

Now when the difficulty goes up 10x, if the BTC price stays the same at $12 (and doesn't go up) that will of course be 1.3% - still seems small ...
Double that power and you get: 2.6% ... still seems small ...
20x difficulty and double power: 5.1% ... still not a big deal yet ...

By the time it matters, anyone who cares about that number may well be using the next generation of ASIC devices ...

... though I don't think anyone has proven any power 'facts' yet ...

Edit: Oh I should add:
20x difficulty, double power, half block reward, still $12: 10.2% - I could deal with that

squid

member

Activity: 112

Merit: 10

Quote from: jamesg on September 30, 2012, 07:55:18 PM

Quote from: squid on September 30, 2012, 06:39:08 PM

Considering gigavps has a vendetta against Tom and btcfpga (and obvious bias towards BFL since that what he invested in) expected him to he hesitant in reporting numbers that are comparable to BFL.

Hi squid,

I have put this thread together to help consolidate information based on provable facts. I am trying to be impartial and objective. If you feel Tom's statement cited earlier in the thread is clear and concise, I will add it to the OP.

Best,
gigavps

Nah, Tom has stated that he will give an estimate once he gets the prototype in his hands.

Bogart

legendary

Activity: 966

Merit: 1000

Quote from: Jutarul on September 30, 2012, 06:44:55 PM

It just amazes me to see BFL claiming to have reached 1J/GH. If it were April 1st I would be assured it's a joke. I simply don't understand how they could push that technology so hard in such a limited amount of time, and competitors don't get even close to it.

http://www.theonion.com/articles/fuck-everything-were-doing-five-blades,11056/

Jutarul

donator

Activity: 994

Merit: 1000

Quote from: mrb on September 30, 2012, 07:47:47 PM

Why can't you understand? Simple math shows that 950-1500 Mhash/J should be possible with a 65nm ASIC: https://bitcointalksearch.org/topic/best-demonstrated-efficiency-1290-mhashjoule-95762

Thanks for the link. Is there any indication that BFL is doing 65nm? AFAIK the lower you go the higher the risk that the chip is a total failure, and you have to go back to the drawing board. Small problems can be fixed after the chip is produced, but that also requires some engineering time.

jamesg

vip

Activity: 1358

Merit: 1000

AKA: gigavps

Quote from: squid on September 30, 2012, 06:39:08 PM

Considering gigavps has a vendetta against Tom and btcfpga (and obvious bias towards BFL since that what he invested in) expected him to he hesitant in reporting numbers that are comparable to BFL.

Hi squid,

I have put this thread together to help consolidate information based on provable facts. I am trying to be impartial and objective. If you feel Tom's statement cited earlier in the thread is clear and concise, I will add it to the OP.

Best,
gigavps

mrb

legendary

Activity: 1512

Merit: 1028

Quote from: Jutarul on September 30, 2012, 06:44:55 PM

Quote from: squid on September 30, 2012, 06:39:08 PM

Considering gigavps has a vendetta against Tom and btcfpga (and obvious bias towards BFL since that what he invested in) expected him to he hesitant in reporting numbers that are comparable to BFL.

It just amazes me to see BFL claiming to have reached 1J/GH. If it were April 1st I would be assured it's a joke. I simply don't understand how they could push that technology so hard in such a limited amount of time, and competitors don't get even close to it.

Why can't you understand? Simple math shows that 950-1500 Mhash/J should be possible with a 65nm ASIC: https://bitcointalksearch.org/topic/best-demonstrated-efficiency-1290-mhashjoule-95762

Jutarul

donator

Activity: 994

Merit: 1000

Quote from: squid on September 30, 2012, 06:39:08 PM

Considering gigavps has a vendetta against Tom and btcfpga (and obvious bias towards BFL since that what he invested in) expected him to he hesitant in reporting numbers that are comparable to BFL.

