Analysis of ASIC earnings, device agnostic

KS

sr. member

Activity: 448

Merit: 250

Quote from: 600watt on May 14, 2013, 07:49:05 AM

Quote from: KS on May 14, 2013, 07:36:27 AM

Quote from: 600watt on May 14, 2013, 05:03:38 AM

if one orders 10 000 avalon chips today and they deliver as they say within 12 weeks
they should be shipped around mid august.

let´s say getting them mounted on boards and running will take 3 to 4 weeks = mid september

that will give you 2800 GH/s with an investment of...(based on current btc rate)

(90 k $ for the chips, + 70 k $ for assembly =) 160 k $

this is massive investment in the best available mining hardware. with a difficulty of 80 mio it would generate 530 btc per month. so it would need about 3 months to pay back.

send cash, i order ! Grin

Time to sell you house

my wife wont let me, but i am working on it Wink

Sell your wife AND the house!

Problem solved Grin

600watt

legendary

Activity: 2338

Merit: 2106

Quote from: KS on May 14, 2013, 07:36:27 AM

Quote from: 600watt on May 14, 2013, 05:03:38 AM

if one orders 10 000 avalon chips today and they deliver as they say within 12 weeks
they should be shipped around mid august.

let´s say getting them mounted on boards and running will take 3 to 4 weeks = mid september

that will give you 2800 GH/s with an investment of...(based on current btc rate)

(90 k $ for the chips, + 70 k $ for assembly =) 160 k $

this is massive investment in the best available mining hardware. with a difficulty of 80 mio it would generate 530 btc per month. so it would need about 3 months to pay back.

send cash, i order ! Grin

Time to sell you house

my wife wont let me, but i am working on it Wink

KS

sr. member

Activity: 448

Merit: 250

Quote from: 600watt on May 14, 2013, 05:03:38 AM

if one orders 10 000 avalon chips today and they deliver as they say within 12 weeks
they should be shipped around mid august.

let´s say getting them mounted on boards and running will take 3 to 4 weeks = mid september

that will give you 2800 GH/s with an investment of...(based on current btc rate)

(90 k $ for the chips, + 70 k $ for assembly =) 160 k $

this is massive investment in the best available mining hardware. with a difficulty of 80 mio it would generate 530 btc per month. so it would need about 3 months to pay back.

send cash, i order ! Grin

Time to sell you house

600watt

legendary

Activity: 2338

Merit: 2106

if one orders 10 000 avalon chips today and they deliver as they say within 12 weeks
they should be shipped around mid august.

let´s say getting them mounted on boards and running will take 3 to 4 weeks = mid september

that will give you 2800 GH/s with an investment of...(based on current btc rate)

(90 k $ for the chips, + 70 k $ for assembly =) 160 k $

this is massive investment in the best available mining hardware. with a difficulty of 80 mio it would generate 530 btc per month. so it would need about 3 months to pay back.

send cash, i order ! Grin

ecliptic

sr. member

Activity: 322

Merit: 250

Quote from: organofcorti on April 23, 2013, 10:45:22 PM

Quote from: chipd on April 23, 2013, 02:15:59 PM

Not sure if already discussed (too lazy to read through the whole thread right now, sorry).
How about GPU miners that might move away when profit is close to 0 for them?

edit:
Nice work, an interesting read!

All I can say is that some point we'll lose GPUs and at another we'll lose FPGAs. Many GPUs will probably leave early to mine LTC, FPGAs will mine on until the bitter end. I'm assuming that by the time this happens, 24Thps will be a very small fraction of the network, and probably won't have much of an impact on the forecast accuracy given all the other sources of error (shipping times being the largest source of error).

Litecoin has been consistently ~1.3-1.5x more profitable than BTC for GPUs for months now. I expect most GPUs have already moved.

(Indeed, a large % of LTC GPU miners were even switching between alt coins because they were more profitable still, but nowhere near as stable as LTC)

KS

sr. member

Activity: 448

Merit: 250

Quote from: organofcorti on May 06, 2013, 10:03:32 PM

Quote from: GodHatesFigs on May 06, 2013, 08:35:10 PM

I can't make heads or tails of those graphs - can you explain what they show please?

Sorry I'm at work right now and don't have time, but if you click on the chart the blog post will open which explains it all in detail.

Somehow, I don't think it will clarify things

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: GodHatesFigs on May 06, 2013, 08:35:10 PM

I can't make heads or tails of those graphs - can you explain what they show please?

Sorry I'm at work right now and don't have time, but if you click on the chart the blog post will open which explains it all in detail.

GodHatesFigs

full member

Activity: 557

Merit: 101

I can't make heads or tails of those graphs - can you explain what they show please?

KS

sr. member

Activity: 448

Merit: 250

Right, a projection on how fast they need to retire their GPUs.

I don't know how anyone could do that right now, given we have no clue how many ASICs are being put online and how many more are to come and when.

