[PENDING] ASICMiner Cubes. 30-38gh/s - USA - 1 BTC - Price protection!

[xoclutch]

Already tried that... Also am attempting to use BFGMINER proxy instead, however i'm getting the same results.
As soon as I click "CLOCK" it basically stop talking to the proxy instantly. 

[1714743633]

1714743633

[sbfree]

Ok crazyguy, I have shipped 1 unit back to you out of 4. And looks like I am not the only one with blades out of position, and/or No network ethernet working but powering on. Anyway....I will check my others just in case...but they are fine.....and would like credit towards another unit from upcoming batch. I will PM you the tracking info.
Thanks,
sbfree.

[sbfree]

My PSU: http://www.amazon.com/gp/product/B00ALK3KEM/ref=oh_details_o00_s00_i00?ie=UTF8&psc=1

That psu includes 2 cables, each with 2 pcie connectors. Make sure you are using a single connector from each cable. I'm not sure that 1 cable can handle the full load required by high clock.

just to throw my 2 cents, but if the PSU is any good one PCIe cable should handle the power to the cube....some video cards can consume 250-300w easily....so 1 cable should do, it work for me on 3 cubes.

[CrazyGuy]

My PSU: http://www.amazon.com/gp/product/B00ALK3KEM/ref=oh_details_o00_s00_i00?ie=UTF8&psc=1

That psu includes 2 cables, each with 2 pcie connectors. Make sure you are using a single connector from each cable. I'm not sure that 1 cable can handle the full load required by high clock.

just to throw my 2 cents, but if the PSU is any good one PCIe cable should handle the power to the cube....some video cards can consume 250-300w easily....so 1 cable should do, it work for me on 3 cubes.

Agreed, but my measurements of the cube were 324 watts at high clock setting. Some have measured it at 350. That's really pushing the limits of a single cable, unless it's thick gauge.

[herener]

Yea, I had five on high through this Kill-a-watt.  1718watts/5 is about 343.6watts each.  I would use both cables to be safe.  I did run two Cubes through one PSU for a bit and the PCIe power cable got much warmer than using a dedicated cable for each plug.

https://i.imgur.com/LmsYjG7.jpg?1

I'm also using the CX750M's on all Cubes.

https://i.imgur.com/MipJ8dq.jpg?1

I did have a problem with two cubes going all x's on me, had to switch out a PSU plug to another circuit.

-herener

[CrazyGuy]

I had a quad 5970 GPU rig ("The Minecart") back in the day running at about 1110 watts and plugged into a killawatt at all times. When I finally retired the machine, I realized the constant current had completely melted the plastic around the killawatt outlet. I'm always a little hesitent to go over 80% of any electrical component limits especially when running constant. Is that a 20 amp circuit you have those 5 cubes plugged into herener? Be careful out there...

[herener]

Yes it is a dedicated 20 amp circuit that had the killiwatt on it.  It was beeping from time to time as it was hitting the 15amp warning on the killiwatt.  Its no longer plugged in, I just wanted to check the draw from the cubes.

I have two cubes on a 15, another one on a different 15, and five plus the switch on the 20.

I am definitely keeping an eye on things, and thanks for the warning. 

-herener

[visdude]

My cube fired right up with nary a hitch and doing 32 GH/s at low clock.  However, it only does 2.5 GH/s more (34.5 GH/s) on high clock, drawing an additional 110WAC at the wall.  Whatever happened to the advertised minimum of 38 GH/s when OC'd?  Is AsicMiner doing a BFL nowadays?

To put it in perspective, 2.5 GH/s drawing 110WAC is a whopping 44WAC/GH (at the wall) or 37.4WDC/GH (with an 85%-efficient PSU).  Is it worth it?

BTW, thanks to CrazyGuy for an excellent service and packing.

[sidehack]

Pull it apart and make sure all your heatsinks are tight. Mine with the stock overclock topped out around 37.5GH, could have been higher if some of the cards cooled better. You can run the chip outputs through a quick script to find the average hashrate per card and see which ones are the turds.

[CrazyGuy]

Are you running it against bfgminer or slush's stratum proxy? I've found cubes/blades don't seem to run as well against the bfgminer implementation. Also, both proxies don't handle multiple units very well. Test out your cube on its own instance of slush's stratum proxy and see if you get different results.

[visdude]

Pull it apart and make sure all your heatsinks are tight. Mine with the stock overclock topped out around 37.5GH, could have been higher if some of the cards cooled better. You can run the chip outputs through a quick script to find the average hashrate per card and see which ones are the turds.

It's the first thing I did upon receiving it.  I also fixed half of the blades that were off the top notches.  Thanks.

[visdude]

Are you running it against bfgminer or slush's stratum proxy? I've found cubes/blades don't seem to run as well against the bfgminer implementation. Also, both proxies don't handle multiple units very well. Test out your cube on its own instance of slush's stratum proxy and see if you get different results.

Thanks.  I'll try Slush's next once I figure out how to connect through it using other pools (I am coding and CLI-challenged).  What port does slush's stratum proxy utilizes BTW?

