Topic: bASIC - speed boost - 54G now 72G, 27G now 36G (~) (Read 3696 times)

Thanks for the info. Any timeline that you are aware of for the info to be released?

@ itsgoldbaby, my apologies. I missed your quotation marks. It is interesting news. I think most people who were not diluting them self ever believed Tom could make the ship date he told the community.

Cool, not sure of the full meaning of this though. If you mean I have ordered from bASIC, your right. I posted the message from the btcfpga over to here for people to see, as well. Thought people might like to see the "news". Have fun with your trolling mate, good luck.

Sock puppet identified, itsgoldbaby.

Next time, dont release get our hopes up about a release date you know you cant pull off. I mean fuck, first BFL, now bASIC...
Really guys?!?! How about, once you actually get chips & boards & assembled .. along w/ numbers out to the community.. Then FUCKING CLAIM YOU CAN SHIP! But not before then.
This is pirateat40 all over again except this time holding our fucking BTC for something these companies cant even prove to us customers..exists....!
Fuck that! Horrible business ethics.

https://www.btcfpga.com/forum/index.php?topic=125.0

"Hello

I really can't express the depth of how incredibly pissed off I am about this project (as I am sure you are as well). After hours of meetings and discussions today with my engineering team I don't see any possible way we will be able to release the bASIC before mid January.

When I first planned this project I was quoted a 2 week design time for the board itself and that was the time we planned for. Due to a conflict of interest we had to find a new design firm, one that came highly recommended and is well known and respected / reputable. This new design firm although incredibly talented is equally thorough and things are taking longer than I had expected.

I am planning now on a mid January release date and I deeply apologize to all of my customers, and I do not blame anyone at all who wants to just cancel their order and get a full refund.

I also think it's very important for everyone to know that Dave is just an employee of BTCFPGA he does not own any part of ownership of this company and cannot be blamed for its problems. I take full personal responsibility for this delay and any other problems anyone has encountered with our company.


REFUNDS:
For those who cannot wait until January and want a refund (or for any other reason) I do not blame you what so ever, and I encourage anyone who has any doubts of this project to please request a refund immediately. We will begin issuing refunds as early as this evening, so for those who have requested a refund you will receive it over the next 1-2 days.

Also please keep this in mind - if you want a refund please just request one by sending an email to [email protected] (he now has the power to issue refunds) You will receive it MUCH faster than by requesting a charge-back through your bank.

Please accept my sincere apologies for not being as available to you as I would like to be,  this whole situation really has me at wits end, however I WILL be available tomorrow 11/28 in afternoon."

It's called "Home Owners Insurance",hope you've got it  

Great discussion on power & cable options,Thanks!!!

Why so many ignores for MeSarah    Look at the post history   He's also a BFL "asshamster" (trademark protected) 

Yes, the more people push the ignore button the more the button turns that funky piss yellow.

If someones ignore button is yellow, does that mean tons of people have ignored them and the forum is suggesting I do so also?

I've only seen two people with yellow Ignore buttons now, electricmucus and MeSarah

ok, who unbanned the above moron?

If your running a bASIC product from Tom and your house burns down, how does that affect your ROI? Lets hope that no one will die.

The exact power numbers have not been released yet so none of us know. Hopefully soon we can get the numbers and start throwing BTC at Cablez

Sorry for the OT post but didnt want to start another thread on this;

Does anyone know the power supply specs and connection requirements for the 72 GH bASIC? Trying to spec out a PSU cable from Cablez but he says he doesnt have enough info to feel confident on setup. Any help appreciated. Feel free to pm me

Ah, OK, that sounds plausible.  Still, there has to be some reason why Molex chose to specify different ratings for different wire gauges and for whatever reason, they've rated an 8981 on 18AWG (the ATX standard) at 8A.   It's difficult to know exactly how the current density will be distributed in the crimp region...

roy

The wire doesn't stay round when you crimp it, it ovals out. Check out my super accurate 60 second paint drawing.


Maybe accuracy could be better, but you get the idea. The smaller diameter the wire, the more it deviates from round and the more surface area you get relative to a circle.

@MrTeal:

Also, just did the math and the figures match perfectly.  18AWG has almost exactly 20% smaller diameter than 16AWG, and hence 20% smaller circumferance.  I don't know exactly how 8981 connectors are crimped, but wouldn't that result in 20% smaller contact area?

And Molex derate the connector by 20% (i.e. from 10A to 8A)

No, sorry, that's pure speculation on my part; I just couldn't think of any other reason why Molex would derate the connectors for 18AWG and smaller...

roy

Do you have a reference that indicates the reduced rating is caused by a smaller crimped contact area? The effect won't actually be as great as you're indicating; you have almost the same contact area crimping to 18 gauge as you do 14 gauge.

It's not the wire run I'm worrying about, it's the reduced contact area between the wire and the connector when you crimp it.

Molex only rates the connectors for the full current draw when you use 16AWG - see http://www.molex.com/pdm_docs/ps/PS-8981-4M4P.pdf

roy

You should be fine drawing 120W through the Molex connector even with 18 gauge wire. If your PSU is pretty standard with the first 4pin being about 1.5ft along the cable, you'll end up with a round trip resistance of a little under 15 milliohms for the single 12V and dual ground wires. You'd only lose 0.15V across the wire, which really isn't a concern. Wire heating should also not really be an issue unless your application is extreme.

