[Group Buy#1] Avalon ASICs CHIPS! Using JohnK as escrow! FINISHED!

[scotjam]

Sorry a bit on edge, coming back from 2 weeks of being too occupied to follow the market, things changed a lot.  I knew chips were late, but suddenly there's a massive influx of unanticipated hashrate, a shit ton more promised and enroute.

Trying to make sense of it all.

Even putting in 150m diff with 60% increase per month (30%/2 weeks seems to be the upper bound thus far) it still makes ROI in under 3 months.  I don't understand what calculations people are using that show negative ROI over FOREVER.

No problem - I think we're all a bit on edge. The current hashrate trends are insane (probably not sustainable long-term, but long enough to crush avalon chips I think).

Are you assuming delivery of assembled boards today? I would assume anything sooner than late September / early October is pretty optimistic at this stage. I would be overjoyed if I thought I could get assembled boards 2 weeks from now, but when you think that chips need to be dispatched (we still haven't received confirmation of tracking info - could be we get them without, but we shouldn't assume we will) ship from China to the US, be repackaged and distributed by ragin, sent to assemblers, assembled then sent to us.... I *really* can't see this happening in less than 3-4 weeks.

In terms of difficulty, mining.thegenesisblock.com shows a 76% monthly increase as the default value at the moment. If you think that's aggressive, look at their past performance table:

Historical Difficulty Increase
30 Day   110 %
60 Day   208 %
90 Day   441 %

Once the ROI starts getting worse each month (due to electricity costs) not better within a 12 month period, you can tell that ROI over forever is not that useful a metric unfortunately...

[1714078426]

1714078426

[1714078426]

1714078426

[1714078426]

1714078426

[1714078426]

1714078426

[scotjam]

Just to elaborate -

Here's an example of the kind of calculations I'm running (this is for a single board - you can multiply it up - same answer but more so):
http://mining.thegenesisblock.com/a/0d1bb81811
Hash power: 5.6gh/s
Power use: 15w
Cost of hardware: (we want to know whether it's worth a refund, so)
- chips: 0.078*16*$120 per btc (i.e. full refund in btc) -> $150
- assembly: c.$100 per board, 50% refund, plus $30 for shipping (full refund) -> $80
Total: $200

Even for overclocked boards (with burnins going for EUR 105):
http://mining.thegenesisblock.com/a/0af9373781
Hash power: 9
Power: 20w
Cost
-chips 0.078*16*120 per btc -> $234
-assembly (in usd) 140 -> $70 refund
total $300

In both cases you're in the red, and by the end of 12m it gets worse not better. So if you're "paying" your refund to get those boards, your ROI is negative, therefore it makes sense to take the refund (assuming you can get 50% refund from assemblers). That is, unless you believe the difficulty projections are bunk, or my other assumptions are BS (if so, please let me know, because I'm basing my decision on these!!)

This also assumes that there's no other delays, you boards aren't broken in the mail (some of burnins customers have had this already with batch 1 chips), no config problems etc.

Even if you get them a bit earlier, you're in the red deep enough that a refund always wins at this stage.

[ecliptic]

76%/month is not possible to sustain.

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

1 Million diff / 2 million diff still exceed electricity costs.  It's extremely doubtful any ASIC will ever fall below electricity cost unless BTC/$ plummets.  the $/hash is simply too massive, especially for anyone who doesn't directly buy chips from the fab themselves.

[wrenchmonkey]

76%/month is not possible to sustain.

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

1 Million diff / 2 million diff still exceed electricity costs.  It's extremely doubtful any ASIC will ever fall below electricity cost unless BTC/$ plummets.  the $/hash is simply too massive, especially for anyone who doesn't directly buy chips from the fab themselves.

Not possible to sustain, for HOW long? If your chips showed up assembled and hashing TODAY, you might break even and make a little money, but we're WELL past the anticipated delivery time with ZERO progress since the order date. What makes you think that the chips will deliver any time soon?

You guys are clinging to a false hope at this point. As others have pointed out and demonstrated, either way we're going to take a loss. You can either take a loss that also involved giving the assholes who screwed us a ton of money, and just bending over for them, or you can take the option of telling companies that this won't be tolerated, and work on finding another viable investment.