Even if I could get it to 38 GH/s, I'm not sure if I'll be running it on high clock because that would still be around 18WAC/GH/s at the wall for the additional 6 GH/s; at least twice as much draw as the USB BEs.  However, I'm still determined to get it to work just to be able to say that my Cube lived up to the hype. 

[sidehack]

mining_proxy.exe -o yourstratumpool -p poolstratumport -gp localgetworkport

The proxy defaults to 8332, same as the default config on the Cube.

[visdude]

Thanks, sidehack.  I'll give it a shot.

[visdude]

You guys are right.  Slush's stratum proxy did get my Cube up to 37.9 GH/s on high clock.  Close enough.  Thanks again.

[AbiTxGroup]

You guys are right.  Slush's stratum proxy did get my Cube up to 37.9 GH/s on high clock.  Close enough.  Thanks again.

I started using the mining_proxy.exe a few days ago to use with my blades and I am very impressed.  I wanted to try it because of the "no midstate" (-nm) option.  I have been running it on my laptop and the blades efficiency are all running from 99.49% up to 99.71%.  I miss the miner stats using this proxy.  I use this laptop all day long to do my normal work and web viewing.

I have not tried running the proxy without the -nm switch to compare yet.  I have not figured out how to setup a backup pool with it yet.

Not to bash bfgminer any, as I use it for my BFL miners and my erupters, but as a proxy, my blades efficiency was 94-96%.  I could not find a compatible switch to use for the no midstate option. 

I did like being able to get the miner stats using bfgminer.  It is also very easy to us another pool for backup mining.

Both have their pros and cons.  Try them both and see which works best for you.

[visdude]

You guys are right.  Slush's stratum proxy did get my Cube up to 37.9 GH/s on high clock.  Close enough.  Thanks again.

I started using the mining_proxy.exe a few days ago to use with my blades and I am very impressed.  I wanted to try it because of the "no midstate" (-nm) option.  I have been running it on my laptop and the blades efficiency are all running from 99.49% up to 99.71%.  I miss the miner stats using this proxy.  I use this laptop all day long to do my normal work and web viewing.

I have not tried running the proxy without the -nm switch to compare yet.  I have not figured out how to setup a backup pool with it yet.

Not to bash bfgminer any, as I use it for my BFL miners and my erupters, but as a proxy, my blades efficiency was 94-96%.  I could not find a compatible switch to use for the no midstate option.  

I did like being able to get the miner stats using bfgminer.  It is also very easy to us another pool for backup mining.

Both have their pros and cons.  Try them both and see which works best for you.

Thanks for the tip.  For some reason, I only get up to 97% efficiency, even with Slush's proxy.  I think I could live with it tough.  BTW, what does the number labeled "Utility" right next to "Efficiency" represent?

[visdude]

I'm an admitted hardware geek and I like to take pictures of them.  I took some while taking apart and prepping my Cube.  I might as well share it.  I hope CrazyGuy doesn't mind:















...and hashing away in good company: 

[hayseed]

I'm an admitted hardware geek and I like to take pictures of them.  I took some while taking apart and prepping my Cube.  I might as well share it.  I hope CrazyGuy doesn't mind:

Thanks for the pic's !!!

Next time anyone has theirs apart I would like to see a pic straight on of the card tops meshing in the slots and of the bottom board again straight on in the slots of the bottom case.

I think that is where the problems are coming from. Asic miner put the cards in too tight in the slots. So any kind of pressure to the case top or bottom is putting pressure on the heat sinks and popping the cards/heatsink screws out. They probably should just flat pack the cards and let a little assembly be done on receipt or increase height of case and use a springy plastic grid retainer with groves to hold the card tops like better workstation class pc's have. Keeps the cards in their slot, but has enough give to accommodate shifts in the motherboard and or heat expansion.

my .0001

[visdude]

I'm an admitted hardware geek and I like to take pictures of them.  I took some while taking apart and prepping my Cube.  I might as well share it.  I hope CrazyGuy doesn't mind:

Thanks for the pic's !!!

Next time anyone has theirs apart I would like to see a pic straight on of the card tops meshing in the slots and of the bottom board again straight on in the slots of the bottom case.

I think that is where the problems are coming from. Asic miner put the cards in too tight in the slots. So any kind of pressure to the case top or bottom is putting pressure on the heat sinks and popping the cards/heatsink screws out. They probably should just flat pack the cards and let a little assembly be done on receipt or increase height of case and use a springy plastic grid retainer with groves to hold the card tops like better workstation class pc's have. Keeps the cards in their slot, but has enough give to accommodate shifts in the motherboard and or heat expansion.

my .0001

You're welcome.

Once the backplane and the blades are aligned to the notches/slots, they slide right in with hardly any effort; at least with mine anyway.  I think it's just an issue of shoddy QC.  I had to do some tightening of the plastic heat sink screws on the USB BEs too because about 80% of them were very loose.  Overall, I grade the Cube's build "A".  It's a nicely made and sturdy unit.  However, I would have to give the chips in it a "C".  Just imagine if they mounted at least 55nm chips on them.  ASICMiner has to get with the times.  But they're not gonna do it while they're still making money off the damn 130nm chips.  They're milking it.