Ah, OK, so that's where the Seasonic's 7A limit comes from - I should have spotted that.


Perhaps true for a premium quality non-modular PSU, but the ATX standard specifies 18AWG on the 8981 so you certainly can't rely on being able to draw 120W from the Molex on a low end PSU.

Good to know that there are PSUs out there that could power the bASIC directly from a peripheral connector - but notwithstanding that I think my point is still somewhat valid.  Many (most?) PSU's will not be able to power the bASIC and remain within spec without the use of custom cables.

I'm also wondering what the current draw of BFL is going to be.  If it's on the high side I may end up deciding to stay with the 13-14V wall wart to minimize the current through the barrel connector.

Right now it's true that we really have no hard information on power requirements for either product - and I imagine we may not for several weeks.  I guess all we can do is wait and see.

roy

Only if you think profits and losses are the same thing.

E.g. one thing returns $10/mo, another option returns $5/mo. Now add a $6 cost to both sides.  Nor can you say that if one or the other was the winner ignoring the constant cost it would still be still be a winner with it as they don't have initial upfront costs.

Personally, for infinite duration things I like to convert constant future operating costs to unfront costs:  $10/mo in today's value forever costs you roughly $3000 (assuming 4% post-tax/inflation return).  Though mining isn't infinite duration and the long term trend of energy costs certainly seems to be upwards.

If both systems require the same variable, then they cancel each other out.

Side note, I'm hoping to cut that host PC down to < 5W. Maybe soon I can actually turn my computer off at night!

You've forgot to consider at least 70W for the PC to control the ASICs (for both systems). On the other hand the 0.09c KWh is really low for most countries. In Europe range is 0.12-0.38 with most of the countries over 0.25c KWh. Considering the 0.25c KWh,  Basic lose over BFL over the 20X with $138/mo versus $139/mo. At 60X Basic will be barely profitable with $8.54/month.

OK. Fine. I'll double the maximum power estimate to 240 watts (which I think is an outrageously high number). The bASIC is still on top.

1. Buy from BFL (~$1300 invested)
    a. Get 60 Gh
    b. Get your ASICs in January/February
    c. Consume 60 Watts ($3.94/mo @ .09c KWh)
    d. Estimated profit at 10x difficulty: $327.52/mo
    e. Estimated profit at 20x difficulty: $161.79/mo
    f.  Estimated profit at 30x difficulty: $106.54 /mo
    g. Estimated profit at 40x difficulty: $78.92/mo
    h. Estimated profit at 50x difficulty: $62.35/mo
    i.  Estimated profit at 60x difficulty: $51.30/mo

2. Buy from Tom (~$1070 invested)
    a. Get 72 Gh
    b. Get your ASICs in January/February
    c. Consume 240 Watts (maximum estimate) ($15.56/mo @ .09c KWh)
    d. Estimated profit at 10x difficulty: $382.09/mo
    e. Estimated profit at 20x difficulty: $183.21/mo
     f. Estimated profit at 30x difficulty: $116.92 /mo
    g. Estimated profit at 40x difficulty: $83.77/mo
    h. Estimated profit at 50x difficulty: $71.66/mo
    i.  Estimated profit at 60x difficulty: $63.88/mo

EDIT: That "bASIC will use more than 120 watts" argument relates to the "BFL can't get 1Gh/watt argument", both of which no one really knows the answers to. It is doubtful any manufacturers have anything more than a prototype.

Do you think there will be any adverse effects of powering through both the 8981 and the barrel at the same time, of course without knowing how the pcb power circuitry is laid out yet?

Warning: Irrational exuberance detected.

Why the heck would you expect that a "maximum estimate" established for 3/4th of the hash power would still hold?

The AX1200 powers the entire 12V wire of a peripheral cable through a single mini-fit jr pin.

Personally, I planned for the worse: 120W power consumption or even higher. Given the poor choice of connectors on the bASIC, the only way to handle a 10 Amp @ 12V load with a modular PSU is to use a Y-cable. So I manufactured myself 10 of them. Each has a custom AX1200 modular connector on one end (populated with 4 mini-fit jr pins), and peripheral + power barrel connectors on the other ends. Both ends will be plugged in the bASIC. I built them out of 16 AWG and made them 4ft long, so they can handle 120W with only a ~165mV voltage drop, or even 168W with only a ~233mV drop.

The fact that Tom recommends a modular PSU (with no specially designed power cables) for the 72Gh/s bASIC means that either (1) he knows power consumption will be much less than 120W, or (2) he is as poorly prepared as all buyers and doesn't realize he will be driving the single mini-fit jr pin on the PSU's end over spec...

This lack of info on power delivery among other things is making it challenging to plan for power delivery/equipment (let alone running purchasing roi numbers). Anyone know when we are supposed to get the info?

Not that I've seen as well.

