Topic: ASIC mining -- is my math right? (Read 5437 times)

I don't get your numbers with a profitability decline of 0.61, but I do get them with 1.0 (no decline). Are you sure on the numbers you posted?

This looks like a good fundamental analysis.

We (many of us) might be being optimistic about the likely increase in difficulty through 2013.
Conversely, we might be pessimistically not considering that BTC price might skyrocket to $50+/BTC in 2013.

Our mileage will vary.

My numbers assume

10x difficulty + 0.5 reward = same as 20x difficulty

+

Moore's Law profitability decline per year:  of 0.61

Your calculations only use a step increase in hash rate. That might work well for a per day calculation right after you get it, but you need to factor in an increase in hashrate throughout the year for per annum income. For instance, if when you get your Minirig the network hashrate is 10x what it is now you would earn $110k in the first year if difficulty stays constant, $79k if it increases linearly over the year from 10x to 20x, $59k if it increases to 40x, etc.

Most likely, if you're looking at buying a BFL product now you won't get it until at least the new year. By that time we can probably expect about 250TH/s from BFL, another 54TH/s from bASIC and possibly ~40TH/s from Avalon to be on the network, as well maybe another 10TH/s that's FPGA and can still mine profitably at high difficulty for awhile. Starting with 350TH/s and assuming the network grows to 1PH/s over the next year, at $12/BTC you would make about $40k from a Minirig SC.

This was my math (back when BTC=$12.53)

Difficulty will become 10x (just add a zero to the end of the current difficulty)
Coin reward in December 2012 will half to 25 coins (effectively doubling difficulty)
So projected profits (after power costs @ 0.15c KW/h and 1 BTC=$12.53 and ignoring equipment cost) are...

Jalapeno (4,500MH/[email protected]) = $0.91pd, $27.30pm, $332.15pa

Little Single (30,000MH/s@30W) = $6.08pd, $182.40pm, $2,219.20pa

Single SC (60,000MH/s@60W) = $12.16pd, $364.80pm, $4,438.40pa

Mini Rig SC (1,500,000MH/s@1500W) = $304.04pd, $9,121.20pm, $110,974.60pa

You can calculate it yourself here: http://bitcoinx.com/profit/index.php

LOL! true story... 

Not as outlandish as all the Illuminati/Rothschild/Bilderberg/jewish/reptillian conspiracy theories that seem to be heavily coupled with the bitcoin community.

After the dust settles I think I will go the bASIC route.. but if BFL improves its shipping times and that fun stuff I may go that way too... but then the Avalon sounds just as great.. so I am all over the place lol

Agreed - I've pre-ordered some BFL - but also going to get some bASIC (since the bigger ones are actually cheaper) just to try them out. I'm not sold on the idea of a bareboard without power supply or enclosure, but we'll just have to see.

I need to find out where he's buying these $5-$6 BTC.

Didn't you know that Satoshi Nakamoto is an alien? 

I like to think of them as these fellows from Dark City:

Who the hell are the "BTC network hashrate managers"? Are they the ones making crop circles??

I agree with dirtycat's position. After ASICs are out the majority of GPU miners are going to quit mining. That means, the only hashpower will be coming from the second gen machines. No matter how many of those are there, they probably will fail to match (instantly) the current network hashpower. Have you ever wondered why they're not shipping single units right now and they wait until they have a mass shipment out?

That's simply because they want the current network hashrate (http://blockchain.info/charts/hash-rate) to be taken over without drops. That means the current 22500GH/s is matched with (at least) 375 BFL's SC singles (or equivalent). They (both ASICs manufacturers & the BTC network hashrate managers) want to make sure the transition will go smoothly.

Now: Just imagine your current 1BTC/day setup that settled you back $2000 to buy, suddenly produces 1/10th the BTCs you make now; won't you just quit? I'd easily say ''yes''; but there's a case I won't. That case will be *IF* I'd still (somehow) manage to make a fraction of my current production at a cost that pays my electricity bills; or even better somehow maintain my current BTC/day production. If MANY people quit GPU mining, I might as well (it's a faint scenario) continue to mine with my 5970s for as long as it's fair for me...

