Cryptocoin Mining?

[Brekyrself]

The latest episode of Let's Talk Bitcoin! #129 starting at ~12 minutes is very, very interesting. Looks like they recorded it a week or so ago as the wallet is now available for download.

http://letstalkbitcoin.com/blog/post/lets-talk-bitcoin-129-dogeparty-and-delegated-proof-of-stake

-Delegated POS
http://wiki.bitshares.org/index.php/DPOS

-Titan anonymous and stealth addresses. Send to a name instead of addresses while remaining anonymous.
http://wiki.bitshares.org/index.php/TITAN

-Market pegged trading. Predictions market trading which follows real world pricing. Think shorting BitUSD or going long on BitGold.
http://wiki.bitshares.org/index.php/Bitshares_X

 

[Mark R]

The alpha technology saga has taken an interesting twist.

They had previously given firm shipping dates in July, and had started taking final payments.

Turns out that they were just taking out, and the tape out failed. Lol.

PayPal then froze their account leaving them with no funding to revise the ASIC design and reattempt tape out.

They've managed to obtain some rescue funding, and are now preparing to tape out again. They now need to start work on the PCB for the miner which could not be built any earlier because the ASIC ballout isn't final until tape out.

Lots of angry customers as no refunds if full payment accepted. Apparently the CEO of alpha t is an accountant and other directors are lawyers, and the alpha t preorder contract is clearly structured as a B2B transaction and is watertight.

 

[Attic]

The alpha technology saga has taken an interesting twist.

They had previously given firm shipping dates in July, and had started taking final payments.

Turns out that they were just taking out, and the tape out failed. Lol.

PayPal then froze their account leaving them with no funding to revise the ASIC design and reattempt tape out.

They've managed to obtain some rescue funding, and are now preparing to tape out again. They now need to start work on the PCB for the miner which could not be built any earlier because the ASIC ballout isn't final until tape out.

Lots of angry customers as no refunds if full payment accepted. Apparently the CEO of alpha t is an accountant and other directors are lawyers, and the alpha t preorder contract is clearly structured as a B2B transaction and is watertight.

I looked at Alpha T at end of 2013. Didn't take me longer than a few hours to decide they were in no way a trustworthy company to send 1000's of dollars to.

It's a scam and they are a deceitful entity to do business with.

 

[dajeepster]

I looked at Alpha T at end of 2013. Didn't take me longer than a few hours to decide they were in no way a trustworthy company to send 1000's of dollars to.

It's a scam and they are a deceitful entity to do business with.

i've learned several things from all these pre-order companies.

1. If they aren't the first to market.. avoid them.. unless you are BFL which you should avoid anyways
2. if they offer a pre-order.. avoid them
3. if they create their own forum instead of using the bitcointalk/litecointalk forum.. avoid them.
4. And just because they were successful once, doesn't mean they'll be successful again.

in fact.. just don't do asics.. don't mine btc/ltc... stay gpu/fpga and mine alts and convert over. i'm still making money, but i have all my gpus condensed down to as few rigs as possible... only mining x11, x13 right now. nist5 and x15 seem to be dead for the time being.

and by making money.. i mean i can buy a cup of coffee at the end of the month.

but am i getting out... No..

i was in distributed computing long before there were bitcoins.. i still hold some of the top scores on TA. it's a hobby.

 

[thilanliyan]

Yeah I seriously toyed with the idea of buying a Scrypt ASIC but decided it wasn't worth it in the end...too much risk. I'd probably be better off just spending that money on some BTC.

 

[pm]

I design integrated circuitry for a living. I know all the tools and flows and I've been designing chips as an electrical engineer for over 20 years now. So a bunch of my co-workers and I got together and decided if we wanted to design a scypt-based ASIC. In our spare time, we coded up the RTL, got it working in an FPGA (it was slow and made no sense to continue down that path), ported it to a verilog simulator simulator, showed that it could basically work... and then we looked at the costs, the risks, the timeline and decided that it was all just too risky for us. We basically decided that we needed about $900-1100k to pull it off in a small way and we decided that in this post-2008 age, we would have a real problem getting the money from a bank. And if you say "$900k? really?" you should look at what even a partial mask set at a foundry runs, plus debug equipment, plus software licenses to synthesis and place-n-route tools

So there's a flip-side to these guys... you can say they are all scammers and that pre-orders are a scam, but the other side of it is that you are basically fronting an incredible amount of risk. You could design a scrypt based ASIC and at the end of the project you could miss a check and tape-in a "dead rat" (industry speak for a chip that doesn't power up due to a mistake), and even if you execute flawlessly you can't be sure of what your competing teams are up to or even what the cryptocurrency market will be.

