Topic: [ANN] US/North American Bitfury sales NEW STOCK ***NOW SHIPPING*** - page 64. (Read 576772 times)

are you laying the M boards down directly onto a wire frame rack?  aren't you worried about an electrical short?

No, use sudo:

Going to get the second rig up tonight and start testing. Need to hit Frys to pick up another beefy power supply and some micro-sized screwdrivers.

 Was playing musical h-cards last night, moving faster cards to the front and slower cards to the back, and got one rig running along near 440gh overnight with one card going to 0gh.

 Will probably ask Dave to order an additional MBoard.

 More troubleshooting tonight... Thanks for all the tips !

  try to run only test on bad hboad alone if hboard isnt working that meant bad.

How I fixed my problems (two rigs stable 473 Gh and 517 Gh- with 1 dead h board on the slower one):

(I don't take credit for these fixes, some of them I found out on my own and some of them I read from other user's experiences.)

I needed to find a way to keep the cards stable, they come with crappy 1x PCI connectors and are really wobbly, especially when being blasted by fans (which is necc. to keep them cool and stable). The movement in the cards was causing the rig to slow down slowly over time until the affected cards eventually ended up at 0 Gh. To fix this I used two rows of electrical tape along the top of the cards to hold them in place better.

Cooling may or may not have been an issue. I added an extra box fan in front (two box fans now) and also two computer fans blowing air down on top of them for added air flow.

I was under the impression that two full rigs would run on one 1200 watt power supply. It turns out that seemed to be overloading the PSU (a nice gold rated ASUS). I ended up using a separate PSU for each rig and that got them both mining. I think this is the main reason why the one was getting 0 Gh and wouldn't mine. As soon as I put it on its own PSU it started working fine.

I put slower H cards in the last slot of each bank to keep them from interfering with other H cards. If you have one bank that is not working but all the rest are, this is likely the issue. If the card in the first slot of each bank is not seated properly, the rest of the bank will not work.

I hope this helps someone and saves them the 16 hours of their life that I just wasted. Now... I wish I would of realized all this before spending all night messing with firmwares... flashing sdcards... installing a different OS... trying BFGminer... trying cgminer... etc. :rolleyes:

My phone messed up while uploading, but a couple pics: http://imgur.com/a/9vDAm#apBFVy5

Consider checking to see if a better PSU gives you better results.  I had this problem and swapped a 1kW Tuniq PSU for a 1.2kW CoolerMaster.  They key is how much amperage you can push over the PCI-E power rails, not just the total wattage of the PSU.

so my miner stopped overnight inexplicably (charts indicate it unning stable between 76-79GH (2 cards) for 36hrs) and I rebooted it and charts indicate that caused it to work for about a minute or 2 then stop again.

using SSH command /opt/bitfury/start-miner.sh, i get:

cp: cannot create regular file `/run/shm/stat.json': Permission denied
cp: cannot create regular file `/run/shm/.chip.cnf': Permission denied

Is this an SD issue or something else?

To prevent confusion, the BFSB Euro M-boards had 4 SPI channels from the start, hence the possibility to test 4 H-boards in one go as described in my linked message.

It was impossible for me to run more than 13 H-boards on a old style v1 M-board. The 14th board will always report zero hash rates.
 I think it's an issue with the SPI clock signal. That is why new H-boards have 4 SPI channels.
 
check line 121 of spidevc.cpp

Here is my setup.  The clear tape creates a tension bridge so the box fan does not vibrate the h-cards as much.

Vibration of the h-cards causes a tremendous hash rate drop.  




and



Thanks to Goxed my 5, h-cards hash rate is 180.5GH/s
*EDIT now 181.534 GH/S


Bank 1   1: 34.002GH/s 2: 34.503GH/s 3: 0GH/s 4: 0GH/s Bank 2 5: 37.939GH/s 6: 0GH/s 7: 0GH/s 8: 0GH/s
Bank 3 9: 36.436GH/s 10: 0GH/s 11: 0GH/s 12: 0GH/s Bank 4 13: 37.653GH/s 14: 0GH/s 15: 0GH/s 16: 0GH/s


Cards #13 and #9 are the only cards fully populated with copper heatsinks.
I have managed to get one card to run 43-44gh/s for days straight, however that card is hot as hell when I clock it so high.

I also had problems with my BFSB old style M-board/H-boards, maybe this will help: https://bitcointalksearch.org/topic/m.3454967

It could be that your M-board has issues.
I bought an extra M-board and am now using it too.
None of my M-boards is filled with 16 H-boards.

H-boards that didn't went over 25GH/s got the pencil mod applied and are now doing better.
I have fans laying on top of the H-boards, blowing down. The fans keep the boards from wiggling.

Good luck.

Chainminer - Didn't stay high long enough to be observed from the pool 

Decided to lock in my case (using v2 spotswood case for this)  Its a bitch dealing with taking the cards out and tuning them, but it seems necessary to keep the cards still in case airflow from the fans causes a slight wobble which causes the boards to cut out. 

Say...when the card cuts out like this...is that the equivalent to "NO HOTSWAPPING!"?  It would seem that the cut off board would either surge the voltage across the other boards or cause the entire bank to fail. 

I also fried my 1st raspberry pi trying to make my own 2in gpio cable so that the pi wouldnt be hanging off the edge.  The pi wasnt totally dead, but it was acting like the card image got corrupted - turns on with red pwr light only but that's it.  Is this Pi done for?

+1 looking for them as well

was this 640 from the pool or in Chainminer?

i'm consistently getting higher readings from the pool.

Anyone here have any V1.2 h-cards they'd be willing to sell? (ie the ones that work with v2.3 m-boards NOT the new PCI connector style m-boards).

I wont lie...I was really pissed off the 1st night I got it because it just kept shutting down...and I was under the presumption that my 700W psu wasnt enough because it kept powering off.  But after it was working and when i saw it reach 640 for a brief moment on this same psu and not shut down...it gave me that boost in morale to "keep calm and carry on" 

 To be absolutely clear, I'm not calling anything bad. Just a bit surprised that we seem to collectively be having some trouble getting these v3 rigs humming along, stably.

 Eagerly awaiting updates and info on how to get our babies nice and happy.

Guys....you need to use a tiny screwdriver and tune down the voltage a bit...also its looking like we need to keep the cards still so I will likely be taking my fans off the top of my cards too if I can.  From what others are saying... I think this will help guys...Would you have wanted the cards to be perfect....7 days from now?  Because it seems like there would be more waiting.  Not here to take MBP's side per say...but putting in one night of work and calling it Bad is sortof shortsighted...I waited at least a few days (contacted MBP for help, did tuning one by one, and I did apply heatsinks as I knew that if it had worked perfectly out the box I wouldve put in the same time to have it hashing high like 600+.  Just try more stuff and report failure after trying our suggestions before coming back on here upset.  

 Ditto here; seeing the exact same thing Have to keep restarting it every 5 minutes to maintain hash-rate. I tried swapping all sorts of combos of H-Cards in different slots with no discernible effect.

 I feel your pain, bro.

I'm done troubleshooting for the evening Shot a "Product Return" email to the MBP folks detailing the issues I've had this evening and hope to get a response about how to proceed.

- Received 32 H-Cards and 2 M-Boards.
- Identified 2 H-Cards as DOA
- M-Board that I thought was DOA I got hashing after further troubleshooting
- Consistent behavior with slots 2 and 3 across both M-Boards throttling down towards 0 (with slots 6 and 8 following shortly after)
- As per the Mining Support thread, I created a best.cnf with aIfDSo 53 set for all 256 entries -- No difference. Boards eventually throttle to 0 and never recover

Help