All in all: I strongly believe that the transition from GPUs to ASICs will not occur instantly. There will be a fair amount of time that miners will be on the verge "to ASIC or not to ASIC". If the BTC production for them make a quick payoff, I personally will quit mining and (for a couple of weeks) will only buy cheap bitcoins (a rough estimation will be about $5 to $6 each). Then I'll use my freshly aquired BTCs to get myself a couple of ASICs...

Cheers.

Cablepair has stated that the bASIC products will be competitive with any other ASIC, not the same or better. He also said he doesn't expect BFL to meet their 1GH/J spec. If I had to guess, I'd say that the bASIC units might be in the range of 100W for the 54GH/s unit.

It's also BFL that has (by far) the most finished product and best packaging... as well as the best specs. (only bASIC has better mh/s/$ numbers and it's very slight - and overshadowed by the fact that you don't get power supplies or an enclosure included in the price).

bASIC might still be able to pull ahead... but the only thing I can think of that would do it, is much lower power consumption, and we won't have hard numbers on that until both products are loose in the wild.

Of all the companies expected to deliver by year end it's BFL with both the first expected delivery date and by far the largest quantity of units to deliver(minimum 3x). It's also BFL that is least likely to meet their time table given their past record. Maybe that means someone else ships right around the same time as BFL, and maybe that means a smaller manufacturer gives their customers a huge advantage by beating BFL's delayed delivery. Such customers would quickly meet ROI with a relatively small number of ASICs out in the wild.

I've said from the beginning that ASIC producers will sale their machines at a price to which people will buy. If the bitcoin/usd price doesnt keep up with the difficulty, ALL manufacturers will drop their prices accordingly. Also, I think those of us that ordered in the first month was a bad idea. BFL has already stated that they want all pre-orders caught up by the years end and by then you will be able to order and have it shipped to you the next day. So, what advantage do all the early purchases have?

I have invested a considerable sum, but I wont purchase any more until I see what happens. I gambled at the beginning and maybe it will pay off and maybe not.. But I surely think its not a wise move to make ANY investment into an asic until you have a good idea what the difficulty is going to be and how fast its rising..

I think we have all vastly underestimated BFL and the others pre-orders, why don't they tell us how many terahashes or petahashes they have sold? because no one would shell out even 1200 bucks for a sc single if there is already a petahash pre-ordered. I think a lot of people are ordering while looking at todays difficulty.. sure, 1.5ths would make a pretty penny right now, but in 2 months? it wont be looking so rosy..

I predict the sc mini rig will cost around 5k within 6 months or they make them 10x faster for 30k My bet is that they will just make them faster...(unless the bitcoin/usd price stays ahead of the difficulty).. I sure as hell wont want to invest 30k into a machine that may never recoup its original investment....

Possibly.

Also, ASICMINER will be coming out in the same timeframe as BFL/bASIC, and the Reclaimer project possibly in the spring. There's going to be a big push in the next couple months, likely well past 10x by the end of January. I'm not sure how fast it will keep up with that pace though, alot will depend on how manufacturers price their units one the orders start slowing down.

I assume there were supposed to be T's in front of those final numbers, not G's

BFL did say $250k, but that was just the BTC orders (not wire transfers) just in the first day. They will have significantly more now. Last count for bASIC is 600@54GH/s and 50@27GH/s. Avalon has sold out two preorders of 300 pieces apiece at 60GH/s.

BFL - Unknown, but almost certainly over 100TH/s
bASIC - 34TH/s
Avalon - 36TH/s

here's the deal I am not trying to step all over your math it looks right and I can agree with it.. however.. there is no account for other companies producing or private groups along with how many will be produced.. on top of that there are no known products out now that show us what they can do.. we can take their word for it if you like but I wanna know where this is going to go.  Until these devices are out in the wild hashing away we don't know anything.

you said:


is it 5 or 6 or 7.2- 8.6...

until they release an actual working product your math is based on hypothesized numbers

your math is not including other companies producing along with who knows how many private groups

there is no telling how long roi will be

No, the manufacturers have released real, actual numbers.  BFL said "we have $250,000 in pre-orders" and bASIC said "we have around 400 pre-orders."  Didn't I just say that?

at 40x with a single - you're look at 7.2 - 8.6 months - ish.  =P

You wont know those numbers are real until the manufacturers actually release the devices.
You don't know how many of these are being manufactured/sold.

without those two pieces of information you wont be able to give a definitive answer to anything related to difficulty/profit without using "it could be" "it should be" "it might be" "could be around" "should be around".

as I stated I love the optimism I would hope to get roi at 5-6 months (again not a exact number) with a sc single at 40x

Nope, they were real numbers released by the ASIC manufacturers (well the 2nd time)

well shit who am I to question a ti84.. I'm on board

It's calculus - you get a nice derived function that you can increment difficulty in and get a lovely graph showing you at what difficulty which hash rates become unprofitable.

