<<  >> (p.6)
    Author Topic: [GUIDE] BitFury Miner Support/Tuning  (Read 148101 times)
    This is a self-moderated topic. If you do not want to be moderated by the person who started this topic, create a new topic.

    • dacman61
    Member
    **
    Offline Offline

    Activity: 87
    Merit: 10



    View Profile
    September 15, 2013, 01:50:06 PM
     #101
    This pencil mod trick has been working great! At first I used too much pencil and had a resistance of 1.08k and the noncerate shot up to 35Gh/s then dropped down quickly... Soon after I wiped away some of the pencil marking, I finally read a resistance of 1.18k, and this turned out to be the sweet spot for me... I've been steady at 30.5-32Gh/s noncerate now. I also put a small turbo desk fan blowing on it at low speed.

    Code:
    16      AIfDSo  55      1.904   2.315   133     7       0       0       219     [0:F]   436     9 9 8 9 9 7 8 8 8 8 8 8 9 9 9 7         0 0 1 0 0 2 1 1 0 0 0 0 0 0 0 2
    speed:880 noncerate[GH/s]:32.098 (2.006/chip) hashrate[GH/s]:34.542 good:2242 errors:190 spi-err:0 miso-err:0 jobs:375 cores:42% good:16 bad:0 off:0 (best[GH/s]:31.525) Sun Sep 15 13:38:13 2013
    0:      880     32.098  34.542  2242    190     0       0       16      0       0       (2.006/chip)    42%

    Code:
    16      AIfDSo  55      1.847   2.304   129     14      0       0       218     [0:F]   363     9 8 7 8 7 8 8 8 8 9 8 8 9 7 9 8         0 1 2 1 2 1 1 0 1 0 1 1 0 2 0 1
    speed:880 noncerate[GH/s]:30.594 (1.912/chip) hashrate[GH/s]:34.584 good:2137 errors:176 spi-err:0 miso-err:0 jobs:375 cores:51% good:16 bad:0 off:0 (best[GH/s]:32.098) Sun Sep 15 13:43:13 2013
    0:      880     30.594  34.584  2137    176     0       0       16      0       0       (1.912/chip)    51%

    I've been running now for the past 30 minutes at the time of me gathering this info.

    How does this look to you?

  • juhakall
    Sr. Member
    ****
    Offline Offline

    Activity: 658
    Merit: 250


    View Profile
    September 15, 2013, 01:55:47 PM
     #102
    Quote from: dacman61 on September 15, 2013, 01:50:06 PM
    This pencil mod trick has been working great! At first I used too much pencil and had a resistance of 1.08k and the noncerate shot up to 35Gh/s then dropped down quickly... Soon after I wiped away some of the pencil marking, I finally read a resistance of 1.18k, and this turned out to be the sweet spot for me... I've been steady at 30.5-32Gh/s noncerate now. I also put a small turbo desk fan blowing on it at low speed.

    Code:
    16      AIfDSo  55      1.904   2.315   133     7       0       0       219     [0:F]   436     9 9 8 9 9 7 8 8 8 8 8 8 9 9 9 7         0 0 1 0 0 2 1 1 0 0 0 0 0 0 0 2
    speed:880 noncerate[GH/s]:32.098 (2.006/chip) hashrate[GH/s]:34.542 good:2242 errors:190 spi-err:0 miso-err:0 jobs:375 cores:42% good:16 bad:0 off:0 (best[GH/s]:31.525) Sun Sep 15 13:38:13 2013
    0:      880     32.098  34.542  2242    190     0       0       16      0       0       (2.006/chip)    42%

    Code:
    16      AIfDSo  55      1.847   2.304   129     14      0       0       218     [0:F]   363     9 8 7 8 7 8 8 8 8 9 8 8 9 7 9 8         0 1 2 1 2 1 1 0 1 0 1 1 0 2 0 1
    speed:880 noncerate[GH/s]:30.594 (1.912/chip) hashrate[GH/s]:34.584 good:2137 errors:176 spi-err:0 miso-err:0 jobs:375 cores:51% good:16 bad:0 off:0 (best[GH/s]:32.098) Sun Sep 15 13:43:13 2013
    0:      880     30.594  34.584  2137    176     0       0       16      0       0       (1.912/chip)    51%

    I've been running now for the past 30 minutes at the time of me gathering this info.

    How does this look to you?

    Great result, though it sounds weird that the hashrate wasn't stable at first with 1.08k. BTW, how do you get these stats: "cores:51% good:16 bad:0 off:0"? I just have the record (it's called 'record' on my stats, not 'best') right after jobs. Are those some extra stats that are shown when autotune is enabled?

