[ActiveMining] The Official Active Mining Discussion Thread [Self-Moderated]

[Entropy-uc]

...

6. your projection shows essentially 1.7MW of power usage for those chips.  that takes into account the chip power usage only.  what are the additional projected power needs for all the other equipment?

At this time, we don't have a projection on the boards power; however, the chips consume >98% of the power.
...

 No doubt thanks to Intellihash(tm)

And this, folks, is your EE

IntelliVRM(tm) efficiency >98% !1!!!1!

I would be fascinated to see these greater than 98% efficient voltage regulators Ken has found.  Not to mention the capacitors with zero impedance at high clock rates, zero power USB and I/O controllers, superconducting board materials,...

Seriously guys I take everything back.  With technology like this it's a wonder Apple hasn't bought the company out.

[1715639772]

1715639772

[1715639772]

1715639772

[1715639772]

1715639772

[klondike_bar]

...

6. your projection shows essentially 1.7MW of power usage for those chips.  that takes into account the chip power usage only.  what are the additional projected power needs for all the other equipment?

At this time, we don't have a projection on the boards power; however, the chips consume >98% of the power.
...

BULLSHIT. Power supplies have a 10-20% loss, and cooling is usually calculated at about 40% of the power usage. 1.7MW of chips will actually draw closer to 2.5MW

[klondike_bar]

Weekly Update 2/5/2013

This last week has been very busy with our Project manager and engineers working with eASIC on our 28nm custom chip.  We have receive the first results back and the chip is expected to run at 30 GH/s, we are working on improving this number.  Also, we have received the quote on our PCIe board and our project manager and engineers are meeting with eASIC and our board engineers this week to keep our project on track.  We are also meeting with UMC to finalize our account setup.  We are also thinking about going to a 20nm custom chip and skipping the 28nm, if we can get it produced fast enough.

totally meaningless drivel. switching processes increases risk and cost, and changes some of the fundamentals of electronics engineering.

to say 30GH on a chip is also meaningless. KNC, Cointerra, Hashfast all created big-die chips that are capable of 10x that per chip, but use less chips. The important number is how many GH in how small a space

[kleeck]

28nm chip running at 30GH/s?  

Is it just me or is that a little uncompetitive now, let alone in Q3/Q4 when we will get our hands on it....?

Who designed this chip? Can't we run more hashing cores in parallel? KnC managed to get 150+GH/s out of their chips back in October 2013

Screw going down to 20nm - i think you should sort out the RTL code on the 28nm chip before you do anything else!

KnC is using 4 dies in one package, so our chip is very close to what they have done.  Our chip with 4 dies would be 120 GH/s.  We are working on increasing the number of cores.

[keepinithamsta]

Weekly Update 2/5/2013

This last week has been very busy with our Project manager and engineers working with eASIC on our 28nm custom chip.  We have receive the first results back and the chip is expected to run at 30 GH/s, we are working on improving this number.  Also, we have received the quote on our PCIe board and our project manager and engineers are meeting with eASIC and our board engineers this week to keep our project on track.  We are also meeting with UMC to finalize our account setup.  We are also thinking about going to a 20nm custom chip and skipping the 28nm, if we can get it produced fast enough.

totally meaningless drivel. switching processes increases risk and cost, and changes some of the fundamentals of electronics engineering.

to say 30GH on a chip is also meaningless. KNC, Cointerra, Hashfast all created big-die chips that are capable of 10x that per chip, but use less chips. The important number is how many GH in how small a space

That's why it's kind of laughable how much real estate ActiveMining needs for 10TH/s at the moment.  KNC's Neptune will run 3TH/s out of 2.6 cubic feet, rounded up slightly.

I can't be the only person that started scratching my head when it's announced that 10.488TH/s takes up 2 42u racks according to Ken.  I know each of my racks takes up between 45-50 cubic feet depending on which rack we're talking about.  I have 2 empty racks at the moment but why would I fill them up with 10.488TH/s when I can just buy three Neptunes that I could literally put on my desk?

Regardless, all this information is useless until there's less mindless announcement being spouted out and real chips being produced with real world figures.

[DeadwoodDan]

My earlier post got deleted from this self-moderated thread somehow, let's see if this gets deleted too. Ken, you owe us an answer.

Can I ask again about those millions of dollars we have in preorders?  Are we still holding the money?  Are the customers all cool with it?

Also, are our own miners included in the figure for preorders?  I'm a bit uncertain on how a virtual entity representing both itself and its profits works when the company owning the virtual entity is based in Belize and the virtual entity is doing business in the United States of America.  Would the virtual entity preorder miners from its and its profits' owner?

[RickJamesBTC]

My earlier post got deleted from this self-moderated thread somehow, let's see if this gets deleted too. Ken, you owe us an answer.

Can I ask again about those millions of dollars we have in preorders?  Are we still holding the money?  Are the customers all cool with it?