My ti-84 is still useful =P

thanks its a cute kitty I stole it off google images lol

Your doing math on hypothesized numbers.  Again love the optimism.

Once all ASIC vendors release their products, it's going to be a hell of a lot more than a 10x difficulty increase.

Well, SOMEONE missed my post where I did do all the math and it's not optimistic at all.  Then another person did the math and got the same result.  Then I redid it and got the same result.  Then someone else did it and got the same result.

My optimism is backed up by having actually done the math. You should try it - it's very reassuring.
The only real variable we can't account for is total hash on the network, but not know what it will be - doesn't stop you from running numbers at different rates (adjust difficulty in your maths).

Ahhhh! *rips hair out* It doesn't just matter how much hardware you ordered or how fast it is!  It matters how much hardware you ordered and how fast it is compared to how much everyone else ordered and how fast everyone else is.  Everyone's like ooooh 300GH/s.  I can run my own pool let alone solo mine!  But guess what, other people ordered 2 TH/s worth.  It's not just the impressive speed numbers that are important.


I love your kitteh avatar! lol

I love your optimism.

the zero commissions is reasonable depending on how much hardware you've already ordered. SC rig, yah 1.5th/s is plenty for solo mining you're talking 1.1% of the network hash rate at somewhere between 5x and 10 difficulty.

sc Single - no you won't be going solo.

I agree with your numbers, we'll see 5x immediately then a quick climb to 10x (as first wavers cash out and reinvest - yes expect a couple days of the btc/usd being compeltely tanked). and a quick step to 10x... then a slower acceleration to 20x or 30x over the next year or so.

I should point out that FPGAs will still be profitable enough at 10x - and will start dropping off when the network hits 15x.
At 40x - an sc single will still make about double what a single is making right now... nobody should be shocked by this... ROI will still be between 5 and 6 months.

Average daily commissions/donations: 0.00000000 BTC ($0.00) <-- this isn't correct.  Everyone thinks that they'll have the GH/s to solo mine.  Guess what, the problem is that EVERYONE thinks that, lol.  You'll probably end up on a pool.

Oh 10x isn't an extreme scenario. There's just pre orders confirmed right now between all the companies that would result in like 6x difficulty (I have re-run those with the new numbers to get a specific difficulty but that's in the ballpark).  Once pre-order people make money and buy more hardware plus other people see that the pre-orders were fulfilled and ASICs work so they start ordering (easily 2x to 3x the count of pre-orderers), that's the end of anyone ever making money with them.

I haven't heard of a trade in on the ASICs.  On the FPGA boards, the most expensive part is the FPGA chip itself, and that can be reused for different things.  Even if the board isn't useful, the chip can be removed and reballed.  An ASIC, however, can't be reused since it is only good for doing the one task, at best it could be resold to someone else interested in using it for bitcoin mining.

Granted it's a concern (endless upgrades) but as long as they keep the 100% trade-in policy. . . I'm not seeing a problem with it... at some point it will advance to the point where it can't continue per upgrade cost vs current state of the art.

 

I have read about the guarantees to not increase the price on the hardware but I have not read any guarantees to not LOWER the price either...
This and the talk of future improved versions cause me some concern. I don't wanna get trapped in a cycle of endless upgrading such as I am with my smart phone...

And what happens if BFL reduces its price by a factor 2x in December? They already increased the GH/$ by 50% without even shipping anything.


You seem to ignore or forget that when BFL and  other vendors reduce their price per GH, it directly affects your mining factor. And you can bet they will reduce the price every time  an equilibrium is approached, because an equilibrium would mean their sales are drying up.

With GPU's or FPGA's you cant keep dropping prices because these chips are expensive and there isnt much you can do about it unless you are AMD or Altera. These custom asics are a completely different kettle of fish; their marginal cost is close to zero, but their market value is extremely high (for now), and  determined solely by your mining factor; there is no demand for this chips besides mining, which isnt exactly true for FPGAs. And the price of these asics will determine the mining factor. Which will determine their market value.. see the loop? Its a race to the bottom, and it will only end once we approach marginal costs of these chips, or electricity cost becomes the determining cost of mining again. We are a very very long way from that today.