So it seems to me that you have to take pre-orders... because where else are you going to get the money? Venture capitalists want a company they can take public or are make an acquisition target... which a crypto-ASIC wouldn't be. Banks won't touch this. The only other option to pre-orders is to pool and gamble your life's savings... and good engineers tend to be risk averse.

 

[bkiserx7]

Wow, thank you for all the insight. That definitely shines new light on things.

 

[Piotrsama]

Yes, very interesting read.

 

[Mark R]

Thanks pm.

In all honesty, I was a bit surprised that the cost would be as little as $1 million. Perhaps, I was thinking more along the lines of KnC who have the model of using a few massive dies, as opposed to bitfury or gridseed who use small (10 mm2) dies. I dread to think what the cost of a 20nm mask set for a 500 mm2 KnC neptune ASIC would be.

Alpha look like they have tried to sub out the entire design, and therefore probably didn't really have anyone on their staff who really understood the design and production process.

Now, AT's CEO is an electronic engineer, but that's quite a broad description. While I've got PCBs fabbed, by finding a house, downloading the DRC rules off the website, and e-mailing off a bunch of Gerbers with a note saying "plz make 50 of these", I suspect that with IC fabbing , the to-and-fro discussion with the foundry and their in-house design consultancy needs to be a bit more nuanced (and expensive).

 

[pm]

Thanks pm.

In all honesty, I was a bit surprised that the cost would be as little as $1 million. Perhaps, I was thinking more along the lines of KnC who have the model of using a few massive dies, as opposed to bitfury or gridseed who use small (10 mm2) dies. I dread to think what the cost of a 20nm mask set for a 500 mm2 KnC neptune ASIC would be.

After some research, we finally decided we wanted to go with MOSIS -and they have a shared wafer approach that dramatically cuts mask costs. I talked to them a bit on the phone about pricing. We were thinking low volumes - again to reduce risk. We were also looking at an older process technology (to reduce mask costs). Our preliminary (relatively wild guess) estimate of die size was very small but we hadn't done any IO, clock generation or any of the stuff that gets abstracted away in an RTL model. We never went through synthesis so our die estimates were based on gate counts out of our verilog simulator and were probably way way off from reality.

Link to Mosis:
https://www.mosis.com/what-is-mosis

I know I got into electrical engineering because it's fun to design things and so there was an element of "lets see if we can get it working" and a bit of research to estimate theoretical costs and profit margins. We never really got too far beyond the stage of "ok, we can do this... do we really want to do this?" before we all got cold feet.

 

[thilanliyan]

After some research, we finally decided we wanted to go with MOSIS -and they have a shared wafer approach that dramatically cuts mask costs. I talked to them a bit on the phone about pricing. We were thinking low volumes - again to reduce risk. We were also looking at an older process technology (to reduce mask costs). Our preliminary (relatively wild guess) estimate of die size was very small but we hadn't done any IO, clock generation or any of the stuff that gets abstracted away in an RTL model. We never went through synthesis so our die estimates were based on gate counts out of our verilog simulator and were probably way way off from reality.

Link to Mosis:
https://www.mosis.com/what-is-mosis

I know I got into electrical engineering because it's fun to design things and so there was an element of "lets see if we can get it working" and a bit of research to estimate theoretical costs and profit margins. We never really got too far beyond the stage of "ok, we can do this... do we really want to do this?" before we all got cold feet.

I'd buy if YOU made one.

 

[dajeepster]

I design integrated circuitry for a living. I know all the tools and flows and I've been designing chips as an electrical engineer for over 20 years now. So a bunch of my co-workers and I got together and decided if we wanted to design a scypt-based ASIC. In our spare time, we coded up the RTL, got it working in an FPGA (it was slow and made no sense to continue down that path),

Have you thought about just doing FPGA for the alt coins such as x11, x13, x15, nist5 and so on?
that way when a new alt comes out, just reload the code. granted, it's slower than asics, but you can always change the code to a new alt when it comes out which doesn't have an asic for.

whereas for the asics, you are just one and done and hope for ROI

i'm an electrical engineer also, but i do process fabrication... i make diodes one at a time... and it's a PITA.

 

[Mark R]

Alpha T have just announced an IPO - 30% of the company being offered to paid up miner customers.

I don't really know what to say other than. Very WTF, Much Wow.

