Topic: The Habanero - 650GH/s - OOS - page 30. (Read 96051 times)

interesting, well i hope for an update :p got the btc ready!

Every time I've checked in the last 36 hours it was still sold out. Figure they probably won't update stock until Monday?

did they re-open it last night with the last of stock and i was asleep? or have they not added the remainder (if any) to the website...

Drat.  Was still open last night and the plan was to add one more today. 

I'm using liquid metal and it's a bit fiddly to apply but that is nothing in comparison to building this:




I'm thinking to buy new waterblocks because at the moment I can't push these boards over 750MHZ.
Which waterblock would you choose?
http://www.overclockers.co.uk/showproduct.php?prodid=WC-002-AB&groupid=962&catid=1519
http://www.overclockers.co.uk/showproduct.php?prodid=WC-230-EK&groupid=962&catid=1519

Whats the max voltage settings?

Specifically, when should I start worrying that I may be over-volting a chip?

Is there a way to tell that I've gone to far before chips start frying?

Mrteal do have the time to write a cgminer setup guide and the rip setup guide with cgminer ? S that all those customer who bought it can go this post , thus saving time to seach and dig . My 2cent.

Ah crap batch 2 sold out

Would love to see this. Or how about a peltier? 

If you decide to do this, of course you must shoot a video and add it to the blog

So you are trying to tell me you don't push your hardware to get the max value out of it? You are just happy with out of the box performance? Well I am not, this is a hobby for me and really enjoy it. Mommy has not paid for anything in a long long time, my 70 hour a week job does that.

Entertainment, mostly. Cheaper more efficient power supplies, trading up hardware, and rigging everything myself instead of paying someone else or using off-the-shelf equipment avoids problems like spending $1500 to power and cool six boards. Having 200A service at the house and $0.09/per KWh helps too. Turning mining profits back in to build hashpower with difficulty has also been working well, instead of throwing a whole lot of money in the pot at the start and hoping to get it back before the hardware is obsolete.

In the spirit of "optimization", I have a board that we swapped out for a customer that had a die down. I'm going to send it back to the assembly house to get it xrayed to see if there is a cracked ball under the ASIC for that die, but before I did I wanted to see how far I could push just 3 dies where in theory thermals should be less of an issue.

The interesting thing is that each PCIe plug is its own rail, with its own 6 phase VRM feeding one die on its own plane. It's not like the VRM for the phase that's down is helping out; it's not even turned on. The only difference between 3 dies as 1050MHz and 4 dies at 1050Mhz is getting the heat away.
I really think I need to buy some of that Cool Labs liquid metal TIM (despite the risks) and hook up a real water block to the municipal water supply, just to see if I can get 1050 on a fully working chip.

I was waiting on a post like this... and I have your answer... though you still wont see anything but a minor payoff unless price of BTC goes up.

Basically... if you look at the landscape of the ASIC technology coming out this year... it looks eerily similar to the HD 7 series cards hitting the market before ASICs finally got dropped by Avalon and others. We are sitting at the 28nm line as of right now just awaiting the mass of miners to unleash. Just like when the HD 7 series arrived and all the CPU miners had to move to scrypt, most of the early ASICs will now be in negative returns in respect to electric costs and will be sold off to people who don't care about electric cost or just plain deactivated, leaving swaths of hashrate dying out of the global pool. this means soon, the difficulty increase will begin it's inevitable plateau leaving some small room for miner ROI.

Why is this true? Because the next steps for ASIC efficiency are into 22nm and 14nm territory, which as right now is something only Intel, Nvidia and AMD have really started mass producing (there are others I am sure but the two foundries making them are all used by these vendors). The foundries with this technology already have their calendar full of multi-million or even BILLION USD roll outs for these major vendors. Intel and Nvidia are definitely already pressing 14nm tech... and AMD is moving out of 22nm now and starting to work at the 14nm level. This means there is not a single SHA256d ASIC developer out there that could get production time anytime soon. At BEST, someone will eventually get in the abandoned 22nm line when INTEL, AMD and Nvidia have all moved the majority of their product to the new 14nm on a solid basis.

What does this mean for us miners? It means you will see one more "double down" in ASIC efficiency in the next 18 months which will kill off 65nm and 55nm miners completely. Look at it this way...

