[GUIDE] BitFury Miner Support/Tuning

[daddyfatsax]

ventilation and power are key with 90% of hashing / card issues.

If you don't have a fan pointed at your cards, they will overheat and shut off.  Turn the rigs on their left hand side (so the SD card is facing down) and aim a good sized fan towards it.

I have 3 120 mm fans pointed right at the cards. I did power everything down and let it sit for half an hour, but that still did not fix my issue. It seems like the cards are not seating correctly in the M-board and therefore not hashing. I have communications with Megabigpower and will probably have to RMA the M-board. It is just weird it all of sudden happened after almost a month of working flawlessly.

[bobcaticus]

ventilation and power are key with 90% of hashing / card issues.

If you don't have a fan pointed at your cards, they will overheat and shut off.  Turn the rigs on their left hand side (so the SD card is facing down) and aim a good sized fan towards it.

I have 3 120 mm fans pointed right at the cards. I did power everything down and let it sit for half an hour, but that still did not fix my issue. It seems like the cards are not seating correctly in the M-board and therefore not hashing. I have communications with Megabigpower and will probably have to RMA the M-board. It is just weird it all of sudden happened after almost a month of working flawlessly.

That would be me then

M-boards will be in around Feb 21st due to the delays around Chinese New Year.   I do get to work with some of the returns, and yes this sadly does happen.  If there's a dead slot or two, i just shuffle cards around until they find a good place.

[daddyfatsax]

ventilation and power are key with 90% of hashing / card issues.

If you don't have a fan pointed at your cards, they will overheat and shut off.  Turn the rigs on their left hand side (so the SD card is facing down) and aim a good sized fan towards it.

I have 3 120 mm fans pointed right at the cards. I did power everything down and let it sit for half an hour, but that still did not fix my issue. It seems like the cards are not seating correctly in the M-board and therefore not hashing. I have communications with Megabigpower and will probably have to RMA the M-board. It is just weird it all of sudden happened after almost a month of working flawlessly.

That would be me then

M-boards will be in around Feb 21st due to the delays around Chinese New Year.   I do get to work with some of the returns, and yes this sadly does happen.  If there's a dead slot or two, i just shuffle cards around until they find a good place.

Then you probably saw my emails from last night and responded rather quickly.

I spent about 4 hours doing that last night. The problem is that it appears that I have more bad slots than I do cards. Right now 5 cards are hashing in 0, 4, 8, 9, and C. I cannot get the 6th card to hash in any spot. Even when I put a card I know hashes in slot 2 or 3, I still get nothing.

[cypherdoc]

ventilation and power are key with 90% of hashing / card issues.

If you don't have a fan pointed at your cards, they will overheat and shut off.  Turn the rigs on their left hand side (so the SD card is facing down) and aim a good sized fan towards it.

I have 3 120 mm fans pointed right at the cards. I did power everything down and let it sit for half an hour, but that still did not fix my issue. It seems like the cards are not seating correctly in the M-board and therefore not hashing. I have communications with Megabigpower and will probably have to RMA the M-board. It is just weird it all of sudden happened after almost a month of working flawlessly.

That would be me then

M-boards will be in around Feb 21st due to the delays around Chinese New Year.   I do get to work with some of the returns, and yes this sadly does happen.  If there's a dead slot or two, i just shuffle cards around until they find a good place.

Then you probably saw my emails from last night and responded rather quickly.

I spent about 4 hours doing that last night. The problem is that it appears that I have more bad slots than I do cards. Right now 5 cards are hashing in 0, 4, 8, 9, and C. I cannot get the 6th card to hash in any spot. Even when I put a card I know hashes in slot 2 or 3, I still get nothing.

i'm having exactly the same issues.

[Keefe]

It's unfortunate that they didn't spend the little bit extra to gold plate the contacts on the cards like most PC expansion cards are. I think that would have prevented alot of problems.