  • dacman61
    Member
    **
    Offline Offline

    Activity: 87
    Merit: 10



    View Profile
    September 15, 2013, 02:00:36 PM
     #103
    I'm running this command as suggested by -Redacted-:

    Code:
    while (true) do tail -3run/shm/.stat.log; sleep 305; done

    Latest reading:

    Code:
    16      AIfDSo  55      1.933   2.325   135     19      0       0       220     [0:F]   197     8 9 10 8 9 9 9 8 9 8 8 8 7 8 8 9        2 1 0 2 1 1 1 2 0 1 1 1 2 1 2 1
    speed:880 noncerate[GH/s]:32.270 (2.017/chip) hashrate[GH/s]:34.753 good:2254 errors:165 spi-err:0 miso-err:0 jobs:295 cores:71% good:16 bad:0 off:0 (best[GH/s]:32.255) Sun Sep 15 13:58:13 2013
    0:      880     32.270  34.753  2254    165     0       0       16      0       0       (2.017/chip)    71%

  • GH
    Member
    **
    Offline Offline

    Activity: 117
    Merit: 10


    View Profile
    September 15, 2013, 02:07:07 PM
     #104
    Quote from: Cablez on September 15, 2013, 12:43:34 PM
    Two quick questions on the goxxed info.  1) What points are you using exactly to measure the altered voltage?   I am leery to probe around the board while powered up for fear of shorting something.
                                                            2) What exactly is reniceing and why is it important to reduce?  This is the first time I have seen this term used in the bitfury threads.

    Thanks very much and awesome work!  Grin

    Meanwhile I have a stable nonce rate of average >35 Gh/s on both boards - so time to give something back:

    You can measure the voltage on nearly every capacitor on the board, I usually use the bigger ones directly beside the DC/DC converter, left to the coil (the bigest part on the board). There are not many ways to do something wrong, just make sure to not shorten your probes, as there are 30+ Amps available. (I have exactly 0.840 Volts there with 55er clocks and it looks very stable, although there is no easy way to measure the real current draw on the core voltage and confirm that everything is inside bounds.)

    BTW: I also reported about a mysterious chip, which usually throws lots of errors, but sometimes recovers completely after a longer uptime. When running this board alone or in switched order with the other one, the errors of this single chip also disappear.

  • Foofighter
    Hero Member
    *****
    Offline Offline

    Activity: 882
    Merit: 547


    BTC Mining Hardware, Trading and more


    View Profile
    September 15, 2013, 02:15:52 PM
     #105
    Quote from: GH on September 15, 2013, 02:07:07 PM
    Quote from: Cablez on September 15, 2013, 12:43:34 PM
    Two quick questions on the goxxed info.  1) What points are you using exactly to measure the altered voltage?   I am leery to probe around the board while powered up for fear of shorting something.
                                                            2) What exactly is reniceing and why is it important to reduce?  This is the first time I have seen this term used in the bitfury threads.

    Thanks very much and awesome work!  Grin

    Meanwhile I have a stable nonce rate of average >35 Gh/s on both boards - so time to give something back:

    You can measure the voltage on nearly every capacitor on the board, I usually use the bigger ones directly beside the DC/DC converter, left to the coil (the bigest part on the board). There are not many ways to do something wrong, just make sure to not shorten your probes, as there are 30+ Amps available. (I have exactly 0.840 Volts there with 55er clocks and it looks very stable, although there is no easy way to measure the real current draw on the core voltage and confirm that everything is inside bounds.)

    BTW: I also reported about a mysterious chip, which usually throws lots of errors, but sometimes recovers completely after a longer uptime. When running this board alone or in switched order with the other one, the errors of this single chip also disappear.

    Hi, thanks for sharing.

    Can you also confirm the Ohm values given above?
    ex official Canaan Distributor (Cryptouniverse)

  • GH
    Member
    **
    Offline Offline

    Activity: 117
    Merit: 10


    View Profile
    September 15, 2013, 03:46:28 PM
    Last edit: September 24, 2013, 09:34:23 PM by GH
     #106
    Not really, because I do not like quick'n'dirty stuff like the pencil mod. I have a real resistance of 4.07k with 1% metal film (not measured mounted inside the circuit), so you can't compare easily. I suppose this is one reason why it is so stable, regardless of temperature. (Carbon an graphite have a negative temperature coefficient, which causes the resistance to drop with rising temperatures, which leads to even more heat and so on...)

    I'm runnning 5 "server grade" fans on the 2 boards, which make a hell of a noise, but keep the boards and surfaces of the components pretty cool.

