Black Arrow 28nm 100Ghash Bitcoin ASIC from $0.49/GH/s

[ernie-]

Speaking as someone who has their products in hand, no they dont. They get effective about 1.1-1.2w/gh.

Rockminer is not a great example, they run their core VDD at 0.72v whereas the BE200 can go as low a 0.55v.

[tempestb]

BFL's Monarch is really good on power, but you know...  you'd have better chances of finding a leprechaun's pot of gold than ever getting your hands on one.

[RoadStress]

Speaking as someone who has their products in hand, no they dont. They get effective about 1.1-1.2w/gh.

Rockminer is not a great example, they run their core VDD at 0.72v whereas the BE200 can go as low a 0.55v.

What's the W/GH ratio for end system at 0.55v?

[ernie-]

Speaking as someone who has their products in hand, no they dont. They get effective about 1.1-1.2w/gh.

Rockminer is not a great example, they run their core VDD at 0.72v whereas the BE200 can go as low a 0.55v.

What's the W/GH ratio for end system at 0.55v?

Who knows, they quote only the chip specs, have to wait until someone under clocks a system properly.

[nexus99]

Power consumption numbers:

1200Mhz - 120-130gh - ~115-120w on my killawatt
1100Mhz - 112 - 118GH - ~107W
1000Mhz - 98 - 109GH - 98W

Have not tested 800 yet.

These results are the same as I am seeing.

I seem to gets lots more HW errors at 1000 MHz compared to 1200 mhz.

[y_boonstra]

Power consumption numbers:

1200Mhz - 120-130gh - ~115-120w on my killawatt
1100Mhz - 112 - 118GH - ~107W
1000Mhz - 98 - 109GH - 98W

Have not tested 800 yet.

Under 1w/gh. Not bad. Not bad at all!

under 1w/gh AT THE WALL (not at chip level) is pretty awesome

.90 watts per GH,not bad.....just not what they were shooting for  

I don't think anyone cept Bitfury came close to predicted specs..............Even ole Friedcat missed by quite abit!!!!! On the second round 

The chip is considerably more efficient in the X3, the X1 has ~15-20w in overhead of the LCD+ controller, where as each X1 has 20w of overhead per 1th.

I dont believe the costum controle and LCD consumes about 15 - 20W no way, they would heel hot to the touch with those amount of power consumption. !!!!!!

The fact remains that the units are overpriced considering W/GHs. + 5 months to late..
They promissed 0,55W/GHs S/S at chip level, thats why they delayed it in the first time (Feb) !! and what do we get.. a unit thats hardly any better than a undervolted antminer ..!

let not even bring the antminer S3 up..

Like many others im screwed to i got a couple of those X1 comming my way direclty from BA.. still dont know when.. maybe NEXT week :S :S

[n00dles0up]

Having been away from these forums for some time now planning my wedding for this August. Can anyone catch me up and summarise any important info that may have been posted? I have been seeing minersource on facebook.

Just curious about delivery times, if anyone in the Uk has got theirs? Compensation? What about customs values etc? Anything worth knowing?

Thanks

[david105396]

Having been away from these forums for some time now planning my wedding for this August. Can anyone catch me up and summarise any important info that may have been posted? I have been seeing minersource on facebook.

Just curious about delivery times, if anyone in the Uk has got theirs? Compensation? What about customs values etc? Anything worth knowing?

Thanks

This should help: http://ecointalk.net/topic/603-weekly-production-update/.
Along with minersource has received the first shipment of X-1's.

[TheMinex]

tempestb I really doubt that BFL which produce the first melting PCIE board in the world can ever produce energy efficient HW.

Speaking about X1 ; david said that heatsink is not so hot at the touch, and all people read the internal chip temperature near 80°C. Heatsink with heatpipes should be very efficient to transfer heat, so deltaT between chip & exaust air shoul be very low. Is termal paste applied correctly???

Reading Bob's mesures what I think is this:

1 - More than 120Gh @ under 120W is good, cause freq can be adjust between 1200 Mhz and 800 Mhz, it seems like BA choose to deliver all hardware OC by default (correct me if I worng).

2 - If we get less than 1 W/Gh in max freq state, there must be room to improve it in lower speed states, but doing it for the moment dosen't seems to increase efficiency significantly.

3 - Possible cause. Chip + power regulators have better dynamic power dissipation than static power dissipation which is good, and this could be why errors increase lowering frequency ( makeing this, chip is forced to pass more time where it dosen't like - especially if it is hot )

Fixes - try anything possible to cool more chip ( this will be good for efficiency & stability at any state ).
        - control voltages (Bob said on ecoin that this will be possible, I suggest david to write this in firsts order on the work to do list for programmers, cause achieving for example 0,7 W/Gh will seriously impove BA situation ).

Bob (or anyone else) , is there any drop in temperatures when you run @ 1000Mhz or lower

[TheMinex]

( lower voltages should decrease drastically static power dissipation while decrease only partially efficiency in dinamic power dissipation where the chip seems good )

[bobsag3]

tempestb I really doubt that BFL which produce the first melting PCIE board in the world can ever produce energy efficient HW.

Speaking about X1 ; david said that heatsink is not so hot at the touch, and all people read the internal chip temperature near 80°C. Heatsink with heatpipes should be very efficient to transfer heat, so deltaT between chip & exaust air shoul be very low. Is termal paste applied correctly???