Honestly, anybody who chooses option A is a fool, and Yifu's bitch.

[scotjam]

76%/month i don't think is possible

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

That's a great link. It doesn't seem to factor in devices from any new / unknown manufacturers though. I'd assume there will be people developing their own equipment and keeping it to themselves as well.

It would be nice if it turned into a bit more of a sigmoid (i.e. leveled out) a bit before the end of next year, but at the moment it's still looking very exponential. It will sigmoid eventually, because at 28nm we're already nearing the edge of today's silicon tech (I think 22nm is best in production?) but the question is more what does Q4 2013 and full 2014 look like rather than what "eventually" looks like. I think we have pretty good visibility that it's going to get hairy in Q4 2013 because there are 3-4 major manufacturers pumping out kit in that time, and it shows no signs of abating.

Keen to see your results...

EDIT: keep moore's law in mind - computing curves are not limited to being sigmoids until you hit silicon's physical limits

[ecliptic]

76%/month is not possible to sustain.

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

1 Million diff / 2 million diff still exceed electricity costs.  It's extremely doubtful any ASIC will ever fall below electricity cost unless BTC/$ plummets.  the $/hash is simply too massive, especially for anyone who doesn't directly buy chips from the fab themselves.

Not possible to sustain, for HOW long? If your chips showed up assembled and hashing TODAY, you might break even and make a little money, but we're WELL past the anticipated delivery time with ZERO progress since the order date. What makes you think that the chips will deliver any time soon?

You guys are clinging to a false hope at this point. As others have pointed out and demonstrated, either way we're going to take a loss. You can either take a loss that also involved giving the assholes who screwed us a ton of money, and just bending over for them, or you can take the option of telling companies that this won't be tolerated, and work on finding another viable investment.

Honestly, anybody who chooses option A is a fool, and Yifu's bitch.

*WHAT* OTHER VIABLE INVESTMENT?

there is NOT A SINGLE ONE THAT LOOKS ANYWHERE NEAR AS GOOD AS THIS ONE FUCKING DID.  BEST CASE ROI WAS UNDER A MONTH

Obviously that was super-optimistic, but that was best case.  Expected case was supposed to be a few months for ROI.  Now it's never?

Run the numbers of anything else, your BEST CASE ROI is on the order of half a year,  what happens when THAT gets fucked over?  Do you think you won't get screwed over by another company, group, or person?

You run the numbers on a "refund", and stop doing shitty wrong calculations with never-ending hashrates being added (again, it's a fucking sigma curve not an exponential, all hardware is paid for.  Do you see people spending money on new hardware anymore?)

Something like 50,000,000$ was sent to BFL

Then 8,000,000$ was sent to Avalon chips.

The guys coming in later?  Less than that.

The money  invested is an inverse-sigmoid curve.  The hashrate is a sigmoid curve

[ecliptic]

76%/month i don't think is possible

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

That's a great link. It doesn't seem to factor in devices from any new / unknown manufacturers though. I'd assume there will be people developing their own equipment and keeping it to themselves as well.

It would be nice if it turned into a bit more of a sigmoid (i.e. leveled out) a bit before the end of next year, but at the moment it's still looking very exponential. It will sigmoid eventually, because at 28nm we're already nearing the edge of today's silicon tech (I think 22nm is best in production?) but the question is more what does Q4 2013 and full 2014 look like rather than what "eventually" looks like. I think we have pretty good visibility that it's going to get hairy in Q4 2013 because there are 3-4 major manufacturers pumping out kit in that time, and it shows no signs of abating.

Keen to see your results...

EDIT: keep moore's law in mind - computing curves are not limited to being sigmoids until you hit silicon's physical limits

moore's law applied to the $100,000,000,000+ industry of microelectronics and computers

The way I see it, very soon there is no financial incentive either to buy hardware or even NRE for hardware.  Fixed, sunk costs will be too absurd, who is going to buy $5,000 - $25,000 hardware (with an 8 figure NRE) that takes ~1-2 years to hit ROI even at a given constant hashrate?  Because that's the market for anything that delivers after 2013.