It just amazes me to see BFL claiming to have reached 1J/GH. If it were April 1st I would be assured it's a joke. I simply don't understand how they could push that technology so hard in such a limited amount of time, and competitors don't get even close to it.

squid

member

Activity: 112

Merit: 10

Quote from: mrb on September 30, 2012, 04:42:11 PM

Quote from: jamesg on September 30, 2012, 04:36:05 PM

Quote from: dust on September 30, 2012, 01:04:49 PM

Cablepair stated the bASIC would consume less power than the mod miner quad (40 watts), but this was back when it was a 27Gh/s unit. It is safe to assume the 54GH/s unit would required twice the power and consume less than 80 watts.

80/54 = 1.48 J/Gh

https://bitcointalksearch.org/topic/m.1183153

Hi Dust,

I think is is fair to say that it is not clear what Tom is saying there. I would rather wait for him to announce the info and then we can add it to the list.

Best,
gigavps

Tom said exactly in the above comment "[the 27Gh/s device] uses less electricity than the ModMiner Quad", so it is pretty clear to me that it means less than 40Watt. Hence 675 Mhash/Joule, or 1.48 J/Ghash.

Now if you wanted to be pointy, one could say that Tom did not promise the 54Gh/s device would have the same efficiency as the 27Gh/s one...

Considering gigavps has a vendetta against Tom and btcfpga (and obvious bias towards BFL since that what he invested in) expected him to he hesitant in reporting numbers that are comparable to BFL.

Inaba

legendary

Activity: 1260

Merit: 1000

I have to agree with fizzisist, ultimately it's going to eventually end up at the cost to run whatever mining hardware you have. The initial costs will be paid back at some point, so if a unit costs 2x as much to run, the value of that equipment is 1/2 comparable equipment that is 2x as power efficient, even if the initial costs are the same or even higher.

But it's even more counter-intuitive than that - your ROI is going to be cut by a factor equivalent to the difference in power consumption over the term of the ROI... I think I phrased that right. So if your unit is 2x a power efficient, that also reduces your ROI time factor, even if you disregard your profit after the ROI has been met.

It's all about the power efficiency in the end.

fizzisist

hero member

Activity: 720

Merit: 525

Quote from: Stephen Gornick on September 30, 2012, 06:01:30 PM

Quote from: jamesg on September 30, 2012, 11:23:57 AM

One of the biggest costs in running mining equipment is going to be power consumption.

So ya, the cost of electricity is becoming much less of a factor than it used to be.

I disagree. Cost of electricity is always the number one concern when mining because you are essentially turning electricity into bitcoins. Cost of equipment is number two, and only matters when you are buying equipment or considering selling it. Since equipment isn't 100% liquid, it's the electricity cost, difficulty, and exchange rate that you are watching on a day to day basis, i.e. "should I mine today or turn it off." Furthermore, if your equipment is more efficient than everyone else's (or your electricity is cheaper) you will be always be able to keep mining, in theory at least.

Stephen Gornick

legendary

Activity: 2506

Merit: 1010

Quote from: jamesg on September 30, 2012, 11:23:57 AM

One of the biggest costs in running mining equipment is going to be power consumption.

A 1,500W BFL Mini Rig SC costs $5.40 per day using the U.S. average residential rate of electricity around $0.15 per kWh.

So that is under $2K USD per year in electricity. For an investment of $30K, $2K going to electricity isn't a particularly huge issue, though definitely operating one where electricity costs are $0.03 per kWh (like it is near Hydro electric generation) is going to be more profitable than running it where electricity costs are above average.

This compares to GPUs whose capital costs were low compared to the cost of the electricity to run them.

For comparison, a $1,100 GPU rig consuming 700W has electrical costs of $2.52 (using U.S. residential average of $0.15 per kWh) and $919 per year.

Of course, figure in the cost of capital per hash and this difference is even more pronounced. Capital costs: GPU rig: $785 per Ghash/s. ASIC rig: $30 per Ghash/s.