Sounds like RAM Guy should stick with RAM then. While I don't rule out the possibility of the AX1200 using two pins on the modular connector for 12V, it would be unusual. Molex doesn't make any Minifit Jr connectors rated at 11A per pin as far as I know.

The AX1200 is a fully modular PSU as well.

The Seasonic rates the 8981 string to 7A because it's a fully modular PSU and those cables have a single pin of a 6pin Minifit Jr provide 12V to the whole string. The problem isn't with the 8981; a non-modular PSU wouldn't have that issue.

The molex 8981 connector standard is 11amp max current, not sure why Seasonic would limit it to 4A under that, nor if that is the standard, but I would suspect not.

EDIT: One product, the Corsair AX1200 has had techsupport answer 11A for their peripherals:

http://forum.corsair.com/v3/showthread.php?t=100140

TBFT (to be fair though), I'm not in love with either molex or PCIE for power.

At this point I am holding my breath until Tuesday to see what Tom's engineers are up to. Depending on the power draw of the device a multi-12v line cable may be needed. <--this is just speculation by the way.

Once I have some solid numbers I can hopefully devise something elegant.

If not, there are tons of connectors out there - one in particular I can think off of the top of my head that I've put 100 amps through on a regular basis at very low resistance that one could replace the stock connector with if he felt uneasy about the one that was included.

Granted you shouldn't have to re-engineer something you spent your cash on, but if this is the least of my concerns - I'd be a happy camper!

Hey Cablez,bASIC's guys have a question & need your help!!!!!

Posted by Roy Badami :

"I'm rather worried about this.  80-120W @ 12V is 6.7-10A.   steve contacted Seasonics in another thread and we know that Seasonic PSUs can only supply a maximum of 7A on their peripheral (8981) connectors - i.e. what's commonly called a Molex connector.  That's only 84W @ 12V.  Other PSUs are unlikely to be any better.  So we seem to have a situation where it will quite likely be impossible to safely power the bASIC from the Molex connector on even a premium quality ATX PSU, but will instead require a custom cable to connect the Molex connector on the bASIC to a PSU connector with multiple 12V rails pins, such as a PCI-E connector or CPU connector.  

This seems far from ideal - I think we really need the bASIC to have a connector with multiple 12V rails pins, such as PCI-E.  I'm not sure how easy it will be to source 10A barrel connectors, either.  Josh has said that BFL have been having difficulty sourcing high current barrel connectors [ETA: and Cablez has likewise said he's had difficulty finding barrel connectors rated at 10A] although given BFL's lower (stated) power specs, and the fact that they seem to be planning to go with 13-14V for their wall warts, connector power draw is probably less of an issue for them.

I'd really like to be sure that the bASIC will come with a connector that is safe to run at 10A for extended periods of time without running so hot that it melts or otherwise creates a safety hazzard. Some clarification on this would be most welcome.  The Molex solultion works fine on the MMQ because it only consumes 40W - which is a current draw of just 3.3A.  I just don't see how it's going to safely work for a miner that's going to consume 2-3 times the power. [ETA: unless you use two of them, of course]"

https://www.btcfpga.com/forum/index.php?topic=104.30

Show'em what you got dude!!!!!!!!!!!!!  

Edit:Here's what we ALL can use 

https://bitcointalksearch.org/topic/wts-atx-psu-to-25-or-2155mm-12v-coaxial-plug-power-cables-new-product-74397

 

1. Buy from BFL (~$1300 invested)
    a. Get 60 Gh
    b. Get your ASICs in January/February
    c. Consume 60 Watts ($3.94/mo @ .09c KWh)
    d. Estimated profit at 10x difficulty: $327.52/mo
    e. Estimated profit at 20x difficulty: $161.79/mo
    f.  Estimated profit at 30x difficulty: $106.54 /mo
    g. Estimated profit at 40x difficulty: $78.92/mo
    h. Estimated profit at 50x difficulty: $62.35/mo
    i.  Estimated profit at 60x difficulty: $51.30/mo

2. Buy from Tom (~$1070 invested)
    a. Get 72 Gh
    b. Get your ASICs in January/February
    c. Consume 120 Watts (maximum estimate) ($7.78/mo @ .09c KWh)
    d. Estimated profit at 10x difficulty: $389.87/mo
    e. Estimated profit at 20x difficulty: $190.99/mo
     f. Estimated profit at 30x difficulty: $124.70 /mo
    g. Estimated profit at 40x difficulty: $91.55/mo
    h. Estimated profit at 50x difficulty: $71.66/mo
    i.  Estimated profit at 60x difficulty: $58.40/mo

So... if you have decent power rates, you can invest $230 less and still make more money by buying with Tom (aslong as you have an old PSU lying around.) Yes, eventually when difficulty sky rockets the BFL unit will still be profitable when the bASIC unit will not be, but I have a feeling 2nd generation ASICs will be out before that happens and everyone will trade in their old units anyways.

EDIT: deleted creepy storage wars meme

A result of hardware changes made.


Source: https://www.btcfpga.com/forum/index.php?topic=104.0

EDIT: Found another thread on this in wrong forum / https://bitcointalk.org/index.php?topic=127490.0;all