[kano]

Or that the original design didn't include any temperature sensors anywhere ....
I think it was 2011 when I first brought that one up with then BFL, still companies make the same stupid mistake today ...

p.s. I finally added tuning to cgminer, options to modify the tuning, and the API stats is full of information about each board ... latest changes are in current git.
I fixed more bugs I copied from chainminer also and found an interesting result the boards return that can be ignored and reduce work checking dramatically.

API summary+stats output of my one BA board:

Code:

   [0] => SUMMARY
   [Elapsed] => 176001
   [MHS av] => 41505.75
   [MHS 5s] => 43373.70
   [Found Blocks] => 0
   [Getworks] => 20828
   [Accepted] => 58864
   [Rejected] => 3328
   [Hardware Errors] => 46211
   [Utility] => 20.07
   [Discarded] => 21684
   [Stale] => 25
   [Get Failures] => 15
   [Local Work] => 2768847
   [Remote Failures] => 3
   [Network Blocks] => 350
   [Total MH] => 7305039289.1326
   [Work Utility] => 579.83
   [Difficulty Accepted] => 1619791.28726888
   [Difficulty Rejected] => 57802.70362325
   [Difficulty Stale] => 352.00000000
   [Best Share] => 808573
   [Device Hardware%] => 2.6451
   [Device Rejected%] => 3.3985
   [Pool Rejected%] => 3.4448
   [Pool Stale%] => 0.0210
   [Last getwork] => 1391604057

