ASICMINER: Entering the Future of ASIC Mining by Inventing It

[rudi]

In fact, AM is trying to change, and some changes have been made, AMHash is a clear signal.

In the future we will see AM will make more positive changes.

Can you already tell us what other changes are being thought of?

[Mabsark]

Thats not rocket science. AM is doing a pilot run or MPW since they expect test chips in december. BF and SP wont.

SP's RockerBox ASIC tape-out date = 13 May, 2014
SP post about receiving RockerBox ASICs = 23 July, 2014

That's the same length of time that AM will be waiting for their 28nm chips so what are you talking about?  

Also, that cryptonews article is a complete joke. Here's a sample from it:

I also see that your pricing is very competitive instead of being price so high like many of the Chinese manufacturers. How were you able to make the units so cost effective?‏

Guy:  We control the entire supply chain.‏
We manufacture in Israel and able to maintain quality. The mechanics are being produced in China and shipped by sea to reduce costs. In addition to ASIC and System design‏ our COO – Kobi Levin is an operational wizard with‏ a lot of experience in previous companies including Solard Edge.

SP20 = 1190 / 1700 = 0.7 USD/G
AM Prisma = (1.39 * 353) / 1400 = 491 / 1400 = 0.35 USD/G

An open and shut case of SP being price competitive with those "expensive" Chinese manufactures.  

Also from that article:

I greatly respect companies like yours that deliver on time‏ and do not do drawn out pre-orders that never deliver as Butterfly Labs and KNC.
‎
Guy:  We do take pre-orders for the 2nd gen, but we 100% we’ll deliver on time and on spec.‏
We are using the same engine, just doing process shrink.
We won’t take pre-orders for the 3rd gen‏ since its higher risk – a completely redesigned engine. We are taping out this week.‏
0.34 W/GHs at the ASIC level‏
0.43 W/GHs at the system level‏ both on the ASIC and on the System level‏
Very efficient DC2DC design‏ 28nm‏ not 20nm ‏
We’re manufacturing at Flextronics Israel‏.

Spondoolies SP30 was not on spec at all, what's Guy talking about? Also, their simulated specs for their gen 3 ASICS are on par with AMs gen 4, 0.34 J/Gh at the ASIC level. SP taped-out a month later though and will get their chips for testing a month later too.

[Blazed]

SP-T is indeed expensive, but AM has yet to make a dominating chip. Gen 1 was just the first en mass chip and they sold over priced gear and did well. Gen 2 failed, Gen 3 also..I do not have much confidence gen 4 will be all that given their history. The competition these days is really good...time will tell.

[hdbuck]

SP-T is indeed expensive, but AM has yet to make a dominating chip. Gen 1 was just the first en mass chip and they sold over priced gear and did well. Gen 2 failed, Gen 3 also..I do not have much confidence gen 4 will be all that given their history. The competition these days is really good...time will tell.

you can only hope they indeed learn from their mistakes.

[Blazed]

I hope they do...they are the only company selling reasonably priced gear. My guess is not a single SP miner has ever gotten a positive ROI. Bitmain is now charging silly prices also. So get it done AM!

Edit: everything BE chip based seems reasonably priced.

[RoadStress]

Also, that cryptonews article is a complete joke. Here's a sample from it:<snip>

SP20 = 1190 / 1700 = 0.7 USD/G
AM Prisma = (1.39 * 353) / 1400 = 491 / 1400 = 0.35 USD/G

Guy:  We do take pre-orders for the 2nd gen, but we 100% we’ll deliver on time and on spec.‏
Spondoolies SP30 was not on spec at all, what's Guy talking about? Also, their simulated specs for their gen 3 ASICS are on par with AMs gen 4, 0.34 J/Gh at the ASIC level. SP taped-out a month later though and will get their chips for testing a month later too.

Mabsark in your blinded AM shareholder view you aren't even paying attention to the article. While it says that it was updated in October the article was published a very long time ago, in May if I remember correctly. How can you compare the SP20 with the AM Prisma when Guy clearly says that they are taking pre-orders for 2nd gen!!! At that the the 0.69$/GH was the best price available. The fact that it didn't materialized it's another story, but it was the best price available at that time.

[Puppet]

SP's RockerBox ASIC tape-out date = 13 May, 2014
SP post about receiving RockerBox ASICs = 23 July, 2014

That's the same length of time that AM will be waiting for their 28nm chips so what are you talking about?  