    Edit: Added NTC explaination, typos, wrong value

  • juhakall
    Sr. Member
    ****
    Offline Offline

    Activity: 658
    Merit: 250


    View Profile
    September 15, 2013, 08:11:17 PM
     #107
    I did the pencil mod on 2 boards, which maxed out at 47 GH/s prior to modding. Both boards' R02F resistors were modified from 1.3k to 1.18k ohm. Now the noncerate is 59-61 GH/s. Here are the results before and after modding. As you can see, chip #10 stopped making errors. Also, chip #24 stopped requiring a 24 hour period after a reboot until it stops making errors. I added a fan and the highest temps I could measure from the boards dropped from ~80C to ~60C. I'll report if anything changes.

  • joris
    Full Member
    ***
    Offline Offline

    Activity: 141
    Merit: 100


    View Profile
    September 15, 2013, 10:05:34 PM
     #108
    Quote from: goxed on September 15, 2013, 05:39:55 AM
    ...

    g) I Use only one pool for mining

    h) renice the miner proxy process to -14 or lower

    Code:
    2100 root        6 -14 33196 18864  4064 R 81.0  3.8 36h27:30 python ./mining_proxy.py -o us3.eclipsemc.com -p 3333 -cu XXXXXXXXXXX -cp XXXX -gp 8332 -v
    h.1) I use htop to do it
    h.1.1) sudo apt-get install htop
    h.1.2) pi@bitfury $ sudo htop
    h.1.3) Press F7 repeatedly to reduce nice.
            

    This did a lot for my performance: boost of 5 GH on a starter kit (regard as highly anecdotical, single measurement). But it seems to have a logic: instead of 3 processes consuming 50-60% of CPU time, now only 1 'large' process exists, using around 20% of the CPU at a high priority. It seems to relieve the raspi a lot...

    Does it make sense to only prioritise the mining_proxy.py? I see many more chainminer processes in htop...
    ;-)

  • zurg
    Hero Member
    *****
    Offline Offline

    Activity: 826
    Merit: 500


    Crypto Investor ;) @ Farmed Account Hunter


    View Profile
    September 16, 2013, 12:43:42 PM
     #109
    Ok, modded my starter kit.

    1.307 on both boards to 1.215 and 1.229(EOL)
    Went from around
    1: 24.234GH/s
    2: 14.379GH/s
    to
    1: 29.234GH/s
    2: 16.779GH/s

    Bitminter now reporting from 42 to 49GH/s (was nearly 50 a sec ago)
    from about ~35 to 39 before.

    I am in a server room at 62F with fan blowing over. Chips are lukewarm.

    Definitely at least a 5Gh/s increase!

  • spiccioli
    Legendary
    *
    Offline Offline

    Activity: 1379
    Merit: 1003

    nec sine labore


    View Profile
    September 16, 2013, 08:50:13 PM
     #110
    Hi,

    I've found, because of an error, that sometimes a bad chip can hash at low speed using "aIFdso" instead of the normal "aIfDSo" flag string.