So ya, the cost of electricity is becoming much less of a factor than it used to be.

jamesg

vip

Activity: 1358

Merit: 1000

AKA: gigavps

Quote from: mrb on September 30, 2012, 04:42:11 PM

Tom said exactly in the above comment "[the 27Gh/s device] uses less electricity than the ModMiner Quad", so it is pretty clear to me that it means less than 40Watt. Hence 675 Mhash/Joule, or 1.48 J/Ghash.

Now if you wanted to be pointy, one could say that Tom did not promise the 54Gh/s device would have the same efficiency as the 27Gh/s one...

I would like to try and make sure that I am not putting specs down that the manufacturers have not released. While you are quoting him, he could have meant that statement in multitude of ways.

For instance, did he mean total power consumption or Mh/s per watt?

mrb

legendary

Activity: 1512

Merit: 1028

Quote from: jamesg on September 30, 2012, 04:36:05 PM

Quote from: dust on September 30, 2012, 01:04:49 PM

Cablepair stated the bASIC would consume less power than the mod miner quad (40 watts), but this was back when it was a 27Gh/s unit. It is safe to assume the 54GH/s unit would required twice the power and consume less than 80 watts.

80/54 = 1.48 J/Gh

https://bitcointalksearch.org/topic/m.1183153

Hi Dust,

I think is is fair to say that it is not clear what Tom is saying there. I would rather wait for him to announce the info and then we can add it to the list.

Best,
gigavps

Tom said exactly in the above comment "[the 27Gh/s device] uses less electricity than the ModMiner Quad", so it is pretty clear to me that it means less than 40Watt. Hence 675 Mhash/Joule, or 1.48 J/Ghash.

Now if you wanted to be pointy, one could say that Tom did not promise the 54Gh/s device would have the same efficiency as the 27Gh/s one...

jamesg

vip

Activity: 1358

Merit: 1000

AKA: gigavps

Quote from: dust on September 30, 2012, 01:04:49 PM

Cablepair stated the bASIC would consume less power than the mod miner quad (40 watts), but this was back when it was a 27Gh/s unit. It is safe to assume the 54GH/s unit would required twice the power and consume less than 80 watts.

80/54 = 1.48 J/Gh

https://bitcointalksearch.org/topic/m.1183153

Hi Dust,

I think is is fair to say that it is not clear what Tom is saying there. I would rather wait for him to announce the info and then we can add it to the list.

Best,
gigavps

Bogart

legendary

Activity: 966

Merit: 1000

Quote from: Mushroomized on September 30, 2012, 01:47:28 PM

Quote from: dust on September 30, 2012, 01:14:46 PM

Quote from: Mushroomized on September 30, 2012, 01:12:36 PM

Quote from: Meatball on September 30, 2012, 01:09:46 PM

Quote from: Mushroomized on September 30, 2012, 11:34:55 AM

Having already ordered a jalapeno, I'd say im pretty happy that the power usage will be so low. If I can run this straight from the usb ports on my raspberry pi, I will be very happy.

If these generate a considerable amount of heat, I might even use it to warm my little sisters hermit crab tank, and heaters like that use around the same amount of power.

Most heat generated in electric appliances, electronics and computer gear is caused by inefficient use of electricity. For example a 100 Watt bulb is drawing 100 watts, but not all 100 watts is going into the creation of light, a lot of it is wasted in heat, and that's why the bulbs get so hot.

I'm guessing the ASIC's must be pretty efficient, but even if you look at the worst case of 100% heat loss, you're only look at a 4.5 watt or 60 watt heater. I'd be shocked if even the SC puts off more heat than a single GPU, and even then, that's a stretch.

Either way I don't need that much heat! Although I forget that it wont give off 100% heat, haha

It will give off 100% of the energy it uses as heat.

It might have a light on it!

And/or fan(s)

Topic: ASIC power consumption estimates - page 4. (Read 15449 times)