[bassclef]

Honestly, anybody who chooses option A is a fool, and Yifu's bitch.

Although I appreciate the wonderful sentiment, I'd prefer that we act like adults here.

I personally don't believe the recent hashrate spike is sustainable. I think it will level off and then progress linearly. A few months ago many predicted difficulty would be over 100 million in July based on exponential extrapolation alone. Soon after it leveled off. People speculate a lot here, and are usually wrong. I'm probably wrong too, but to have an x-petahash difficulty in December would require way too much capital and gives mining companies too much credit. Fact is we live in a cyclical, imperfect world and its not as easy as putting numbers into an equation. I recall reading somewhere that predicting difficulty with any accuracy is a fool's errand; I tend to agree.

Given the recent price bump and good press, it's possible the window may not have shut on profitability completely, but it's closing quickly. We need these chips very soon, or I'll be on the refund bandwagon too. (I'm not paying assembly costs so I'm willing to give it more time.) As others have pointed out it will be easy enough to buy cheaper chips elsewhere, however to take that additional gamble I need proof that Yifu is processing refunds quickly AND that prodigits/ragingazn will be on the ball to promptly refund everyone in the group buy the proper amounts.

We're not quite there.... yet.

[redphlegm]

I will defer my decision depending on what ryepdx wants to do. I think he has a significant amount of sunk cost in this.

[bigbeninlondon]

We also have no evidence yet of how quickly those refunds will be processed.  If we opt for refund, many will be left with no chips and no btc for who knows how long.

[ecliptic]

Sigh

So, burnin is offering 50% refund on his assembly? 

Or is that you can convert it to bitfury chips on a burnin PCB assembly?  or something?

Maybe a refund does make sense, even if the ROI between these and other things are a wash.  There's still a degree of unmitigated risk with chip delivery and whatnot, including in the group buy

[Vycid]

Honestly, anybody who chooses option A is a fool, and Yifu's bitch.

Although I appreciate the wonderful sentiment, I'd prefer that we act like adults here.

I personally don't believe the recent hashrate spike is sustainable. I think it will level off and then progress linearly. A few months ago many predicted difficulty would be over 100 million in July based on exponential extrapolation alone. Soon after it leveled off. People speculate a lot here, and are usually wrong. I'm probably wrong too, but to have an x-petahash difficulty in December would require way too much capital and gives mining companies too much credit. Fact is we live in a cyclical, imperfect world and its not as easy as putting numbers into an equation. I recall reading somewhere that predicting difficulty with any accuracy is a fool's errand; I tend to agree.

Given the recent price bump and good press, it's possible the window may not have shut on profitability completely, but it's closing quickly. We need these chips very soon, or I'll be on the refund bandwagon too. (I'm not paying assembly costs so I'm willing to give it more time.) As others have pointed out it will be easy enough to buy cheaper chips elsewhere, however to take that additional gamble I need proof that Yifu is processing refunds quickly AND that prodigits/ragingazn will be on the ball to promptly refund everyone in the group buy the proper amounts.

We're not quite there.... yet.

http://en.wikipedia.org/wiki/Overshoot_(signal)
http://en.wikipedia.org/wiki/Overshoot_(population)

I disagree. We are gonna hit breakeven for ASICs... and then go beyond,  as people keep their pre-ordered hardware running in a doomed attempt to recoup investment.

Also - I'd rather take 50% refund than struggle for no return. This requires effort and none of you guys are considering the cost of the labor of actually running the ASICs.

128 chips voting for refund. Apologies to those wanting to stick it out, but I at least have to try to cut my losses.

[ecliptic]

Has ragingazn said anything on this? 

Or has he still been pretty hard to get ahold of..

[bigbeninlondon]

I've only got 20 chips worth of say.  Do what you will.   

[scotjam]

As others have pointed out it will be easy enough to buy cheaper chips elsewhere, however to take that additional gamble I need proof that Yifu is processing refunds quickly AND that prodigits/ragingazn will be on the ball to promptly refund everyone in the group buy the proper amounts.

