Topic: Official Thread: AMT - page 243. (Read 678353 times)

Watch the video again. The pool auto-adjusts the difficulty. When they reach stable performance of about 209 GH/s the difficulty is 512.

Not really. Industrial psu is needed plus possible design issues. 6 module design is not option in their current stage of development is what I think

So basically we are talking for paid share and reject rate
I am so stupid please excuse me. The point was about hash rate shown by cgminer right?
So like a resume cgminer will show same. The paid shares rate will be different and I do agree that mining with higher difficulty do all things you mention. And mining at higher diff is always better for sure. But it is not making your miner from 1t to 1.2t if you follow my thoughts. What will make your miner more powerful is higher voltage and higher chip clock which hit power consumption badly

Well if Bitcoin.ch advertised their units at 1 THs systems and are shipping 5 modules, then I do expect AMT to ship with 6 modules at 1.2 THs.

Will this break the bank?  At $90 per chip x 48 = $4,320 add $500 for everything else... $4820... list price $5999... still enough room to make decent profit.

AMT will need to make a 48 chip system profitable at $5,999.   Cointerra is already at $5,999 and is advertised at at least 1.6 THs.   

Whoa... impressive wisdom here on Bitcoin mining.   Saved for posterity.

Obviously you don't mine with an army, like I have...

You submit 10,000,000 diff-1 shares and tell me your miners don't run slower submitting crap...
You put the work-load higher on big miners, and put the "submit shares" to 256 for THs miners, not 16-diff, which slows down the network, sending millions of share-hashes. Sending 1,000 a second is too much, stops CGminer from actually processing blocks, lowering your average. That is why they changed the pools to auto-adapt to flooding of crap.

Want a tip... if you are in a pool... use small shares with small miners and smaller workload sizes of 64. You finish work faster, and submit a few more shares than the guy with a 256 workload size. He is still "processing" while you have submitted 4 shares. When the next block arrives, or your solution, you have already submitted, and his 256 workload still processing gets rejected for being late, by the time it actually submits. However, you will be considered a flooding miner, and big pools will auto-adjust to higher shares if you have too many small miners.

For a THs miner, that is death, submitting too many orders too fast. You got speed, you need those bigger orders to stop from flooding the threads and the miner and the network and the wallet and the pools processing all those hashes to confirm they are valid before accepting them as "credit". Networks buffer like crazy, you will lose more credits and "found blocks" due to your ISP buffering the hell out of your puked data.

I no longer mine on pools... I like getting my actual blocks and tx-rewards, and don't like having my hard work rejected because someone found a block, and so it throws-out your hard worked shares, as if you didn't just work them. Pools are a scam. For some, it is the only way to mine. I choose lower diff coins I can solo-mine, more rewarding in the end.

I do not know if you are a paid amt spokesmen with lack of knowledge or just a troll but among all stupid things posted the undersized share and diff 64 made May Day. As native English speaker you should be aware of the luck and variation meaning.
In short they used diff 1 or undersized share as you call just  to avoid the variation and show the performance for a minute. With 64 share it will be the same but more time is needed. I do not have free time to comment all by one all of your bright and shiny thoughts. Please take my apology for that
 

Let me see if I can put some reality into this...

The unit with 40 chips, (5x boards), shows about 212GHs average per board running an undersized share/hash. So, If running with the appropriate diff-rate of 64, which that unit should be doing, I'll say it runs closer to the peaks I saw of just under 222GHs per board. Total, that is 5x 222GHs = 1110 Peak (Which is what a company advertises as a units "Ability" or "Potential". Not to mention that it takes about 30 min of mining to settle-in to a decent average, which is higher than start-up averages and peaks.)

The power consumed read as, 1059-1061 Watts, "from/at the wall". (Since power does not move "to the wall".)

They also admit that the "unit will run with lower wattage", once they actually optimize it. (Obviously, this is just a RAW setup, which has not been tuned, it was just made to function for the demo.) For now, lets assume they don't get any better, with 40 chips, running at an unspecified operating voltage. You don't know if those are slightly voltage-bumped, which I assume they are, by the specs of the units and chips.

The units being sold to us are 1.2THs units. (Still does not specify if that is a maximum potential, or normal operation potential, but I will assume "normal for the unit, at the stock settings of the unit". Also I will assume that is a peak potential for that mode. Eg, ideal cooling and power delivery of your home. Thus, potential. Individual use may vary, obviously.)

So this would require more than 40 chips, as 40 is only what is required for the 1.0 advertised units. With each card being roughly 200GHs (222GHs peak, at non-optimized settings.) This would be 6 cards, which is 48 chips.

6 cards (48 chips), running non-optimized deliver 222GHs peak (at an unspecified operating voltage). This is a total of 1,332GHs or 1.332THs peak, in normal mode. (Not Max, because that would be "turbo", and that is not how these are being setup.)

Voltages are variable, and settable by steps, so how AMT sets-up the voltage as "normal", will ultimately matter. However, I will assume, rightfully, that the video showed just the chips running at the "normal" chip-voltage.

Quick recap of the 5x units values...
Power-saver: 750GHs {est: 400W}
Normal-mode: 1000GHs {act: 1060W} (1110GHs peak shown)
Turbo-mode: 1500GHs {est: 1875W}

Translate up to the 6x cards (48 chips) AMT's minimum possible chip requirement
Power-saver: 900GHs {est: 480W}
AMT-tuned: 1200GHs {est: 990W} [900W + 10%] <- voltage below "chips normal mode".
Normal-mode: 1200GHs {act: 1272W} (1332GHs peak shown)
Turbo-mode: 1800GHs {est: 2250W}

Now, provided AMT uses this Minimum possible chip design, and does not add more chips... (More chips would allow them to run at lower voltages, delivering more GHs per the same wattage.)