Why does that matter if you purchased with $?

The reason why people complain about electricity costs is because prior to these numbers from ASIC's being realized, electricity costs to run GPU's are pretty high.  A single GPU can suck up 300 Watts of power, and when you add in the other equipment it adds up quickly.  For example, with a 5850/6950, you're lucky to get around 400 MH for about 180-200 watts draw.  A 7970 is probably closer to about 250 watts for about 700 MH. 

Let's run with the 7970.  You'd need 42 of them to get the 30 GH in your ASIC example, or a power draw of 10500 watts.  At .10 a kWh, you're looking at about $25 a day in power costs compared to your 65 cents with the ASIC's.  And that's not counting all the extraneous power consumption you'd need for 7-8 mining PC's and cooling costs.

Still profitable?  Yes, but GPU power consumption can add up.

As we can see from the mining factor charts on bitcoinx.com/charts , the factor has stabilized at around 0.45 for the last year. That's 5-6 months ROI if you plug in the stats for a BFL single, at about $100-120/month mining profits each. Certainly this factor will drop off dramatically once the ASICs arrive. But despite this, I think we can safely say that this extremely consistent mining factor indicates a level of risk the mining community is willing to accept when purchasing new equipment, such that 5-6 month ROI's will remain constant even if some think this is still "insanely profitable". Miners aren't negligent of the fact that usd/btc has been rising, and will temper their mining purchases with btc holdings, waiting for the price to rise to another plateau (and with it, the mining factor), and THEN will buy more equipment, pushing the mining factor back down to equilibrium. The new ASICs are about 40x as efficient as last generation fpga's, so perhaps the new mining factor equilibrium will be 0.45/40=0.01125.

In short, I think pre-ordering ASICs was a bad move due to the usd/btc price runup, and that buying in the next month or so is best when shipping times are confirmed along with potentially faster shipping times from competitors.

I said marginal profit:
marginal profit is the term used to refer to the difference between the marginal cost and the marginal revenue for producing one additional unit of production.

The development cost of the asic is a sunk cost.Yes its big, even if Im pretty sure the preorders alone will have paid for it, but its irrelevant. If BFL no longer sells any volume because these rigs are no longer profitable to miners (ie, difficulty is too high) or because their competitors undercut them significantly,  they, as well as other ASIC suppliers will lower price as long as price is well above marginal cost. And its orders of magnitude above marginal cost right now. That they may or may not recover their sunk costs doesnt matter here. Deepbit will probably be the last to join, perhaps they will not recover their investments, but that doesnt mean they will keep prices where they are now. Its better to sell something at a "loss" as long as its well above marginal costs,  than sell almost nothing at all.

Sorry I never bet unless I'm sure to win 


Building an ASIC is really cheap, but the cost of the project is high (I actually know some people doing that kind of work and I've asked for a custom ASIC: too many hours of project are need to obtain something so is not profitable enough).
Selling them at the price you see around is not a 100000% margin, but probably something in the 20-25% if they sold hundereds of ASICs. Using and external fab is not so simple: logistic, cababilities of the farm and so on can reduce you profit to near 0 unless you sold 1000s of ASIC.


I know that there is 5 or 6 competitors in the arena and maybe another couple can add before the end of Q2, and I agree that is important to buy the ASIC now if you want to mine with them, but still remain on my own opinion that before the last Q of 2013 we don't reach a 50x in difficulty: the point is that a 20X the sc RIG has an hardware break even point of 228D (halving the reward and at change of 12.5$/Bitcoin). At 50x there is no convenience in buy it, more than 2 year only to repay the hardware (and no warranty on how long can  run an ASIC before fail) unless the price drops well below 10.000$ and/or bitcoin jumps over 35$  (still more than 6 month to repay the HW) and both the events have a low probability: the market is not so deep to sustain a 35$ bitcoin, and IMHO the price of the ASIC can halves but not more than this.