   [STATS] => 0
   [ID] => BaB0
   [Elapsed] => 176001
   [Calls] => 0
   [Wait] => 0.000000
   [Max] => 0.000000
   [Min] => 99999999.000000
   [Version] => 2
   [Chips] => 16
   [Boards] => 1
   [Banks] => 1
   [Chips Per Bank] => 0 16 0 0 0
   [Missing Chips Per Bank] => 0 0 0 0 0
   [Device Elapsed] => 176003
   [History Enabled] => true
   [Chip History Limit] => 300
   [Nonces 0-15] => 117283 112518 111587 107190 109769 113981 106979 106443 107113 112412 106065 106157 114316 106283 104793 104409
   [Good 0-15] => 111519 106752 109167 105769 108942 109436 103353 105949 106447 104870 105433 105540 109630 104876 101620 101544
   [Bad 0-15] => 5749 5751 2405 1406 812 4530 3611 479 651 7527 617 602 4671 1392 3158 2850
   [Conf 0-15] => 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b
   [Fast 0-15] => 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55
   [Spie 0-15] => 14 8 12 8 9 10 12 7 13 14 11 10 83 73 70 58
   [Miso 0-15] => 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
   [HW% 0-15] => 4.902 5.112 2.156 1.312 0.740 3.975 3.376 0.450 0.608 6.697 0.582 0.567 4.087 1.310 3.014 2.730
   [GHs 0-15] => 2.721 2.605 2.664 2.581 2.658 2.671 2.522 2.585 2.598 2.559 2.573 2.575 2.675 2.559 2.480 2.478
   [Sum GHs 0-15] => 41.506
   [Cont-Bad 0-15] => 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
   [Max-Bad 0-15] => 4 5 3 5 3 4 4 4 4 4 4 3 8 6 5 5
   [History Good 0-15] => 179 175 227 186 174 195 185 194 170 192 176 184 193 174 167 167
   [History Bad 0-15] => 13 18 3 4 0 10 9 0 3 15 1 0 8 0 3 5
   [History HW% 0-15] => 6.771 9.326 1.304 2.105 0.000 4.878 4.639 0.000 1.734 7.246 0.565 0.000 3.980 0.000 1.765 2.907
   [History GHs 0-15] => 2.600 2.506 3.244 2.693 2.501 2.785 2.641 2.770 2.434 2.751 2.521 2.627 2.775 2.492 2.383 2.383
   [Sum History GHs 0-15] => 42.105
   [Total History GHs] => 42.105
   [Total History HW%] => 3.036
   [History Speed 0.0 Dead] => 0
   [History Speed 0.8 V.Slow] => 0
   [History Speed 1.6 Slow] => 0
   [History Speed 2.2 OK] => 0
   [History Speed 2.8 Good] => 15
   [History Speed >2.8 Fast] => 1
   [History Dead] => None
   [History V.Slow] => None
   [History Dead Boards] => None
   [History Dead Chains] => None
   [Nonce Offset 0xff800000] => 890671
   [Nonce Offset 0x00000000] => 566858
   [Nonce Offset 0xffc00000] => 243318
   [Discarded E0s] => 2728775
   [Tested] => 1747298
   [OK] => 1700847
   [Total Tests] => 8977596
   [Max Tests] => 16
   [Avg Tests] => 5.278
   [Untested] => 0
   [Work Links] => 3676660
   [Work Processed Links] => 3613284
   [Max Links] => 8
   [Max Processed Links] => 7
   [Total Work Links] => 6358964
   [Avg Links] => 2.162
   [Avg Proc Links] => 2.124
   [Avg Work Links] => 3.739
   [Fail] => 46211
   [Fail Total Tests] => 405639
   [Fail Avg Tests] => 8.778
   [Fail Work Links] => 173878
   [Fail Total Work Links] => 173878
   [Initial Ignored] => 240
   [Ign Total Tests] => 471
   [Ign Work Links] => 0
   [Ign Total Work Links] => 0
   [WFree Total] => 1024
   [WFree Count] => 943
   [Available Work] => 16
   [SPI Work] => 0
   [Chip Work] => 65
   [SFree Total] => 8
   [SFree Count] => 8
   [SPI Waiting] => 0
   [SPI Sent] => 0
   [RFree Total] => 256
   [RFree Count] => 256
   [Result Count] => 0
   [NFree Total] => 102400
   [NFree Used] => 3030
   [Delay Count] => 170550
   [Delay Min] => 0.899001
   [Delay Max] => 5.202787
   [Delay Bands] => <0.5=0 <0.7=0 <0.9=5828 <1.1=164717 <1.3=2 <1.5=0 <1.7=0 <1.9=0 <2.1=0 <2.3=0 <2.5=0 >=2.7=3
   [Send Count] => 170551
   [Send Total] => 14938.829633
   [Send Avg] => 0.088
   [Send Min] => 0.085179
   [Send Max] => 0.113610
   [Reply Wait] => 160
   [Reply Waits] => 5303
   [MaxSpeed] => 55
   [DefaultSpeed] => 54
   [MinSpeed] => 52
   [TuneUp] => 1.000000
   [TuneDown] => 10.000000
   [SPISpeed] => 96000
   [SPIDelayuS] => 0
   [TransferDelayuS] => 0
   [USB Pipe] => 0
   [USB Delay] => r0 0.000000 w0 0.000000
   [USB tmo] => 0 0

The Paid GH/s there (how much the pool accepts) is 39.5GH/s

[darkfriend77]

I just want to thank you guys ... for the interesting and super exciting time I got with you and the bitfury chips since I started this thread in september ...

I will sell my bitfury rigs as long as there is still power in there mining capability (ebay) ... I got the rigs to 420 GH each with 12 slots occupied ...

best results with max. 3 slots filled each column ... I never achieved a stable full setup with 16 cards to go over 500 GH/s for a long period ...

but filled at 75% they are stable and run at 35GH each card ... let's see if the nextGen bitfury will be available for the community ...

[CCCrypto]

Is there a diagram anywhere online showing all the components of the V1.2 H-cards? (The chips and capacitors are easy enough to spot, but I'd like to know specifically where the regulator is, and the inductor.) I tried reading the schematic, but it wasn't very helpful for me.

Thanks.

[spegelius]

Or that the original design didn't include any temperature sensors anywhere ....
I think it was 2011 when I first brought that one up with then BFL, still companies make the same stupid mistake today ...

p.s. I finally added tuning to cgminer, options to modify the tuning, and the API stats is full of information about each board ... latest changes are in current git.
I fixed more bugs I copied from chainminer also and found an interesting result the boards return that can be ignored and reduce work checking dramatically.

