Bitcoin Forum

Hi all,

I've not been around on the scene much this past year, so I've been catching up on the current news and trends.
I'm glad to see the ASIC's dream is almost realized :) that make much more sense that churning through ATI cards...

So I've been checking the suppliers out... and unlike the GPU wars... this is heating up a little more on the design front than I've ever imagined it would.

I've looked through the main options and it seems to me that technical / power wise... the BFL SC single comes out tops ( please don't flame me if im wrong... im still switching from card based thoughts to self contained units :P )

So i need some opinions from you more experienced guys and gals:

Does anyone have like an up to date comparison table for pwr Vs. GH/s on the respective competitors like this?

http://organofcorti.blogspot.co.uk/2012/11/92-asic-choices-update-2nd-november.html


I could imagine some sort of league table.. fantasy football style going with these guys now :P and seeing who trumps who and who can make that all important come back. It does seem that there have been a few teething issues, and for those of us who are less technical than the designers themselves would benefit from some ASIC comentator on the upcoming ASIC race :)


I really like the idea of having the coffee cup warmers...  on a few powered USB hubs being controlled by my Raspberry PI's, does anyone have any thermal details on these puppies? I can't imaging max 5v really pushing the heat out.. maybe long term the would bu nothing like a 400w psu would do.

Also.. can anyone answer why the other ASIC designers are spec'ing such high power supplies on their offerings? is this to cater for future cluster scaling ?

Cheers

Rock

No one knows yet. It's all vaporware and promises that have been changing daily or weekly.

On paper BFL looks the best, but as usual their execution is meh and will probably take 3-4 revisions after they ship the first lot to make them work like they say they will. If they don't default and fold.

Basic looks good, but is apparently restarting from scratch, who knows.

Avalon looks to be the first out of the gate. It may not be the most efficient, but it is unique that it's a full stand alone unit. It also has more parts that could fail. But it is also open source so you can tweak things.

Powerwise initially power wont matter, since even the worst asic is not going to pull the power of a gpu rig while doing 10-100x the hashrate. Whoever hits first will "win". Once difficulty rises, power will play a larger role, but again the worst asic rig will only pull the power of a desktop PC. You wont be able to heat your home with them.

The best thing to do is diversify if you can. Avalon is sold in waves and is currently sold out. Basic and BFL will take orders, and make on demand. At this point if you aren't already ordered, you are probably better off sitting on your cash or buy bitcoins. It's going to be a bumpy ride for the next 6 months or so.

I expect asic prices to drop drastically since competition will be fierce, and difficulty skyrocketing. The asics who hit after the "winner" makes it to market may take an incredibly long time to break even. It's a much much poorer investment than gpus were.

http://4.bp.blogspot.com/-Mg0lrwAi9jI/UJOpzXMl4vI/AAAAAAAAC8Y/4UaoTiCfH_8/s1600/ASIC.table.2012-11-02.png
I don't know if any updated versions were made, but here are the changes to bring you up to date:

BFL info is mostly unchanged. $1299 for 60GH/s with a claimed 1GHs/Watt.
Unique features: Will probably be the most power efficient unit. No-questions-asked lifetime warranty. Possible trade-up program in the future.

Avalon is still $1299 for 66GH/s, but it's power usage is still unknown.
Unique features: It is a stand-alone unit - no host PC required. Can possible even host other ASICs. Could be the first to ship.

The bASIC is still $1069, but it's speed has been increased from 54 to 72GH/s. Power consumption is still unknown, but it will not be sold with it's own PSU.
Unique features: Has the highest MHs-per-$ / performance-for-the-price / bang-for-the-buck.

I probably shouldn't say this (for my own behalf, I want to discourage others from buying ASICs as much as possible), but I will anyway.  ASICs have yet to be proven as a much poorer investment than GPU's, especially GPU's after the 2011 crash.

I expect that, after all preorders for ASICs are met, we'll still be looking at around 200 TH/s of total hashing power.  This gives an estimated difficulty of 27,852,735.  Even at that difficulty, a $1,299 BFL miner @ 60 GH/s would generate $436/month.  It would fully pay for itself in just 3 months.  After the crash, and all the way up to today, there wasn't any way for a GPU to pay for itself in less than 6 months that I saw.  And even if you double the estimated hashrate, break even is still only 6 months.