900W, the "estimated" power consumption, plus 10% = 990W (That is the "complete adjusted estimate".)
1200GHs, the "estimated" peak production ability, minus 10% = 1080GHs (That is the "complete adjusted estimate".)

By those estimates, the projected 6x (48 chips), would operate within specs, slightly below the "chips" "normal voltage". Which, for the AMT 1.2THs miner, would be the units "normal operating voltage", or "stock setting".

It would only take about 60 chips, operating at lower voltages (but not power-saver low), to produce the desired 1.2THs at 600W, but I am sure the 600W +10% (660W) is the consumption in power-saver mode, which should be about 900GHs. (480W estimated by the producers of the chip, for their completed units. Which would be about 600W if they used 60 chips, instead of 48, and produce 1.2THs, not 900GHs.)

So... If this is only the 48 chip design (6x boards), that allows me to push the unit to 1.8GHs potentially, by adding one more PSU. Nice... if it is 60 (doubt that), wow, that would be about 2.4GHs... Actually, that makes 60 chips sound like a more realistic design. (Still doubt I will get 60 chips though. )

40 chips... Not... that is not the system they sold us, 40 chips is the 1.0GHs unit. That is why I would have never gotten the unit from coincraft, besides having to pay crazy VAT for delivery.

I don't see these units being impossible to fall into the specs. My view has not changed. The chips have exceeded the design specs themselves. So should most of the units.

Hope AMT is generous with the first production of non-optimized and heavily-populated boards. (Would like to know what the chip-count would be for the units still... Anything over 48 would be freaking great! But 48 would be plenty still. 1.8THs potential from a 1.2THs machine, and I am happy.)

1. Congrats gys I am impressed -no irony
2. 1068x1.2 is around 1.3 KW at the wall doubling projected 600w min consuption and with 44% over 900w
3 the bet that I was about to make was for 900 w at the wall so I still feel winner here. Unfortunately no one including the guy with jockes and even you did not take it. I am still wandering why? That we're quickest 10k usd you were going to make or I am mistaken?
4. I have some doubts about performance stability and espesialy if it will mine stable for more than one minute after cgminer startup
5. I would be glad to make an review with miner in hand. I know that there is no free lunch so I would like to know what shall I pay in btc and when I will be getting my miner
6. A video of stab ale 1th after 24 hours of mining will convince me that job is done well
Good luck to all of you
Ps:
After carefully watching video I can tell you that:
1. the 5 modules are carefully chosen with 1 or two dead cores per module. Nothing bad about that in reality one dead core equals to 1 GH and common thing is that you gona loose almost one chip or 20 -25 gH for 40 chip setup. Not your fault for sure. The chips are as they are
2. The interesting part for me is why 5 th module is underperforming so badly? Some technical difficulties when adding more modules? Flaw in design?
3. You know that turbo mode is a fiction and too power hungry and it will blast away your projected 1.3 kW estimations right?
4 if you add 6 th module how badly it will underperform? Let me put it other way does the design allows more modules to be added at all?
5. What psu are you going to use to provide at least 100A/12v (110a with spare)
Thanks for reading.

I'm assuming it is 40 chips.

How many chips on an AMT 1200?

Loshia did you see it, did you see the video update, here it is.. http://bitmine.ch/?p=5176  did you see the link?  it's this link loshia http://bitmine.ch/?p=5176 .  1060W.. so is .9 so impossible? is it..

Ah, the swiss posted their video..  came in at 1060W on 40 chips - TO THE WALL!

Review?  I can do a review too.   Website is http://www.legitcoinage.com

Loshia stop making problems. I said I liked the member that offered that bet, that it was funny, to give an over under on a wattage spec and then parlay it with a hashrate. He made a joke and it was humorous.

Wattage is 0.9 nominal as projected. Loshia, we use a different board in this miner than the one board you've had experience with thus far. You of all people should know that the board makes the difference.  If you'd like, we can have you review our miner as well. Would that be something you'd like to do Loshia?

Bitmine.ch now shipping out of warehouse!



Okay, AMT,  can we expect the same?

Now for some good news:


5 modules...  8 chips per module... 40 chips per unit at 1 TH/s?  1060W?  So is AMT going to run a turbo mode to hit 1.2 THs?

I do not know in my opinion no but they are getting close +20% If they put 30% more chips yes it is possible but accourate figure in my appinion is arrond 2

So the AMT official says it is at 1680 watts for 1.2 THs miner?  Is that correct?

So is 1680 watts possible here?

Dude,
They just admitted the fact that 900 is fiction and made first steps to the real info 1680 WATS. I just do not get your comment as long as i quoted AMT_miners which supposed to be official AMT account. Did i miss something?

Maybe a wager is called for here! AMT is claiming a much better wattage than what loshia claims is possible.   

WOW...Misleading info? Be prepared guys 1.1W + .3W = 1.3Wx1200GH = 1680 WATS! What about 600-900W figures? Let us make them 2KW right?

And then we can talk about whats "at the wall". ? What do you actually producing at full speed WALLS or what? We can talk in "the wall also"
This is getting ridiculousness guys it seems like top secret NASA project dealing with walls