API summary+stats output of my one BA board:

Code:

   [0] => SUMMARY
   [Elapsed] => 176001
   [MHS av] => 41505.75
   [MHS 5s] => 43373.70
   [Found Blocks] => 0
   [Getworks] => 20828
   [Accepted] => 58864
   [Rejected] => 3328
   [Hardware Errors] => 46211
   [Utility] => 20.07
   [Discarded] => 21684
   [Stale] => 25
   [Get Failures] => 15
   [Local Work] => 2768847
   [Remote Failures] => 3
   [Network Blocks] => 350
   [Total MH] => 7305039289.1326
   [Work Utility] => 579.83
   [Difficulty Accepted] => 1619791.28726888
   [Difficulty Rejected] => 57802.70362325
   [Difficulty Stale] => 352.00000000
   [Best Share] => 808573
   [Device Hardware%] => 2.6451
   [Device Rejected%] => 3.3985
   [Pool Rejected%] => 3.4448
   [Pool Stale%] => 0.0210
   [Last getwork] => 1391604057

   [STATS] => 0
   [ID] => BaB0
   [Elapsed] => 176001
   [Calls] => 0
   [Wait] => 0.000000
   [Max] => 0.000000
   [Min] => 99999999.000000
   [Version] => 2
   [Chips] => 16
   [Boards] => 1
   [Banks] => 1
   [Chips Per Bank] => 0 16 0 0 0
   [Missing Chips Per Bank] => 0 0 0 0 0
   [Device Elapsed] => 176003
   [History Enabled] => true
   [Chip History Limit] => 300
   [Nonces 0-15] => 117283 112518 111587 107190 109769 113981 106979 106443 107113 112412 106065 106157 114316 106283 104793 104409
   [Good 0-15] => 111519 106752 109167 105769 108942 109436 103353 105949 106447 104870 105433 105540 109630 104876 101620 101544
   [Bad 0-15] => 5749 5751 2405 1406 812 4530 3611 479 651 7527 617 602 4671 1392 3158 2850
   [Conf 0-15] => 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b 0x1b
   [Fast 0-15] => 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55
   [Spie 0-15] => 14 8 12 8 9 10 12 7 13 14 11 10 83 73 70 58
   [Miso 0-15] => 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
   [HW% 0-15] => 4.902 5.112 2.156 1.312 0.740 3.975 3.376 0.450 0.608 6.697 0.582 0.567 4.087 1.310 3.014 2.730
   [GHs 0-15] => 2.721 2.605 2.664 2.581 2.658 2.671 2.522 2.585 2.598 2.559 2.573 2.575 2.675 2.559 2.480 2.478
   [Sum GHs 0-15] => 41.506
   [Cont-Bad 0-15] => 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
   [Max-Bad 0-15] => 4 5 3 5 3 4 4 4 4 4 4 3 8 6 5 5
   [History Good 0-15] => 179 175 227 186 174 195 185 194 170 192 176 184 193 174 167 167
   [History Bad 0-15] => 13 18 3 4 0 10 9 0 3 15 1 0 8 0 3 5
   [History HW% 0-15] => 6.771 9.326 1.304 2.105 0.000 4.878 4.639 0.000 1.734 7.246 0.565 0.000 3.980 0.000 1.765 2.907
   [History GHs 0-15] => 2.600 2.506 3.244 2.693 2.501 2.785 2.641 2.770 2.434 2.751 2.521 2.627 2.775 2.492 2.383 2.383
   [Sum History GHs 0-15] => 42.105
   [Total History GHs] => 42.105
   [Total History HW%] => 3.036
   [History Speed 0.0 Dead] => 0
   [History Speed 0.8 V.Slow] => 0
   [History Speed 1.6 Slow] => 0
   [History Speed 2.2 OK] => 0
   [History Speed 2.8 Good] => 15
   [History Speed >2.8 Fast] => 1
   [History Dead] => None
   [History V.Slow] => None
   [History Dead Boards] => None
   [History Dead Chains] => None
   [Nonce Offset 0xff800000] => 890671
   [Nonce Offset 0x00000000] => 566858
   [Nonce Offset 0xffc00000] => 243318
   [Discarded E0s] => 2728775
   [Tested] => 1747298
   [OK] => 1700847
   [Total Tests] => 8977596
   [Max Tests] => 16
   [Avg Tests] => 5.278
   [Untested] => 0
   [Work Links] => 3676660
   [Work Processed Links] => 3613284
   [Max Links] => 8
   [Max Processed Links] => 7
   [Total Work Links] => 6358964
   [Avg Links] => 2.162
   [Avg Proc Links] => 2.124
   [Avg Work Links] => 3.739
   [Fail] => 46211
   [Fail Total Tests] => 405639
   [Fail Avg Tests] => 8.778
   [Fail Work Links] => 173878
   [Fail Total Work Links] => 173878
   [Initial Ignored] => 240
   [Ign Total Tests] => 471
   [Ign Work Links] => 0
   [Ign Total Work Links] => 0
   [WFree Total] => 1024
   [WFree Count] => 943
   [Available Work] => 16
   [SPI Work] => 0
   [Chip Work] => 65
   [SFree Total] => 8
   [SFree Count] => 8
   [SPI Waiting] => 0
   [SPI Sent] => 0
   [RFree Total] => 256
   [RFree Count] => 256
   [Result Count] => 0
   [NFree Total] => 102400
   [NFree Used] => 3030
   [Delay Count] => 170550
   [Delay Min] => 0.899001
   [Delay Max] => 5.202787
   [Delay Bands] => <0.5=0 <0.7=0 <0.9=5828 <1.1=164717 <1.3=2 <1.5=0 <1.7=0 <1.9=0 <2.1=0 <2.3=0 <2.5=0 >=2.7=3
   [Send Count] => 170551
   [Send Total] => 14938.829633
   [Send Avg] => 0.088
   [Send Min] => 0.085179
   [Send Max] => 0.113610
   [Reply Wait] => 160
   [Reply Waits] => 5303
   [MaxSpeed] => 55
   [DefaultSpeed] => 54
   [MinSpeed] => 52
   [TuneUp] => 1.000000
   [TuneDown] => 10.000000
   [SPISpeed] => 96000
   [SPIDelayuS] => 0
   [TransferDelayuS] => 0
   [USB Pipe] => 0
   [USB Delay] => r0 0.000000 w0 0.000000
   [USB tmo] => 0 0

