Topic: [Guide] Dogie's Comprehensive ASICMiner Blade Setup - page 26. (Read 580761 times)

these are located in a server room with all high end equipment. I can't fathom stuff that can handle mega traffic would trip up on this stuff... but its gotta be something...
I have both going to the same worker, maybe slush's auto diff thing is setting the diff too high...
I am gunna try changing to 2 workers

I had similar issues with 1 Blade.  After fiddling with everything, someone informed me that my Dlink router was garbage and could have been the issue.  I upgraded to new Belkin Dual band, Miner has been stable for weeks now. No random disconnects or proxy changes.

Looking for some Ideas
Gen 2 blade GREEN issues:
I am now more perplexed as to what is the issue with my blades.
I now have 2 blades, both are using power supplies with well more than the 8.3amps on 12v that is needed triple the amount needed in fact.
I pointed 1 to a bfg proxy and 1 to slush's mining proxy program but neither will exceed 7.2gh/s.
If it were network traffic issues adding the 2nd blade in "theory" should have dropped both?!
They are in the same network.
They both insist on changing proxy servers about every 40 work units requested.
You can not put a blank or the same ip for both servers or they will stop requesting work after 40 until they are rebooted.

I put a bfg proxy on a server and pointed them to 2 ip's on the server which works. but they do not exceed 7.2gh/s

I setup another proxy on another server and pointed the other blade at it. still will not exceed 7.2gh/s.
I changed one of the proxys to the slush miner and the other to bfg, still wont exceed 7.2gh/s

Neither are running hot, they have more than enough cooling on them.
in BFGMiner the first column says 10ish/7.2/7.0gh/s
What does the first column mean?


Total MHS:   07318
Received:   0000257291
Accepted:   0000203803
Per Minute:   100.00
Efficiency:   079.21%
Up Time:   1d,09h,57m,53s

Current Server: 10.1.3.144:8334
Chip: OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO


Total MHS:   07217
Received:   0000259925
Accepted:   0000202522
Per Minute:   098.61
Efficiency:   077.91%
Up Time:   1d,10h,13m,39s

Current Server: 10.1.3.72:8333
Chip: OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Make that 8 blades * 120 watts (at peak) = 960 watts.... woopsie.  Either way, still looks like it's unwise to populate that back plane past 8 Gen2 boards on that recomended HP rack-mount PSU.  Any thoughts?

H@shKraker

xstr8guy,

Thanks for the info.  The HP rack mount server PSU I'm using (which is the recommended one) says to not exceed 900watts at 120VAC.  If I do my math correct that would indicate that a backplane with that PSU can only support 8 blades in any case (8 blades * 1120watts).  Is that why you only have 8 in your back plane OR do you plan to add two more?

H@shKraker

I have a gen 2 backplane with 8 blades so obviously I can't separate them.  I have 2-140mm fans pushing air from one side and 2 fans pulling air from the other side.  And I don't have any overheating problems.

The heatsinks do a great job as long as you have adequate airflow so the blades don't just bake in their own heat.

Yea, was what I was thinking ..... populate every other slot to allow for air flow.  Thx for the feedback.

H@shKraker.

I only have gen1 blades, so no personal experience with the new ones.

If it were me, regardless of spacing, I would definitely have fans blowing top->down on the blades. This ensures cooling. Most definitely fans if they're not spaced. Spacing is always a good thing, to keep the hot air buildup to a minimum.

vesperwillow,

Speaking of heat issues, I have a Gen2 power backplane with two of the gen2 blades plugged into it.  I've got them separated by a slot.  That seems to allow the blades to shed heat well.  The thing is that since there are 10 slots in that backplane it seems like one *should* be able to populate every slot.  I've noticed the heat sink on those Gen2 blades gets hella hot.  If I *did* populate all 10 slots do you think there would be enough air flow spacing to shed the heat effectively *OR* would it be smarter for me to only populate every other slot?

H@shKraker

I thought of that. I figured it won't be too much of an issue considering the test will only be for about a minute, and I'm going to wrap it around the rack (not right around the blade) which allows for a couple minutes before the thermal issue will be a problem.