    Code:
    18 aIFdso 47 0.687 0.888 48 26 0 0 84 [1:1] 288 4 3 3 4 1 2 3 2 2 3 4 2 4 4 3 4 1 2 1 0 3 2 1 2 3 2 1 3 1 1 2 1
    47 aIFdso 40 0.272 0.835 19 26 0 0 79 [2:E] 443 1 2 1 1 2 1 1 1 2 2 0 1 1 1 1 1 2 1 2 1 0 1 2 2 1 1 3 2 2 2 2 2
    72 aIFdso 47 0.487 0.930 34 19 0 0 88 [4:7] 288 2 3 2 2 2 1 3 2 1 3 2 3 2 2 2 2 1 0 1 1 1 2 0 2 3 1 2 1 1 1 1 1
    97 aIFdso 47 0.515 0.814 36 24 0 0 77 [6:0] 288 1 1 2 2 2 1 2 3 2 3 3 2 4 1 3 4 3 2 1 1 1 3 2 1 2 1 1 2 0 3 1 0
    143 aIFdso 40 0.344 0.909 24 22 0 0 86 [8:E] 504 2 2 1 2 2 2 1 2 0 2 1 1 1 2 0 3 1 1 2 1 1 1 2 1 3 1 2 1 1 1 3 0
    151 aIFdso 47 0.458 0.814 32 14 0 0 77 [9:6] 288 2 2 2 3 3 2 2 3 2 1 1 1 2 2 3 1 1 1 1 0 0 1 1 0 1 1 1 2 1 1 0 2
    162 aIFdso 47 0.558 0.814 39 17 0 0 77 [A:1] 288 2 3 1 4 4 1 2 2 3 3 3 3 1 1 3 3 2 1 3 0 0 2 1 1 0 0 0 0 2 3 1 1
    178 aIFdso 40 0.329 0.824 23 20 0 0 78 [B:1] 552 3 0 2 2 1 1 1 1 0 3 1 2 1 2 2 1 0 3 1 0 1 1 1 1 3 0 2 1 2 1 1 2
    182 aIFdso 40 0.086 0.835 6 45 0 0 79 [B:5] 656 1 1 0 0 0 1 0 0 0 2 0 0 1 0 0 0 2 2 3 3 4 3 4 3 3 1 3 3 2 3 3 3
    200 aIFdso 48 0.716 0.877 50 0 0 0 83 [C:7] 36 3 3 3 3 3 3 3 3 3 3 3 3 4 4 3 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
    205 aIFdso 44 0.544 0.698 38 0 0 0 66 [C:C] 36 2 3 3 3 3 3 3 2 2 2 2 2 2 2 2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
    225 aIFdso 40 0.501 0.835 35 19 0 0 79 [E:0] 288 3 3 1 2 1 3 2 2 0 3 3 2 3 3 2 2 1 1 2 1 2 0 1 1 3 0 0 1 1 1 2 2
    238 aIFdso 40 0.415 0.708 29 15 0 0 67 [E:D] 289 1 3 1 2 2 2 3 2 2 3 1 2 2 0 1 2 2 0 2 1 1 1 0 1 1 0 1 0 0 2 2 1
    254 aIFdso 40 0.229 0.846 16 42 0 0 80 [F:D] 596 0 0 0 1 3 2 1 0 2 1 1 1 1 1 1 1 4 4 4 3 1 1 2 3 1 2 2 3 3 3 3 3
    256 aIFdso 44 0.415 0.877 29 14 0 0 83 [F:F] 324 2 1 2 1 3 3 3 1 2 2 1 2 1 2 1 2 1 2 1 2 0 0 0 2 0 0 1 0 1 1 2 1

    What does it really change using "Internal Fast" instead of "Internal Divided Slow" ?

    I've made tests lowering speed down to 10 without conclusive findings, btw boards are extremely "instable", simply restarting miner can be enough for a good chip to start throwing errors that slow it down/shut it down.

    I'm going to change miner so that best.conf speed is the minimum it can go irregardless of errors of any kind while keeping the auto-tune flag active.

    I'm doing this to see if I can stop the spiralling down of the unit which tends to slow down if left mining alone with auto-tune on.

    spiccioli

    ps. note that I've changed my miner to accept speeds lower than 52 and that I've not pencil-modded them (yet).

  • EvilLizardApparel
    Full Member
    ***
    Offline Offline

    Activity: 133
    Merit: 100



    View Profile
    September 16, 2013, 10:03:47 PM
     #111
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.

  • goxed
    Legendary
    *
    Offline Offline

    Activity: 1946
    Merit: 1006


    Bitcoin Crypto mining Hardware.


    View Profile
    September 16, 2013, 10:11:42 PM
     #112
    Quote from: EvilLizardApparel on September 16, 2013, 10:03:47 PM
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.
    My guess-
    The regulator autoshutting
    Revewing Bitcoin Crypto mining Hardware.

  • btc4life
    Newbie
    *
    Offline Offline

    Activity: 18
    Merit: 0


    View Profile
    September 16, 2013, 10:12:22 PM
     #113
    I just added some heatsinks to one of my cards today to test them out before I committed to buying enough for all 16 cards.

    They seem to fit the surface area of each chip perfectly.

    http://www.amazon.com/Cosmos-Aluminum-Cooling-Heatsinks-cooler/dp/B007XACV8O/ref=sr_1_1?s=electronics&ie=UTF8&qid=1379368738&sr=1-1&keywords=Cosmos+%C2%AE+20+PCS+mini+Aluminum+Chips

    http://i106.photobucket.com/albums/m264/atunicus/DSCN0803.jpg

  • zurg
    Hero Member
    *****
    Offline Offline

    Activity: 826
    Merit: 500


    Crypto Investor ;) @ Farmed Account Hunter


    View Profile
    September 16, 2013, 10:15:53 PM
     #114
    Quote from: EvilLizardApparel on September 16, 2013, 10:03:47 PM
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.

    What  did you set it to?
    Mine has been rock solid all day. Few rare errors, like 150 out of over 200k.

  • -Redacted-
    Hero Member
    *****
    Offline Offline

    Activity: 574
    Merit: 501


    View Profile
    September 16, 2013, 10:21:53 PM
     #115
    Quote from: btc4life on September 16, 2013, 10:12:22 PM
    I just added some heatsinks to one of my cards today to test them out before I committed to buying enough for all 16 cards.