The Paid GH/s there (how much the pool accepts) is 39.5GH/s

Just to confirm, this doesn't support Bitfury v1.2 hcards/mcard combo? Compiled with --enable-bitfury and it doesn't find anything. And no mention in ASIC-README about hcards...

[kano]

...
Just to confirm, this doesn't support Bitfury v1.2 hcards/mcard combo? Compiled with --enable-bitfury and it doesn't find anything. And no mention in ASIC-README about hcards...

--enable-bab

Bitfury never supported my writing the software and wouldn't even sell me hardware - so it wasn't until someone sent me a BlackArrowBitfury board that I could write and test a driver for it.

The detection code and work send command is almost the same logic as chainminer - just less bugs
If they've updated the code and not released (i.e. don't want anyone to support their hardware) then yeah my code won't know about the updates
https://github.com/bfsb/chainminer is their code as far as I know ...

[spegelius]

...
Just to confirm, this doesn't support Bitfury v1.2 hcards/mcard combo? Compiled with --enable-bitfury and it doesn't find anything. And no mention in ASIC-README about hcards...

--enable-bab

Bitfury never supported my writing the software and wouldn't even sell me hardware - so it wasn't until someone sent me a BlackArrowBitfury board that I could write and test a driver for it.

The detection code and work send command is almost the same logic as chainminer - just less bugs
If they've updated the code and not released (i.e. don't want anyone to support their hardware) then yeah my code won't know about the updates
https://github.com/bfsb/chainminer is their code as far as I know ...

