[GUIDE] BitFury Miner Support/Tuning

[arorts]

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]

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]

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]

Added:

7. Hashrate droping regularly, despite bitfury admin showing steady noncerate?

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:

*/10    *       *       *       *       sudo /opt/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.

[pbigmoon]

Does Bitfury have any stratum logs i cant seem to find any and i keep having problems with Bitminter, Overnight the speed always drops to around 1gh/s tho bitfury admin  still shows 40gh/s.

Edit

Ill try https://bitcointalk.org/index.php?topic=287590.msg3172013#msg3172013 see if it makes any difference

[evilscoop]

Does Bitfury have any stratum logs i cant seem to find any and i keep having problems with Bitminter, Overnight the speed always drops to around 1gh/s tho bitfury admin  still shows 40gh/s.

Edit

Ill try https://bitcointalk.org/index.php?topic=287590.msg3172013#msg3172013 see if it makes any difference

no id doesnt by default, but you can load up the screen instance (see screen -ls)

You could use tee on the startup to pipe to a logfile and the screen, I guess, but you aint got a lot of space there

[kano]

Pipe it through a socket to another machine?

[cscape]

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

The chips have 3 oscillator options: external, internal slow and internal fast. In addition, there's a /2 divider that can be switched on. Bitfury recommended the slow internal oscillator if it was fast enough. The /2 divider makes the clock slightly less jittery, and also allows smaller steps in clock speed.

The H-cards have an option for external clock, but it's not used on the m-board.

[spiccioli]

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

The chips have 3 oscillator options: external, internal slow and internal fast. In addition, there's a /2 divider that can be switched on. Bitfury recommended the slow internal oscillator if it was fast enough. The /2 divider makes the clock slightly less jittery, and also allows smaller steps in clock speed.

The H-cards have an option for external clock, but it's not used on the m-board.

Thanks,

so there is no problem in using internal fast if it works any better? Is the speed range the same 53-57?

spiccioli

[btc4life]

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...

Thanks I didn't realize the heat was designed to come out the back, I moved the heatsinks to the other side and the chips are staying much cooler now.

[cscape]

so there is no problem in using internal fast if it works any better? Is the speed range the same 53-57?

No, there's no problem. Not sure about the speed range. I would just experiment with different values and see what works best.

[dresdenreader]

I didn't see this in the thread so I thought I would ask.

Is there any specific power supply that people recommend?

I'll be using it in conjunction with one of the open air metal cases that Spotswood makes, so I'd like it to be able to fit in that.

[dracore]

Where can we find an image of the SD card?  My filesystem is messed (fsck currently trying to repair...) because I powered off the rpi too often without proper shutdown.  I'm worried I may need to reflash my SD card.

[darkfriend77]

Added:

...
Raspberry Pi SD Image
- 2013-Aug-29_BitfuryRasPi_v0.1.img (Magnet-Link)
...

...
8. How to update on the most actual chainminer version (gitpull)?

... first .. if it ain't broke don't fix it ...

do backup of your actual bitfury chainminer version...

Code:

pi@bitfury ~ $ cp -a /opt/bitfury /usr/bitfury.backup

Then move to chainminer get latest version clean make ... restart ...

Code:

pi@bitfury ~ $ cd /opt/bitfury/chainminer
pi@bitfury ~ $ git pull
pi@bitfury ~ $ make clean
pi@bitfury ~ $ make

restart the miner ... and hope it will still hash ... :-)
...

[-Redacted-]

Where can we find an image of the SD card?  My filesystem is messed (fsck currently trying to repair...) because I powered off the rpi too often without proper shutdown.  I'm worried I may need to reflash my SD card.

It's my understanding that US users with version 1 m-boards should not upgrade or compile from a git pull.  Make sure you use the right image...

[dracore]

Where can we find an image of the SD card?  My filesystem is messed (fsck currently trying to repair...) because I powered off the rpi too often without proper shutdown.  I'm worried I may need to reflash my SD card.

It's my understanding that US users with version 1 m-boards should not upgrade or compile from a git pull.  Make sure you use the right image...

But the image (.zip) provided on the 1st page should be ok for v1-m boards?

[americandesi]

I just posted dropbox links to the V1 & V2 rPi images and chainminer code if you guys need to remake your SD Cards.  Please be sure to use the correct image - especially make sure not to use the V2 image or chainminer code on a V1 board!

Check the OP for the support thread I posted (see my sig for link).

The M-boards we shipped out today are all V2, so remember that these boards use 4 distinct SPI channels.  If you have 4 hashing cards, you can load them in the slot 1 of each bank for best results.

Also be sure to use two separate PCI-e cables from your PSU to the plugs on the M-board.  Using one cable that has two plugs will just put extra load on that cable, possibly dangerous as you increase H-cards.

Tracking numbers should have come to you via email (when FedEx label was printed).  If not, I'll check with the shipping team to see whats up.

With regards to Dave's message above. If i have 2 EOL boards, does anyone know if i can connect them to the same M-board (v2.2) in separate SPI channels / banks.
Will that work.?

[-Redacted-]

Yes, it should.  Instead of a single SPI path, there are four completely separate "banks" of 4 boards each.

[americandesi]

Yes, it should.  Instead of a single SPI path, there are four completely separate "banks" of 4 boards each.

Perfect. Thanks.!

[Isokivi]

Do we know which chip is which # in the .stat.log ?