Reading Bob's mesures what I think is this:

1 - More than 120Gh @ under 120W is good, cause freq can be adjust between 1200 Mhz and 800 Mhz, it seems like BA choose to deliver all hardware OC by default (correct me if I worng).

2 - If we get less than 1 W/Gh in max freq state, there must be room to improve it in lower speed states, but doing it for the moment dosen't seems to increase efficiency significantly.

3 - Possible cause. Chip + power regulators have better dynamic power dissipation than static power dissipation which is good, and this could be why errors increase lowering frequency ( makeing this, chip is forced to pass more time where it dosen't like - especially if it is hot )

Fixes - try anything possible to cool more chip ( this will be good for efficiency & stability at any state ).
        - control voltages (Bob said on ecoin that this will be possible, I suggest david to write this in firsts order on the work to do list for programmers, cause achieving for example 0,7 W/Gh will seriously impove BA situation ).

Bob (or anyone else) , is there any drop in temperatures when you run @ 1000Mhz or lower

I dont think the software based temp sensors, or their software is not working. Our X1s report either 0, 60 or 80c, nothing else.

Clocking to 800mhz only drops the speed, does not seem to impact power consumption all that much. Seems the sweet spot is 1000-1100mhz.

[ninjaboon]

all this talk and BA is still behind on shipping.

[CryptoGuy]

Any word on the power cords bobsag? They should have arrived right behind the miners if they shipped within a day or two from them... or was this one of the "We'll ship them next week" situations with the cords? I seriously don't understand this delay. People should be getting their tracking numbers today, you've had a day to test them now.

[bobsag3]

Any word on the power cords bobsag? They should have arrived right behind the miners if they shipped within a day or two from them... or was this one of the "We'll ship them next week" situations with the cords? I seriously don't understand this delay. People should be getting their tracking numbers today, you've had a day to test them now.

People got tracking numbers yesturday for all the units we had to ship, I only have 1 unit left here for testing and writing the guide.

I have not heard back about the power cords, so I purchased 500 US- Cords for my units on Monday, so I will have more than enough US cords for all my Batch 1 orders at least, with plenty time to order more cables if needed.

[sorehammer]

Bobsag3 so seeing as I didn't get tracking info when will the next batch arrive to you.

[bobsag3]

Bobsag3 so seeing as I didn't get tracking info when will the next batch arrive to you.

I received a reply to this from Bobsag, and was told that units would be sent out Friday/Monday (more than likely Monday is my guess).

It will be Monday most likely, and shipment will contain at least all of Batch 1.

[sorehammer]

Bobsag3 so seeing as I didn't get tracking info when will the next batch arrive to you.

I received a reply to this from Bobsag, and was told that units would be sent out Friday/Monday (more than likely Monday is my guess).

It will be Monday most likely, and shipment will contain at least all of Batch 1.

Cheers but will that include the lcds and uk plugs also or are they still waiting for the lcds.

[TheMinex]

tempestb I really doubt that BFL which produce the first melting PCIE board in the world can ever produce energy efficient HW.

Speaking about X1 ; david said that heatsink is not so hot at the touch, and all people read the internal chip temperature near 80°C. Heatsink with heatpipes should be very efficient to transfer heat, so deltaT between chip & exaust air shoul be very low. Is termal paste applied correctly???

Reading Bob's mesures what I think is this:

1 - More than 120Gh @ under 120W is good, cause freq can be adjust between 1200 Mhz and 800 Mhz, it seems like BA choose to deliver all hardware OC by default (correct me if I worng).

2 - If we get less than 1 W/Gh in max freq state, there must be room to improve it in lower speed states, but doing it for the moment dosen't seems to increase efficiency significantly.

3 - Possible cause. Chip + power regulators have better dynamic power dissipation than static power dissipation which is good, and this could be why errors increase lowering frequency ( makeing this, chip is forced to pass more time where it dosen't like - especially if it is hot )

Fixes - try anything possible to cool more chip ( this will be good for efficiency & stability at any state ).
        - control voltages (Bob said on ecoin that this will be possible, I suggest david to write this in firsts order on the work to do list for programmers, cause achieving for example 0,7 W/Gh will seriously impove BA situation ).

Bob (or anyone else) , is there any drop in temperatures when you run @ 1000Mhz or lower

I dont think the software based temp sensors, or their software is not working. Our X1s report either 0, 60 or 80c, nothing else.

Clocking to 800mhz only drops the speed, does not seem to impact power consumption all that much. Seems the sweet spot is 1000-1100mhz.

I didn't said sw is not working, only suggest BA implement voltage controls as fast as they can (I see temperatures only HW related)   .  The fact that 800Mhz drops only speed, fully support my thesis.
Would be a shame we can't acieve the better energy consumption only for those little details  

Sry ninjaboon but I see this as a general X1 topic and I don't want to make bob and david start to loose time jumping along btctalk. Those detalis could be usefull of every X1 future older too.  

[sorehammer]

Well hopefully some one hacks the firmware and maybe improves it.

[bobsag3]

Bobsag3 so seeing as I didn't get tracking info when will the next batch arrive to you.

I received a reply to this from Bobsag, and was told that units would be sent out Friday/Monday (more than likely Monday is my guess).

It will be Monday most likely, and shipment will contain at least all of Batch 1.

Cheers but will that include the lcds and uk plugs also or are they still waiting for the lcds.

No LCDs, possibly on the cords. Still waiting for an answer on them, I was supposed to get the cabled in another shipment separate from the units, but nothing yet.