Thanks, got it working, doing quite nice, only 10-15GH/s slower than what Bifury's management page says when mining with chainminer. With chainminer i have handtuned the chip speeds. 248GH/s on average (10 hours of runtime) versus 260 - 267 in chainminer. I wonder if manual tuning of those autotune parameters would bring it up.

I'll let this run for a while.

[kano]

...
Just to confirm, this doesn't support Bitfury v1.2 hcards/mcard combo? Compiled with --enable-bitfury and it doesn't find anything. And no mention in ASIC-README about hcards...

--enable-bab

Bitfury never supported my writing the software and wouldn't even sell me hardware - so it wasn't until someone sent me a BlackArrowBitfury board that I could write and test a driver for it.

The detection code and work send command is almost the same logic as chainminer - just less bugs
If they've updated the code and not released (i.e. don't want anyone to support their hardware) then yeah my code won't know about the updates
https://github.com/bfsb/chainminer is their code as far as I know ...

Thanks, got it working, doing quite nice, only 10-15GH/s slower than what Bifury's management page says when mining with chainminer. With chainminer i have handtuned the chip speeds. 248GH/s on average (10 hours of runtime) versus 260 - 267 in chainminer. I wonder if manual tuning of those autotune parameters would bring it up.

I'll let this run for a while.

The --bab-options will allow it to go higher also.
Mainly the first parameter, the Max speed - it defaults to 55 but will allow up to 57
(but that depends on if your board can handle it that high also)

I set the defaults to match a 'slower' BitFury board so that people who tried it didn't blow anything up without messing with options
I've not actually been able to run it on my 6 board for a while now - so I don't know how well it goes with a lot more boards
(the 6 board setup has been out of my hands for a couple of weeks now)
It may still require some tuning (of the 'Delay' code to get the average down when there are a lot of boards) but until I have a larger board setup to test with, I can't really sort that one out.
Shouldn't be too much longer.

It checks each chip results roughly every 5minutes and then tunes them up or down based on HW% for that 5 min of data.
default is down if >10% HW and up if <1%

I do still have other ideas for tuning (and dealing with dead chips coz they use double or more CPU than a live chip)

[spegelius]

The --bab-options will allow it to go higher also.
Mainly the first parameter, the Max speed - it defaults to 55 but will allow up to 57
(but that depends on if your board can handle it that high also)

I set the defaults to match a 'slower' BitFury board so that people who tried it didn't blow anything up without messing with options
I've not actually been able to run it on my 6 board for a while now - so I don't know how well it goes with a lot more boards
(the 6 board setup has been out of my hands for a couple of weeks now)
It may still require some tuning (of the 'Delay' code to get the average down when there are a lot of boards) but until I have a larger board setup to test with, I can't really sort that one out.
Shouldn't be too much longer.

It checks each chip results roughly every 5minutes and then tunes them up or down based on HW% for that 5 min of data.
default is down if >10% HW and up if <1%

I do still have other ideas for tuning (and dealing with dead chips coz they use double or more CPU than a live chip)

Had to switch to chainminer, the speed was tanking to 220Gh/s with cgminer, occasionally going higher and then back. I'll do more testing when i have more time to concentrate on tuning.

[Slander]

My sd card got corrupted so I imaged it again not realizing I used an old version of the image file from megabigpower. Nothing but low hash rates and 0 hash rates for 2 days.

When I realized this I put a new image on the sd card from megabig, the one linked in their help file pdf.

Its been running rocksolid for 2 days at full hash rate, about 30-34 ghash per card on an 8 card setup.

so, make sure you are running the most current image file as it makes a huge difference

Make sure you are using the correct image.  On the M-Board it will say what version the board is.  You can fit the V1 and V2 image on a 4GB SD card, however you will need an 8GB SD card for the V3 image. 