I believe the funds went through JohnK, so presumably all we would need to do is get an updated ownership spreadsheet together with refund addresses. prodigits/ragingazn would get their refunds from JohnK (who placed the order, presumably from the escrow address) just like everyone else, no?

[ecliptic]

Hmm.. that is true, refund would only need JohnK to act (with whatever critera, probably 51% of chips..)

[scotjam]

Hmm.. that is true, refund would only need JohnK to act (with whatever critera, probably 51% of chips..)

Yeah, probably. And it's probably possible to get refund requests from 51% of chip holders, because the chips are quite concentrated in this group buy (top 10 on the spreadsheet would be enough, and I know a lot of other people below those will be keen on a refund too).

Anyone keen to volunteer to update the spreadsheet for ownership, and then go through the refund requests that have already been posted in the thread?

[scotjam]

I have pm'd JohnK to see what is possible but he is moving this week and will probably only catch up with this thread and his PMs on his return (on Tuesday 3rd September)

[wrenchmonkey]

76%/month is not possible to sustain.

the absolute MAX is 6-7 petahash/sec at the end of the year

https://bitcointalk.org/index.php?topic=283820

What difficulty is that.. ~1 billion at the end of 2013 diff

2 billion at the end of 2014 diff.

Hash rate will be a sigmoid curve.  it's not exponential except for a short window.  I have not found a calculator that correctly does this, all do an exponential curve.

Let me try a few things

1 Million diff / 2 million diff still exceed electricity costs.  It's extremely doubtful any ASIC will ever fall below electricity cost unless BTC/$ plummets.  the $/hash is simply too massive, especially for anyone who doesn't directly buy chips from the fab themselves.

Not possible to sustain, for HOW long? If your chips showed up assembled and hashing TODAY, you might break even and make a little money, but we're WELL past the anticipated delivery time with ZERO progress since the order date. What makes you think that the chips will deliver any time soon?

You guys are clinging to a false hope at this point. As others have pointed out and demonstrated, either way we're going to take a loss. You can either take a loss that also involved giving the assholes who screwed us a ton of money, and just bending over for them, or you can take the option of telling companies that this won't be tolerated, and work on finding another viable investment.

Honestly, anybody who chooses option A is a fool, and Yifu's bitch.

*WHAT* OTHER VIABLE INVESTMENT?

there is NOT A SINGLE ONE THAT LOOKS ANYWHERE NEAR AS GOOD AS THIS ONE FUCKING DID.  BEST CASE ROI WAS UNDER A MONTH

Obviously that was super-optimistic, but that was best case.  Expected case was supposed to be a few months for ROI.  Now it's never?

Run the numbers of anything else, your BEST CASE ROI is on the order of half a year,  what happens when THAT gets fucked over?  Do you think you won't get screwed over by another company, group, or person?

You run the numbers on a "refund", and stop doing shitty wrong calculations with never-ending hashrates being added (again, it's a fucking sigma curve not an exponential, all hardware is paid for.  Do you see people spending money on new hardware anymore?)

Something like 50,000,000$ was sent to BFL

Then 8,000,000$ was sent to Avalon chips.

The guys coming in later?  Less than that.

The money  invested is an inverse-sigmoid curve.  The hashrate is a sigmoid curve

I'm not your investment advisor, or your broker. Figure it out yourself. If you're going to myopically focus ONLY on bitcoin mining equipment as be the only "viable investment" then I'm afraid you're in for a rough ride, friend. Some people have enough sense to be able to spot other viable investments, even if they don't look identical to what YOU apparently consider to be the only definition of a viable investment.

Putting it in a low-yield savings account, or spending it on hookers and blow would be more viable investments than this is, at this point. 

[ryepdx]

I will defer my decision depending on what ryepdx wants to do. I think he has a significant amount of sunk cost in this.

I appreciate your consideration. I won't be seeking a refund, though I realize that may simply be due to the amount I've invested into this emotionally. Please make your own decision based on what you think is best for you.

Also, for those of you going through my assembly service, I've just posted in our thread regarding refunds: https://bitcointalk.org/index.php?topic=203388.msg3047235#msg3047235

The TL;DR: Full refunds are out of the question, but I'll see what I can do in the way of partial refunds for those of you who want them.