Im talking about the fact that AM taped out in september, expects engineering samples in december and mass production nearer the end of Q105. The only explanation for that is that december is a MPW or a small test run. Thats not by definition a stupid idea, especially if you are not confident in your design, but it doesnt exactly help time to market and the competition dont seem to be bother with it. 

While we are it at, from mid december to end of february is barely enough time to just process a wafer, leaving no time for testing, if need be refining the design and do a respin. Which begs the question why do it at all. If the chip works as expected, you wasted 2-3 months. If it doesnt, you dont have time to fix it with that schedule.

[hdbuck]

SP's RockerBox ASIC tape-out date = 13 May, 2014
SP post about receiving RockerBox ASICs = 23 July, 2014

That's the same length of time that AM will be waiting for their 28nm chips so what are you talking about?  

Im talking about the fact that AM taped out in september, expects engineering samples in december and mass production nearer the end of Q105. The only explanation for that is that december is a MPW or a small test run. Thats not by definition a stupid idea, especially if you are not confident in your design, but it doesnt exactly help time to market and the competition dont seem to be bother with it.  

While we are it at, from mid december to end of february is barely enough time to just process a wafer, leaving no time for testing, if need be refining the design and do a respin. Which begs the question why do it at all. If the chip works as expected, you wasted 2-3 months. If it doesnt, you dont have time to fix it with that schedule.

obviously they are basically being very cautious now.

[RoadStress]

obviously they are basically being very cautious now.

Why would they be if they have identified the problem and fixed it?

[hdbuck]

obviously they are basically being very cautious now.

Why would they be if they have identified the problem and fixed it?

tss you can do better than this.

[freedomno1]

I actually meant December 16th + 14 days, the most important 2 weeks in AM history....... waiting for results from chip simulations. Sorry you were to slow to get that.......

That wasn't being slow mate 
That was just me thinking you were slower since two weeks from now is roughly November 16 

[Fordee]

I actually meant December 16th + 14 days, the most important 2 weeks in AM history....... waiting for results from chip simulations. Sorry you were to slow to get that.......

That wasn't being slow mate 
That was just me thinking you were slower since two weeks from now is roughly November 16 

In the ASIC industry, I think we all know how long 'Two Weeks' (tm) can be!  Lol!

[EternalWingsofGod]

An Asiminer wafer clock. You know you want one.

Interestingly enough I sort of would want one
Nothing like diversifying the product line a bit kind of like the original USB sticks being fashionable looking
(Perhaps not a clock)
But something cool looking is always nice.

[jjdub7]

So I'm practicing pulling data via php with the Havelock API and gretl.  I pulled the entire transaction history of AM1 trades on Havelock, including days divs were paid and teased out day-of-week and month variables and ran a price regression weighted on volume.  Dates ranged from 7/10/13 to the present.  BTC-USD and BTC network data from Quandl.

*** denote degrees of statistical significance on t-normalization.

Code:

Model 1: WLS, using observations 79-8426 (n = 8348)
Dependent variable: Price
Variable used as weight: Quantity
Omitted because all values were zero: DMonth_6 DMonth_5 DMonth_4 DMonth_3
Omitted due to exact collinearity: DDayofWeek_7 DMonth_12

                                        coefficient      std. error     t-ratio    p-value 
                                ----------------------------------------------------------------
  const                                 -64.0891       7.37374       -8.692   4.27e-18  ***
  Date                               -0.0200663    0.00281633    -7.125   1.13e-12  ***
  Time of Day                       3.51516e-06  2.03557e-06    1.727   0.0842    *
  Last Div Amt                         104.355        1.31024       79.65    0.0000    ***
  Days Since FC Post         -0.000763632  0.000133520   -5.719   1.11e-08  ***
  
  BTCPrice24hAvg$USD        -0.000432      2.06688e-05  -20.90    1.50e-94  ***
  DailyVolume BTC$USD     -4.32020e-07  2.54834e-08  -16.95    2.18e-63  ***
  ln(BTCNetworkHashrate)   -0.276515      0.0162034     -17.07    3.44e-64  ***
  AvgBlocksize_MB_             0.215881       0.0597392       3.614    0.0003    ***
  Network Revenue               3.85365e-08  4.20636e-09    9.161   6.37e-20  ***
       (daily, USD)
  Total BTC in Circul              5.93937e-06  6.56003e-07    9.054   1.70e-19  ***