V1 (3.64GB)
https://mega.co.nz/#!GZEwQRiJ!TqG52DBj07C_zGO38dhXuyeNhwZ42VB2qk2ErfsrUZ8
 

V2 (3.64GB)
https://mega.co.nz/#!iVEwFKpT!TqRndnWurfcQ0V2gJb2sQlNcSGoMF1B7YpUoGDvVebI​

V3  (3.69GB)
https://mega.co.nz/#!7AFVhLwQ!BmgAIx32aQQxyadT1WKdJzNOUGQAQcQgYemfSJ6kMNM

Mirror: https://www.dropbox.com/s/skr9vitueeyxwmw/miner-v3-image-11-15-2013-customer.img

[bobcaticus]

My sd card got corrupted so I imaged it again not realizing I used an old version of the image file from megabigpower. Nothing but low hash rates and 0 hash rates for 2 days.

When I realized this I put a new image on the sd card from megabig, the one linked in their help file pdf.

Its been running rocksolid for 2 days at full hash rate, about 30-34 ghash per card on an 8 card setup.

so, make sure you are running the most current image file as it makes a huge difference

Make sure you are using the correct image.  On the M-Board it will say what version the board is.  You can fit the V1 and V2 image on a 4GB SD card, however you will need an 8GB SD card for the V3 image. 

V1 (3.64GB)
https://mega.co.nz/#!GZEwQRiJ!TqG52DBj07C_zGO38dhXuyeNhwZ42VB2qk2ErfsrUZ8
 

V2 (3.64GB)
https://mega.co.nz/#!iVEwFKpT!TqRndnWurfcQ0V2gJb2sQlNcSGoMF1B7YpUoGDvVebI​

V3  (3.69GB)
https://mega.co.nz/#!7AFVhLwQ!BmgAIx32aQQxyadT1WKdJzNOUGQAQcQgYemfSJ6kMNM

Mirror: https://www.dropbox.com/s/skr9vitueeyxwmw/miner-v3-image-11-15-2013-customer.img

Hey, thanks for reading my PDF

Technically, i've been told that you can run the latest version on everything - but i'm pretty pessimistic on that still.  Putting in a new SD card does wonders, as it has been noted in here time and time again that they do putz out.

I know that kits say overclockable - but due to the different card types, I fear that many people will try to tweak their systems using the wrong autochip commands and so on.

V1 cards can go up to 57
V2 cards can go up to 55
V2.2 cards (with the metal backing) should stick around 53

I've spent so many hours fiddling with these units and in my opinion: keeping it simple seems to be the best solution. 

If you don't know what you are doing, just run with the basic setup.  I think a bunch of people are trying to squeeze blood out of a rock (so to say), and frying their cards in the process.  I know the difficulty is going up crazy, and yes - it bums me out - but that is the way bitcoin was setup from the start. 

Dave in Germany right now - then he's going to finally meet Putin in real life(!).  Those two are going to get something hashed out concerning next-gen rigs and I can't wait to see what happens.


-Jason

[klondike_bar]

^really hope it involves actual competitive pricing. For a more-stable, stackable/enclosed bitfury I would gladly add it to my growing collection of antminers - but only if they can get their shit together and
1) price in BTC - for the love of god.
2) price no higher then 10% above the antminer - those things are simple, elegant tanks that dont wobble, dont blow sd cards, and only require the single fan that they ship with. The only upside to the bitfury is saving ~1w/GH

[Keefe]

^really hope it involves actual competitive pricing. For a more-stable, stackable/enclosed bitfury I would gladly add it to my growing collection of antminers - but only if they can get their shit together and
1) price in BTC - for the love of god.
2) price no higher then 10% above the antminer - those things are simple, elegant tanks that dont wobble, dont blow sd cards, and only require the single fan that they ship with. The only upside to the bitfury is saving ~1w/GH

And saving 1W/GH only saves $0.88/GH per year at $0.10/kwh. Bitfury 55nm will be obsoleted by even more efficient tech within a year anyway, so it's not worth the huge premium Punin seems to think it is.