I will take bets on that.
Its rather simple, if we are over estimating the amount of ASICs preordered, then these devices will be insanely profitable when they do arrive, which will spur more sales until they no longer are. Once sales dry up, what do you think BFL will do? They will lower their prices and it starts all over. And again, and again.  IN the end asic's will be priced close to marginal cost, which at least for the chip itself, is negligible (literally a few dollar per chip). It will take a while to get there, mostly due to manufacturing delay, but if you are counting on that to guarantee your profitability, it seems like a very dangerous gamble. BFL is no longer a tiny startup, IIRC they are employing 22 people now and they have the funds to outsource anything they want. They would be crazy not to buy external manufacturing capability for a device with an initial marginal profit of something like 100000%.

Also keep in mind BFL isnt the only player, by early next year there will most likely be 5 asic vendors competing (BFL, bASIC, asicminer, Nzhang, Deepbit...).

Buying an asic seems like an excellent bet if you know you will get yours early so  you can recover your cost in at most a few months. Anything beyond that will look like a bloodbath IMO.

A 50x in difficulty means that network power will be of 1000THash, so more than 600 SC Rig. But all the preorders of BFL and Avalon are only in the 80THash range. Even if they sell 5 times that power in mini rig in few months we arrive at 20x in difficulty (and other 8 milion of $ invested in rigs). To reach your 100x with the SC mini rig we need an investment from miners of 40 milion of dollars. Even considering an halving of the prices we still need 20 milion $ (near 2$ for every bitcoin currently circulating).
I think that a 50x in difficulty is not probable before at least 8-10 months after the delivery of the first batch of ASIC.

I would say that is not an extreme scenario IF JUST BFL ONLY meet their target speeds of 1.5 TH/s in the SC mini rig. More like 50 to 100x difficulty could happen before things calm down.

Keep in mind you can't be certain WHEN you will receive your order so that could dampen your profitability time frame.

IMHO

Yeah Hash power should be 5x-15x for the next few months but If I was to predict long term hash rates (1-2 years maybe that's the medium term) I'd look at the MHs/$ rate.

With GPU it was arguably 1.7(my own rate) the ASICs are about 50
50/1.7 = ~x30

I think people will buy hardware until it stabilizes at that rate... maybe block halfing will keep it closer to half that (x15), but BTC price increases may bring it up...

Dollars in to dollars out I think the ROI will settle in at 10 +/-2 months so long as there are no new quantum leaps in technology again.
40% mining margin is palatable given the speculative nature of bitcoins...

It's a bit sobering... as much as I would love to get a loan for $30,000 and make ~$20,000 in revenue every month...

My most conservative estimates are 20x (and that's just based on BFL) it's looking more and more likely that by first quarter 2013, we'll see 40x difficulty - assuming that other producers ship on time.

Hrmm. I thought I saw a couple of threads where people predicted 5-10x based on present statistics for ASIC preorders, even if their efficiency is 40x last generation. Slow transition perhaps.

The efficiency is a couple orders of magnitudes greater with ASICs, not a single order of magnitude, so you should predict the network hash rate will increase 25- to 100-fold.

Hence average daily profit post-halving I would wager to be close to 0.521 to 2.084 BTC a day within a month of ASICs coming onto the scene.

I did factor in block halving and I increased difficulty ten fold.

These allows you to configure the reward:
https://bitclockers.com/calc
http://tpbitcalc.appspot.com/

You forgot to factor in block halving pretty soon and at a minimum, double the difficulty.

First time miner here, looking to get some BFL ASIC's soon and doing some calculations on expected returns.

Using: http://www.alcula.com/calculators/finance/bitcoin-mining/

Assuming: 5 BF SC Singles: ~$6k for 300 Gh/s
Power consumption: 1W/Gh @ 300Gh = 300W
Electricity cost: $0.10/kW

AND assuming price remains same (worst case scenario, as I'm pretty bullish), block halving, and 10x difficulty (again, extreme scenario):

Total Hash Rate:300000 MH/s
Average time to find one block: 4.75 days
Average daily revenue (pre-halving): 10.53539662 BTC ($130.98)
Average daily revenue (post-halving): ~5.26 BTC ($65.50)
Average daily commissions/donations: 0.00000000 BTC ($0.00)
Average daily electricity cost: $0.65 (0.05212355 BTC)

Average daily profit pre-halving: 10.48327307 BTC ($130.33)
Average daily profit post-halving: ~5.21 BTC ($64.85)

Are those electricity costs really that low? I always hear miners complaining about electricity costs but really these seem pretty negligible.