  D_Sunday               0.00152049   0.00508091     0.2993  0.7648   
  D_Monday              -0.0274101    0.00459092    -5.971   2.46e-09  ***
  D_Tuesday             -0.0385198    0.00470833    -8.181   3.23e-16  ***
  D_Wednesday         -0.0382756    0.00441648    -8.667   5.31e-18  ***
  D_Thursday            -0.0228972    0.00470844    -4.863   1.18e-06  ***
  D_Friday                  0.00268289   0.00446526     0.6008  0.5480
   
  D_January                 0.232385    0.00761263    30.53    6.22e-194 ***
  D_February               0.235075      0.0119877     19.61    1.04e-83  ***
  D_July                      1.80466         0.0266744     67.65    0.0000    ***
  D_August                  1.33486         0.0209490     63.72    0.0000    ***
  D_September            0.424720     0.0155508       27.31   2.90e-157 ***
  D_October                0.0509887     0.0100955      5.051   4.50e-07  ***
  D_November             0.0939666    0.00638384    14.72    2.00e-48  ***

R-squared            0.978181

Granted, the dataset is still very limited for a thin market such as Havelock, but a few things to point out: 

- the variable "Days Since FCposts" is what is implies - the number of days since FC's last update to the forum.  On average, since AM shares have traded on Havelock, each day that FC does not post produces a ~76k satoshi drop in share price.  Granted, this is expected and the effect would become more drastic

- Total BTC in Circulation:  this makes sense...as the total number of BTc rises, there are more available to buy shares, leading to an increase in the share price ceteris paribus

- Weekends are a good time to sell, it seems.  Prices are higher, possibly because there are the most active users online/free at the same time over the weekend days (days' Correlation to measured relative to Saturday's contribution to predicted values).

- Since AM's capture of the high-water 30% network share, the price has decreased approximately 0.27 BTC per order of magnitude that the network hashrate has grown (that'd be your log effect)


Of course many of these variables are closely interrelated, but the Chinese exchange rate, the Chinese and American M0 supplies, and the LIBOR rate all produced negligible explanatory results when added to the model.  The Chinese exchange rate would, of course, be pegged to the USD, so perhaps this is not entirely unexpected.

Feel free to PM/reply with suggestions/comments on improving this and other models.  One useful variable I don't have is a time series of the shares outstanding on Havelock's exchange - this will definitely be useful in trying to explain, interpret, and rationalize buying and selling behavior over the past year.

[trek27]

A few words about AMhash (taken from AMhash thread):

As for AMhash. Buying it is similarly cost effective as buying Prismas - details varying on specific individual cases. It could give positive ROI (though probably not that much).

From a perspective of AM shareholder situation is a bit different because buying AMhash helps AM to resurface as successful operation. So, an AM shareholder buying AMhash while retaining realistic chance to earn a few coins effectively gives AM a hand. Frankly, the last conclusion increases my personal risk tolerance. 

[BitHub]

I would like to make a 1000 btc bet that asicminer wont be payin divs this year.

[Dexter770221]

I would like to make a 1000 btc bet that asicminer wont be payin divs this year.

I think they will. December for sure...

[cs54]

AMHash1 IPO is 18% filled with a week to go, and the topic thread is rather quiet: https://bitcointalk.org/index.php?topic=833704.60

It seems like an unsuccessful IPO is a real possibility, which would not only be a depressing outcome for the offering, but could also lead to a negative feedback loop of lost confidence in AM.

Let's face it -- IPOs are about intrigue.  If I were FC, I would declare a special dividend to AM shareholders as soon as possible, making clear that it's one time only. The amount isn't important -- but going through the motions of a dividend will turn some heads and be a great opportunity to put the spotlight on AMHash1.

I don't think it's unfair to say that lots of people have forgotten about AM and regard Havelock as a place where investments go to die. If AMHash1 is actually a quality offering then it should sell itself once people know about it. I think an AM1 special dividend is the way to get that done.

[Dexter770221]

AMHash1 IPO is 18% filled ....

I see ~3.7% (186650/5000000)

[hdbuck]

IPO is going down for the same reason every asset in bitcoinland is going down.. which is because bitcoin itself is crawling. no more momentum, enthusiasm or confidence caused by the previous bubble.