[daddyfatsax]

My sd card got corrupted so I imaged it again not realizing I used an old version of the image file from megabigpower. Nothing but low hash rates and 0 hash rates for 2 days.

When I realized this I put a new image on the sd card from megabig, the one linked in their help file pdf.

Its been running rocksolid for 2 days at full hash rate, about 30-34 ghash per card on an 8 card setup.

so, make sure you are running the most current image file as it makes a huge difference

Make sure you are using the correct image.  On the M-Board it will say what version the board is.  You can fit the V1 and V2 image on a 4GB SD card, however you will need an 8GB SD card for the V3 image.  

V1 (3.64GB)
https://mega.co.nz/#!GZEwQRiJ!TqG52DBj07C_zGO38dhXuyeNhwZ42VB2qk2ErfsrUZ8
 

V2 (3.64GB)
https://mega.co.nz/#!iVEwFKpT!TqRndnWurfcQ0V2gJb2sQlNcSGoMF1B7YpUoGDvVebI​

V3  (3.69GB)
https://mega.co.nz/#!7AFVhLwQ!BmgAIx32aQQxyadT1WKdJzNOUGQAQcQgYemfSJ6kMNM

Mirror: https://www.dropbox.com/s/skr9vitueeyxwmw/miner-v3-image-11-15-2013-customer.img

Hey, thanks for reading my PDF

Technically, i've been told that you can run the latest version on everything - but i'm pretty pessimistic on that still.  Putting in a new SD card does wonders, as it has been noted in here time and time again that they do putz out.

I know that kits say overclockable - but due to the different card types, I fear that many people will try to tweak their systems using the wrong autochip commands and so on.

V1 cards can go up to 57
V2 cards can go up to 55
V2.2 cards (with the metal backing) should stick around 53

I've spent so many hours fiddling with these units and in my opinion: keeping it simple seems to be the best solution.  

If you don't know what you are doing, just run with the basic setup.  I think a bunch of people are trying to squeeze blood out of a rock (so to say), and frying their cards in the process.  I know the difficulty is going up crazy, and yes - it bums me out - but that is the way bitcoin was setup from the start.  

Dave in Germany right now - then he's going to finally meet Putin in real life(!).  Those two are going to get something hashed out concerning next-gen rigs and I can't wait to see what happens.


-Jason

The V3 image works just fine on a V1 board. I bought goxed's 10 card V1 unit and it is working perfectly. My idiot self corrupted his card, that I know he spent a lot of time tuning, after a week.

[adib]

My 2 cents, thought i should post here as well, given the unavailability of M-boards.

After turning this: into this: and this: into this: ,
because there were no 0603 220k resistors in stock.


The result: , ,

P.S. Thanks intron for the help and support

4 boards fixed and accepting chips

Schematic:
TOP VIEW


BOTTOM VIEW


NOTES:
- ONLY USE V2 CHAINMINER SOFTWARE AS THE SCHEMATIC IS BASED ON THAT, its actually a bank of the v2 m-board
- tried to keep air wires to a minimum (8 to be exact)
- LM2596 power module should be placed exactly where it is placed on the layout, otherwise youll have trouble in placing the H-cards, youll have some tolerance if you use flexible wires.
- Scale is 1:1 so i did not bother to place dimensions
- full resolution images here:  https://www.dropbox.com/sh/yl0fov35cmv3r7p/gR4W5vdL6W
- pins on the slots are a bit tighter then they should be so youll have to bend the first 4 pins on each side inwards a bit (made them from the hip)

I have decided to stop making other version... aka the 8,12 or 16 slot versions although you can adapt as you see fit.
You can also connect 4 m-boards with 4 slots toghether like so:

full image here: https://www.dropbox.com/sh/yl0fov35cmv3r7p/gR4W5vdL6W
 

[daddyfatsax]

Is there anyway to use a different stratum proxy with chain miner? Resetting it every 10 or 60 min still leads to issues after several days. Anyway to put slush's proxy in it, or is that what it is using?