I should go and see a psychic, then I will let you know

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: KS on May 06, 2013, 06:26:28 AM

I would look at BTC/GH/s per $ invested.

As your projections show, BTC/GH/s is falling rapidly (nice blog BTW) but when you factor in CAPEX, the numbers look more mundane.

Now, to know whether you'll be making money in $ terms is another issue (as always, it depends on the market). My guesstimate is that the BTC should hover in the 50-70 USD range when all the crazyness has dissipated but things are very much in flux right now so I wouldn't bet on it Wink

At 50 USD/BTC, you're looking at a profit of circa 20 USD/day in August for 50GH/s (0.008 BTC/GH/s per day), not factoring power consumption. If you paid 50K for your ASIC (and got it in August), you just lost a ton of money. If you paid 2.5K, you'll get the same ROI as ppl used to get on GPUs when spending the same 2.5K on a PC.

I would plug in some CAPEX numbers in the tables/graphs to have a guesstimate as to how much one can invest and when, if one wants to make money (but it looks like you're doing it much better already :p).

I've included exchange rate in previous posts in the series, and produced tile plots of exchange rate / ROI with days to ROI as the tile colour. It's interesting, but at the moment everyone seems to want their estimates based in BTC, and no one seemed to know how to interpret the charts.

I agree it's an important factor, but back in January I'd only included exchange rates up to $40 and even then people though I was being optimistic. Point is, BTC market is too volatile to use an estimate - you have to produce a large number of estimates for a very wide range. If readers want that back, I'll bring it back next time.

KS

sr. member

Activity: 448

Merit: 250

I would look at BTC/GH/s per $ invested.

As your projections show, BTC/GH/s is falling rapidly (nice blog BTW) but when you factor in CAPEX, the numbers look more mundane.

Now, to know whether you'll be making money in $ terms is another issue (as always, it depends on the market). My guesstimate is that the BTC should hover in the 50-70 USD range when all the crazyness has dissipated but things are very much in flux right now so I wouldn't bet on it Wink

At 50 USD/BTC, you're looking at a profit of circa 20 USD/day in August for 50GH/s (0.008 BTC/GH/s per day), not factoring power consumption. If you paid 50K for your ASIC (and got it in August), you just lost a ton of money. If you paid 2.5K, you'll get the same ROI as ppl used to get on GPUs when spending the same 2.5K on a PC.

I would plug in some CAPEX numbers in the tables/graphs to have a guesstimate as to how much one can invest and when, if one wants to make money (but it looks like you're doing it much better already :p).

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: KS on May 06, 2013, 05:08:11 AM

@organofcorti: I was referring to "your" numbers table.

Long story short: I think ASIC mining will be where GPU mining now is, profit-wise, after the "ASIC-wall" hits.

The GH/s power required will simply have multiplied by a factor more or less equivalent to the price reduction in equipment of similar GH/s power. So, you will still invest the same amount for the same or slightly lower BTC return but you will also have cut your electricity costs.

The money-making window is NOW for those lucky few with a shiny new ASIC and it will close when the masses can finally get their hands on them too. Your bet is as good as mine as to when that will happen (given BFL should have shipped in 2012 already...).

Hope this clarifies the previous post a bit.

Ignoring electricity costs (which hopefully wont be an issue for a while unless we hit the petahash mark soon), is this what you mean:

KS

sr. member

Activity: 448

Merit: 250

@organofcorti: I was referring to "your" numbers table.

Long story short: I think ASIC mining will be where GPU mining now is, profit-wise, after the "ASIC-wall" hits.

The GH/s power required will simply have multiplied by a factor more or less equivalent to the price reduction in equipment of similar GH/s power. So, you will still invest the same amount for the same or slightly lower BTC return but you will also have cut your electricity costs.

The money-making window is NOW for those lucky few with a shiny new ASIC and it will close when the masses can finally get their hands on them too. Your bet is as good as mine as to when that will happen (given BFL should have shipped in 2012 already...).

Hope this clarifies the previous post a bit.

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: KS on May 05, 2013, 06:09:09 PM

meh... if you look at capital expenditure (CAPEX) and operating expenditure (OPEX), ASICs are way better than GPUs.

For the same 2500USD spent, you'd now get a PC with 4x 7970 GPUs, giving you about 2.8GH/s for a total of 0.11 BTC/day (according to the table above, as of May 11th) and using up about 1400-1500W of energy.

Compare that to a 2500USD 50GH/s machine in August that will give you 0.40 BTC/day at 100W.

Even at 10.000USD per 50GH/s, serious miners won't loose money with respect to the current situation (if the table is correct)...

Sooo, my humble guess is... not much will happen.

I'm not following you. What post are you referring to?

KS

sr. member

Activity: 448

Merit: 250

meh... if you look at capital expenditure (CAPEX) and operating expenditure (OPEX), ASICs are way better than GPUs.