High CFM fans with 30*F ambient air keep the blades plenty frozen lol.

vesperwillow,

Foil will work but it might also hold in heat because it's a continuous membrane (whereas screen isn't a continuous membrane).  It should work fine in regards to RF shielding so long as the foil is connected to the PSU ground *BUT* you might notice heat build and as a  result, hashing efficiency drop-off.

H@shKraker

I've had several phones within a meter of my blades, including one or 2 hosting wifi, and there was never a noticeable issue. Perhaps on a case by case basis, or orientation?

I have some shielding sitting around (aluminum foil), I might actually try this for giggles and report back the data.

Yes, i'm lazy as fuck 

And to keep it profitable, i don't invest more money than necessary...

Beutelschneider,

If you *REALLY* wanted to test out whether it's a RF shielding problem you could make yourself a Faraday shield and put your blades inside of it.  The way I'd do it is I'd head over to the hardware store and get to unpainted, uncoated window screen and some wire.  I'd get enough to make a dome or at least an enclosed cover over my blades.  Make sure there are no gaps in the shield.  Finally I'd solder a wire from the cage to the ground on the power supply.  The reason you go to the ground of the same power supply powering your blades is 1) it easy and 2 it's the same ground that serves as the return path for the blades' electrical supply.  This will in effect shield anything inside from RF on the outside and it will shield anything on the outside from RF originating from the blades.  For reference take a read at http://en.wikipedia.org/wiki/Faraday_cage.  If you've got an "RF is getting to the blades and F-ing them up" problem eth Faraday enclosure will nip that in the bud for sure.

H@shKraker

I'm using a HTC DHD, and there is no real shielding when you see the bare electronic parts on the PCB. By now, my phone with enabled WLAN is only 50cm (about 2ft) away from the blades and they still run smooth. So i'm a bit uncertain if its the non-existent shielding of the blades or the fact they had to connect via my WLAN router to the proxy. As i said before, all stuff is connected by cables now.

This is so weird. Have two V2 Blades that had been working fine for a couple months, then recently both seemed to stop working. All chips showing enabled but they just weren't receiving and accepting many shares (would submit some I think they were hashing at a couple hundred MH/s or so) and kept rebooting. Tried resetting and power cycling the Blades, restarting Stratum proxy, etc. with no luck.

Then got on here and saw your post. My HTC Thunderbolt being connected to my home WiFi ended up being the cause (I very rarely get on my home network with it, usually just use it on the 4G network, but last night connected it to the WiFi to transfer some files). Turned WiFi on the phone back off and Blades started running and hashing away at ~10.9GH/s again. Really, really weird. These things must have really poor shielding or something. But just wanted to say thanks for posting this and wanted to post my experiences, may help others having a similar problem.

Greetings folks. I have a number of BE blades up, talking to a slush proxy and hashing against btcguild awesomely.  In my population of blades I have two Gen2 blades in a power back-plane with one of those HP server power supplies powering the pair.  This is cool and all except for the part where the Gen2 blades seem to make a TON more heat each than the older Gen1 blades did (and with much less spacing in the power back-plane).  I currently have my two Gen2 blades separated by one slot (i.e blade1 is in slot 1 and blade2 is in slot 3).  Even in this config there's a TON of heat.  If I were to fully populate that back-plane in all 10 slots (with that very close clearance between blades) would I be able to properly extract the heat with a 20" box fan blowing through the blades?  It seems to me that since those Gen2 heat sinks get hella uncomfortable hot that maybe I should only ever populate every other power slot.  What say ye?

H@shKraker

^-- Good thought.

BFG wouldn't be able to bind the port if bitcoin or another client had the port open, and would throw an error. That's why it's important to make sure you see the http-listening line when BFG fires up.

If you are running a bitcoin client on the same machine which uses the default port, it is going to conflict.  I would use a different port than 8332.  Personally, I use 8330.

For simplicity I would make sure the blade and proxy are on the same subnet, as was mentioned above. If that's too much to do, put a computer on the same subnet as the blades and use it to test the proxy connection to see if you can actually reach the proxy from that subnet.