Also, are our own miners included in the figure for preorders?  I'm a bit uncertain on how a virtual entity representing both itself and its profits works when the company owning the virtual entity is based in Belize and the virtual entity is doing business in the United States of America.  Would the virtual entity preorder miners from its and its profits' owner?

Why do people keep saying "we" There is no "we" you are not partial owners in this company, you "maybe" own some unregulated, unknown, untradeable shares in something, but since so many of you people were so excited to throw your money at bitcoin to try and get rich, you didn't get any details about these questions up front. You basically own nothing, and until you realize that and start demanding your money back, nothing will be exactly what you get.

[arousedrhino]

Weekly Update 2/5/2013

This last week has been very busy with our Project manager and engineers working with eASIC on our 28nm custom chip.  We have receive the first results back and the chip is expected to run at 30 GH/s, we are working on improving this number.  Also, we have received the quote on our PCIe board and our project manager and engineers are meeting with eASIC and our board engineers this week to keep our project on track.  We are also meeting with UMC to finalize our account setup.  We are also thinking about going to a 20nm custom chip and skipping the 28nm, if we can get it produced fast enough.

totally meaningless drivel. switching processes increases risk and cost, and changes some of the fundamentals of electronics engineering.

to say 30GH on a chip is also meaningless. KNC, Cointerra, Hashfast all created big-die chips that are capable of 10x that per chip, but use less chips. The important number is how many GH in how small a space

That's why it's kind of laughable how much real estate ActiveMining needs for 10TH/s at the moment.  KNC's Neptune (Shipping in late Q2 possibly Q3, they already have a backup plan if they miss the Q2 deadline) will run 3TH/s+ (probably at least around 3.5TH/s given the latest news) out of 2.6 cubic feet, rounded up slightly.

I can't be the only person that started scratching my head when it's announced that 10.488TH/s takes up 2 42u racks according to Ken.  I know each of my racks takes up between 45-50 cubic feet depending on which rack we're talking about.  I have 2 empty racks at the moment but why would I fill them up with 10.488TH/s when I can just buy three Jupiters that I could literally put on my desk?

Regardless, all this information is useless until there's less mindless announcement being spouted out and real chips being produced with real world figures.

Corrected in bold^^

[SoylentCreek]

We are also thinking about going to a 20nm custom chip and skipping the 28nm, if we can get it produced fast enough.

No... Just no. Do you honestly hear yourself, Ken?  One, has eASIC even publicly announced that they are capable of making a 20nm chip in the near future?  Stick with 28, let's mine the shit out of them, and THEN we can look into expanding into 20nm. 

[VolanicEruptor]

"skipping 28nm" completely is just more evidence that there was no 28nm to begin with.

[drawingthesun]

I want to trade my shares please.

[eikzbtc]

Omg , please don't go 20nm , finish 28nm first.

[VinceSamios]

I can't be the only person that started scratching my head when it's announced that 10.488TH/s takes up 2 42u racks according to Ken.

Considering ActM has plans to host close to 50TH/s in the same racks within months, I'm not scratching my head at all. Why the hell do you care about hashing density? It's one of the least important factors.

Most important: Time to market/rollout speed
Second most important: Energy Efficiency
Third most important: Energy costs
Fourth most important: Hashing density

[JimmyJazz]

Weekly Update 2/5/2013

First off Ken for the last few updates you have had the wrong date, but now its taking the biscuit. It's 2014 now - maybe your confused because we are meant to be hashing.

Second don't start changing plans it will cause a mass fuck up.

And last but no least let me trade and get out before the stock dies, I was going to buy more but now I see the light and want out at a small loss at best!

[sparky999]

So let's get this right, you couldn't get a 28nm chip on time. Your full custom 28nm performs worse than KNC's released last year and now you want to go to 20nm something that nobody has done successfully yet.

Ken you have failed to meet every single deadline you have set that was important to shareholders - try fulfilling some promises before you go down the 20nm route.

[drawingthesun]

Can we stop focusing so much on how fast the chips are? Someone else said it best. The most important thing is time to hash and the price per GH/S.

Ken, 30gh is poor but what's the cost per GH?


also please let us trade the damn shares. COME ON!

[VinceSamios]

This 20nm malarkey,

The benefit is not enough to justify the delay, cost and risk.

Ken - don't do it.

[wasubii]

For those who dont know; important info over here:

https://bitcointalk.org/index.php?topic=297530.msg4965446#msg4965446

[zumzero]

If there is the remotest chance that 20nm chips could not be produced within the same time frame as the 28nm and if Ken is unable to give an absolute guarantee supported by hard facts and figures that he could pull it off, then in my opinion, pursuing the 20nm at this stage would be a grave error for the company.

There are people out there who didn't buy Bitcoins when they were $30 because they decided to wait for a better deal.  The same people then waited when the price hit $270 and then $600, and so on.  I can see similarities with ActM having not produced the original chips and now talking about skipping the full custom in favour of 20nm.

[runam0k]

I want to trade my shares please.

By "trade" you mean "sell" and yes, everyone wants the same thing.