For the same 2500USD spent, you'd now get a PC with 4x 7970 GPUs, giving you about 2.8GH/s for a total of 0.11 BTC/day (according to the table above, as of May 11th) and using up about 1400-1500W of energy.

Compare that to a 2500USD 50GH/s machine in August that will give you 0.40 BTC/day at 100W.

Even at 10.000USD per 50GH/s, serious miners won't loose money with respect to the current situation (if the table is correct)...

Sooo, my humble guess is... not much will happen.

Signus

newbie

Activity: 28

Merit: 0

I will do that, thanks.

At least an ASIC will pay itself off at a relatively decent rate given that most people will receive their ASIC devices once the difficulty is 45-55mil.

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: Signus on April 24, 2013, 01:33:40 AM

Quote from: organofcorti on April 23, 2013, 10:52:43 PM

Quote from: Signus on April 23, 2013, 10:32:59 PM

Quote from: railadner on April 23, 2013, 12:54:17 PM

With increasing difficulty and reducing miners, which the price of 1 BTC in August? $ 500 +?

The change in the price depends on the interest and the trading whereas the ability to mine profitably depends on the difficulty, market, electricity, etc.

Just so.

Signus' statement implies that difficulty is affected by the USDBTC price and not visa versa. I have a number of posts on this subject, and a weekly forecast of the network hashrate and difficulty that does not rely on ASIC shipping predictions, just previous hashrates and exchange rates. It is surprisingly (to me) accurate.

I said the ability to mine depends on the difficulty and many other factors.

I'm not saying the difficulty is necessarily affected unless the market implodes the market like it has done with ASICs. Essentially the price spike got people interested in bitcoin, and then because of that people are buying large amounts of mining power that escalates the difficulty rise. It's a cycle.

OK, well I'm going to have to disagree with you then. Even before the price spike there's been quite a nice relationship between the exchange rate and network hashrate. Check the posts I linked to.

Difficulty / network hashrate is predominantly affected by the exchange rate. If the things that the money buys are reduced or increased in cost, then then these external influences change the relationship away from expected.

Signus

newbie

Activity: 28

Merit: 0

Quote from: organofcorti on April 23, 2013, 10:52:43 PM

Quote from: Signus on April 23, 2013, 10:32:59 PM

Quote from: railadner on April 23, 2013, 12:54:17 PM

With increasing difficulty and reducing miners, which the price of 1 BTC in August? $ 500 +?

The change in the price depends on the interest and the trading whereas the ability to mine profitably depends on the difficulty, market, electricity, etc.

Just so.

Signus' statement implies that difficulty is affected by the USDBTC price and not visa versa. I have a number of posts on this subject, and a weekly forecast of the network hashrate and difficulty that does not rely on ASIC shipping predictions, just previous hashrates and exchange rates. It is surprisingly (to me) accurate.

I said the ability to mine depends on the difficulty and many other factors.

I'm not saying the difficulty is necessarily affected unless the market implodes the market like it has done with ASICs. Essentially the price spike got people interested in bitcoin, and then because of that people are buying large amounts of mining power that escalates the difficulty rise. It's a cycle.

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: Signus on April 23, 2013, 10:32:59 PM

Quote from: railadner on April 23, 2013, 12:54:17 PM

With increasing difficulty and reducing miners, which the price of 1 BTC in August? $ 500 +?

The change in the price depends on the interest and the trading whereas the ability to mine profitably depends on the difficulty, market, electricity, etc.

Just so.

Signus' statement implies that difficulty is affected by the USDBTC price and not visa versa. I have a number of posts on this subject, and a weekly forecast of the network hashrate and difficulty that does not rely on ASIC shipping predictions, just previous hashrates and exchange rates. It is surprisingly (to me) accurate.

organofcorti

donator

Activity: 2058

Merit: 1007

Poor impulse control.

Quote from: chipd on April 23, 2013, 02:15:59 PM

Not sure if already discussed (too lazy to read through the whole thread right now, sorry).
How about GPU miners that might move away when profit is close to 0 for them?

edit:
Nice work, an interesting read!

Good question. I'm not including GPU miners leaving, since of the ~ 24Thps that were present prior to ASICs I don't know how much was attributable to GPUs and how much to FPGAs. So I don't know how much the hashrate would drop once GPUs are unprofitable. Another thing is that I've very carefully avoided guessing what the exchange rate will be, which is needed to determine GPU profitability.

All I can say is that some point we'll lose GPUs and at another we'll lose FPGAs. Many GPUs will probably leave early to mine LTC, FPGAs will mine on until the bitter end. I'm assuming that by the time this happens, 24Thps will be a very small fraction of the network, and probably won't have much of an impact on the forecast accuracy given all the other sources of error (shipping times being the largest source of error).

Topic: Analysis of ASIC earnings, device agnostic (Read 9831 times)