    They seem to fit the surface area of each chip perfectly.

    http://www.amazon.com/Cosmos-Aluminum-Cooling-Heatsinks-cooler/dp/B007XACV8O/ref=sr_1_1?s=electronics&ie=UTF8&qid=1379368738&sr=1-1&keywords=Cosmos+%C2%AE+20+PCS+mini+Aluminum+Chips


    I thought these chips were designed to put heat out through the bottom, and that heatsinks should go on the back of the board over the thermal vias...

  • EvilLizardApparel
    Full Member
    ***
    Offline Offline

    Activity: 133
    Merit: 100



    View Profile
    September 16, 2013, 10:37:08 PM
     #116
    Quote from: goxed on September 16, 2013, 10:11:42 PM
    Quote from: EvilLizardApparel on September 16, 2013, 10:03:47 PM
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.
    My guess-
    The regulator autoshutting

    ....Is there a way to run it and not have it autoshutdown? It was hashing like a boss...Heat was fine...is it the auto-tune, or a hardware limit I'm hitting?

  • arorts
    Sr. Member
    ****
    Offline Offline

    Activity: 408
    Merit: 250


    View Profile
    September 17, 2013, 12:44:21 AM
     #117
    Quote from: EvilLizardApparel on September 16, 2013, 10:37:08 PM
    Quote from: goxed on September 16, 2013, 10:11:42 PM
    Quote from: EvilLizardApparel on September 16, 2013, 10:03:47 PM
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.
    My guess-
    The regulator autoshutting

    ....Is there a way to run it and not have it autoshutdown? It was hashing like a boss...Heat was fine...is it the auto-tune, or a hardware limit I'm hitting?

    Wow, how can it increase hashing so much while stating that "heat was fine"? Did you actually measure with a temperature probe placed on the board?
    The board is self-protecting itself from being fried and by pencil modding it you also just voided its warranty.

  • dacman61
    Member
    **
    Offline Offline

    Activity: 87
    Merit: 10



    View Profile
    September 17, 2013, 01:08:29 AM
     #118
    Quote from: zurg on September 16, 2013, 10:15:53 PM
    Quote from: EvilLizardApparel on September 16, 2013, 10:03:47 PM
    Using pencil trick, I had an H-Card go from 21 GH/s to 39 GH/s . However, after 60 seconds the hashrate slowly decreases to lower than 21 GH/s. Is this a result of the auto-tuning?

    EDIT: When miner is restarted, card goes back to 39 GH/s .. but then does the same thing.

    What  did you set it to?
    Mine has been rock solid all day. Few rare errors, like 150 out of over 200k.

    Make sure the you're reading no less than 1.1k Ohm on your multimeter. I screwed up the first time and got a reading of 1.08k which was too little resistance which caused some chips to shutdown and kill my hashrate... Then after redoing it, I got a reading of 1.18k. Since then, I've been rock solid between 32-34Gh/s for a day and a half now.

  • EvilLizardApparel
    Full Member
    ***
    Offline Offline

    Activity: 133
    Merit: 100



    View Profile
    September 17, 2013, 01:16:07 AM
    Last edit: September 17, 2013, 05:34:40 AM by EvilLizardApparel
     #119
    I have it running at 480 465 GH/s stable now. Thats a 400 GH/s kit. So I'm pretty happy with that...I'm staying at 1.2 .

  • darkfriend77 (OP)
    Sr. Member
    ****
    Offline Offline

    Activity: 434
    Merit: 265


    View Profile WWW
    September 17, 2013, 09:21:08 AM
     #120
    Added:

    Quote from: darkfriend77 on September 04, 2013, 12:17:20 PM
    7. Hashrate droping regularly, despite bitfury admin showing steady noncerate?

    Quote from: Trupik on September 17, 2013, 08:39:09 AM
    Someone may find this useful... If your pool reports your hashrate dropping regularly DESPITE bitfury admin showing steady noncerate:

    1) ssh into your pi
    2) run command: crontab -e
    3) add this line at the end of file:
    Quote
    */10    *       *       *       *       sudoopt/bitfury/start-stratumproxy.sh
    4) save & exit

    It will restart your stratum proxy every 10 minutes from now on. If you want to cancel it, just run the command again and delete the line.
    -:| www.DOTMog.com |:-
    Pages: « 1 2 3 4 5 [6] 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 »
      Print  
Page 5
Viewing Page: 6

IRLBTC™ © 2025 IRLBTC.com