1. 65nm BFL chips are turning LESS than 2 cents per day per GH/s at $.10 per kWh at 5-6 J/GH/s
2. 55nm Bitmain chips are turning LESS than 2.5 cents per day per GH/s at $.10 per kWh at 2+ J/GH/s
3. 28nm HF Habaneros are turning LESS than 2.6 cents per day per GH/s ay $.10 per kWh at 1 J/GH/s

Nothing that I have seen (that I believe anyways) has proven to really mine below the 1 J/GH/s mark yet... that can be had by consumers at this point. I won't believe the Monarch figures until someone third party verifies it in their own hands. The odds of us seeing a 22nm product that is double the efficiency of these 28nm products is laughably impossible. Not until 14nm ASICs will we see a double down from these Habaneros and other 28nm miners. The time and funds it is going to take to develop 22nm would almost be for naught except it is going to be the second halving before anyone can say they have a 14nm chip lined up at the foundry.

Put all that together and you should see, by the end of the year, a price increase in BTC, a plateau in difficulty... and people ditching their old gear for 28nm as quickly as the vendors can stock the miners... all of which should enable a Habanero to ROI this year.

If anyone disagrees or has anything to add or clarify on the above... i welcome feedback.

~Daemon

Hey guys, I too snoozed on the 2nd batch, I wanted to get 6 boards, but after plugging the numbers in to the bitcoinwisdom.com calculator I just couldn't get anywhere, all the time coming out in a loss, unless I upped the bitcoin price, but that is not the case at this point, if the price stays at $6xx then I will never even return my investment, even $700 doesn't seem like it would be enough.  How do you guys look at this, if I get 6 boards(~700 Gh/s x6)=4.2 Th/s, power .15 kWh, difficulty inc. - 20%, hardware cost - 875 x 6 = ~$5,250 + 6 PSUs(~120 x 6 = $720) + 6 Cooling systems(~120 x 6 = $720) ---  Total cost = ~$6,690.
I agree that yes it will be very fun to play around with the Habaneros while putting them together and configuring/perfecting them, but at the end of the day or week or month I want to make some money as well.  So I guess what I am not understanding is how would I be able to make my investment back and make some money with the Bitcoin miners such as "The Habanero".

Another difficulty I'm seeing is, of course most people are setting these things up at home, but at 1000 watt a board you could only have a limited amount of these boards set up in the house and even then in different locations of the house(if whoever you live with will even allow these devices in their spaces of the house or a mess of very long power cables running all over the house to the rig station) so that each board uses 1 circuit, so at around 9 boards my circuit breaker wouldn't be able to handle any more. 
How are you guys coping with these electrical limits?
What still keeps you in the Bitcoin mining game with the increasing difficulty?

I think you are confusing "fooling around" with "optimization". The Habaneros provide an opportunity to optimize a solution based on the specific variables you are dealing with (electricity cost, total load, ambient temps, space). This takes a little time but is a much better option than receiving a shitty build with a bunch of shitty components that many hardware manufacturers have shipped. Those products often require tearing them apart and tweaking in order to fit ones environment.

In theory I would support your angle on this, but given the fact that the guys behind PepperMining have taken the best qualities of failed projects to create something as beautiful as these boards with an almost completely open source approach, I can't.
In the true spirit of homebrew, these guys have done something amazing and anyone who wants plug and play should probably go find another miner.

I was reading through some random part of the forums last November or so and someone said "the only way to make a profit in bitcoin is to sell hardware, or to get hardware for free". To which I responded, hardware for free? Challenge accepted. In the last 9 months I've turned a single AM Blade into 2.6TH, utilities paid, with zero out-of-pocket because I played with the hardware and turned what I learned into ways other folks could benefit. My mother didn't pay for a thing. Just saying, not everyone has lots of money riding on good setups turning quick bucks and not everyone is in the game just to make a fiscal reward. Some of us are just really having fun seeing what's even possible. For example, I'm working on building a custom setup good enough to keep about half a dozen of these Habaneros cool (if I can get my hands on that many, right now I'm hoping for one) for less than about $300 in external equipment. $400 if you include power supplies.

Some people have big money involved and every second you are "fooling" around with shit it's not hashing making a reward. If the whole idea is to just fool around with hardware and squeeze a hash point higher then everyone else with a custom type setup then mommy is paying for everything or you just have bank.

Ha. Yeah, with the tools provided it really does make the monkey work way easier than any other hardware I have had and I think I have had damn near one of just about everything.

Hah, thats what i get for cut and paste and not watching the freaking html tags that came along with it.. fixed and thanks for pointing it out.