 

[dajeepster]

I don't really know what to say other than. Very WTF, Much Wow.

my thoughts exactly...

they don't have a working product, because they failed on the first tape out. so the people that paid money are screwed (so far).
they're trying to raise more money, by trying to incentivize current customers to pay for a product that is an initial failure, by offering an IPO in a failing company... they don't have a board layout or anything beside a back plane of a powersupply board that just looks like an extension for PCI-E power plug (how about just putting the PCI-E connectors on the boards.. like everyone else does.. less connectors, less points of failure).... (maybe i missed the other boards)...they've got nothing

Sure... i'll do that.. -_-

i'm glad I only did the small asic.. i'll just write it off as a loss on next years taxes... and i only did the initial 30%
Alarms went up a when they started to send out invoices, and my invoice had a VAT on it, even though I'm in the US.
I emailed them about it and i never got a response back. A couple weeks later they resent all the invoices with the correction on the speed because they had increased the speed, but the original invoice didn't reflect the new invoice (and the forum pointed that out)
well, the new invoice still had a VAT on it... i'm holding onto the rest of my bitcoins.

 

[dajeepster]

x11 has been rocking it the last couple of days

i.e. Iocoin

 

[njdevilsfan87]

Crypto is not looking too good right now... :hmm:

LTC just feel under 0.01 and the entire altcoin market is in shambles (DOGE at 24 sats) while BTC is starting to fall again. Time to get rid of some. I think BTC has to test a long term double bottom of $350-$400. There was so much good news recently about it, and it barely moved in price. And if does LTC is going to end up like $2-$3 in price.

 

[WhoBeDaPlaya]

I know I got into electrical engineering because it's fun to design things and so there was an element of "lets see if we can get it working" and a bit of research to estimate theoretical costs and profit margins. We never really got too far beyond the stage of "ok, we can do this... do we really want to do this?" before we all got cold feet.

I talked to a couple of product definers awhile before BTC took off and got laughed out of the room ("You're just a physical designer, a lowly automated polygon pusher, what do you know?")
They're not laughing so hard now

 

[Elfear]

So I'll be away on business for ~4months and I have my mining rig with me. It looks like the most profitable coin to mine right now with free electricity is Vertcoin or Spots coin. Are those my best bet?

 

[T_Yamamoto]

I haven't mined in so long. Once I move in, I think I'll jump onto a multipool.

 

[thilanliyan]

About mining on Nicehash/Westhash...so difficulty is lower at Westhash than Nicehash. So should I in theory get paid more with Westhash (assuming payout BTC/MH/day is the same)?

How can I test it out? Would WU be different if difficulty is different?

 

[gorobei]

http://www.thecoinsman.com/2014/08/bitcoin/inside-chinese-bitcoin-mine/

seems avalons are only worth the psu inside. $60,000 per month in electricity costs, is coal power that cheap?

 

[KingFatty]

http://www.thecoinsman.com/2014/08/bitcoin/inside-chinese-bitcoin-mine/

seems avalons are only worth the psu inside. $60,000 per month in electricity costs, is coal power that cheap?

After reading that article, it struck me that perhaps the electricity cost is due to poor planning/engineering?

I mean think about how many times the author mentions the windy conditions, the curtains blowing etc. Well, that indicates to me how poorly the airflow is planned out, to where you end up spending lots of money to just blow the air around in ways that are not directly contributing to cooling. I bet a lot electricity is spent simply blowing air around in a circle within the building, e.g., to cause the curtains to flutter. Look at the huge distance between the fans and the heatsinks they are actually directed toward - so much waste to blow the air across the entire aisle before hitting the cards.

They need to learn about basic layout and having hot aisle/cold aisle and directing the air accordingly.

 

[Mark R]

KnC have announced that they have received first silicon for their 28nm scrypt ASICs.

2284 cores per die, with 8 scrypt threads per core. I bet the die size is monstrous with all the RAM requirement.

I must admit, I don't really get see how you can get 8 threads on a core with scrypt - unless, they hold 8 RAM buffers and run the salsa engines faster than the RAM. Or perhaps, it's pipelined, but I don't really see how you can do much useful pipelining on scrypt, because every stage is designed to be sequential and dependent on RAM.

Oh, and on the alpha-T front: their website has gone!

 

[njdevilsfan87]

LTC is $4.40. Lol. That's the price it was when I first got into mining in April/May 2013 when it had its first mini-bubble. I bought some at $5.30, then some more at $4.25, and will have two more cost average as $3.35 and $2.60 if it gets that low. Either I'm going to lose it all soon or be very happy in a few months. Useless it may be doesn't change the fact this market is still based off speculation mainly as opposed to fundamentals. It's all profit money I've made off other alts I'm playing with now anyway. One can always dream of it going to back to $48 again.

 

[x3sphere]

Unless there's a massive BTC price explosion I don't think LTC will ever hit its previous ATH, there's way too many alt coins now and the money flowing in is getting more spread out.

There's probably another rally left in it, though, whenever BTC takes a spike back up

I have all my alt coin money in XCP and XMR now... both have went down a little these past few months but nowhere near as much as most of the other alts. That to me a is good sign they'll do well once this BTC downtrend ends.