If I had more money to invest in ASICs right now, I would do so in a heartbeat.

ASIC adoption will continue unabated for at least 6 months.

Network capacity will rise to 700 TH in 6-8 months.

At least one ASIC mfg will reduce $/GH by 30-50% in 6 months.




That is all.

And by then (6-8 months), any ASICs purchased now will have paid for themselves already.  So what's your point?

Ok, Thanks for this.. that makes the situation much clearer.

Does anyone know what the generations path is for Asic Design... Kinda crystal ball really... but does anyone in the know, know what or able to estimate what the likely outcome is going to be regarding the FAB time for the current ASIC chips? i mean, is there any forward info on which chips are in design which are in planning... etc.. like the spartan family..

Problem is, your hardware cost is a sunk cost unless you sell it to another bitcoiner. GPU's have intrinsic value for other computing purposes.

So once the GPU "paid itself off" you were actually at a net profit if you sold the hardware then and there.

I'd say 6 months has been 6 months will be six months for mining hardware ROI give or take a month or two.

The few who get their hands on the first 20-40 TH will hit big, the next 40-60 will do okay and once we exceed 200-300 TH it will be same as before or worse depending on how price adjusts as usual.

I'm out of the ASIC race and honestly it feels good.

As that (current ASIC miners)? You mean (possible asic miners).

When they claim "lifetime" they actually mean as long as they sell the product. Once they stop selling it, all the warranties end.

BFL states that warranty for even end-of-life products is to be no less than 1 year from date of sale (this can be found in the FAQs).    I believe that BFL has stated that a new product line is only when a re-design of the chip is done on a lower die size, so that is obviously a long time away from now.  65nm is still pretty new for a small company to make a custom chip on, and recouping the development costs will most likely take some time. 

When I was a kid I always found it amazing how the Americans could give "lifetime warranties" for their marvellous products like teflon pans or Dyson vacuum cleaners. "Wow - a 80 year guarantee!!".

Until I realized that it's only for the lifetime of the product. And that a new version of that product can come within a couple of month.

For me this was very disillusioning.

I see nothing referring to changing die size. It's pretty clear. Once they stop manufacturing that model, all warranties revert to a 1 year warranty. It really is just a 1 year warranty.

What I mean is,  I can see no reason for a total re-design of an ASIC product except for production technology advancement such as die size.  Other than that, there can be a revision - such as a fix to a connector or whatever may come up, but given BFLs excellent history with backing their products, I am sure this is a non-issue when it comes to warranty support.

There are already smaller process sizes in use today, for those with deep enough pockets.  Intel's Ivy Bridge chips use a 22nm process, vs BFL's 65nm.  Remember that the size advantage saves space in two dimensions, such that halving the process size results in a quartering of the die area.

Another redesign possibility is on the logic level, designing the hashing engines to fit into a smaller die area, to scale to higher clock speeds, and most importantly, to consume less energy.

There are probably also other areas where the design can be improved upon, such as improved thermal coupling of the chip die to the chip package.  Using a flip-chip design for instance.

65nm is the smallest process any other small company making fully custom designed chips in any industry can be expected to use.  Intel designs and produces on a scale that justifies not needing to outsource their chip fabrication, for example.  I'm not saying Intel is better than BFL, but only that they might have some resources that are not accessible to companies producing at BFL's level.  It is after all an application specific chip, and that application is hashing.
 
Who knows what the final chip package will be? There might be changes to that too, but to expect smaller than 65nm -  when a few months ago the speculation was that no small company like BFL could afford to go 65nm?    I just want to press fast forward to about 2 weeks from now when the picture is more than clear on how things stand with the various ASIC offerings and initial shipment.

I'm just saying that the technology is already out there.  I was talking about subsequent generations, not implying that we should expect anything better than 65nm in the first-gen chips.

BFL themselves have hinted at a move to 45nm chips for the next generation in the Bitcoin Magazine interview.