BAMT - Easy persistent USB key based linux for dedicated miners/mining farms

[hugolp]

I'm using the exact same MB.  

Please, send a report to Gigabyte. They will probably ignore us, but nevertheles if they receive a bunch of complains the might solve it. If we dont tell them anything they sure wont solve it.

[mmartoccia]

I'm using the exact same MB.  

Please, send a report to Gigabyte. They will probably ignore us, but nevertheles if they receive a bunch of complains the might solve it. If we dont tell them anything they sure wont solve it.

I plan to as soon as I know what the differences are.  Something's up with the way the boot is handled between UNetbootin and that which is used in BAMT's .IMG.  I dont think tech support will get anywhere without know what the problem is.

[fitty]

Rebuilding the USB via Terminal on OS X:

Code:

dd if=bamt_v0.4b.img of=/dev/disk1 bs=8192

Same result as doing it with Win32 Disk Imager:

Code:

Loading Operating System......
Boot Error

Need to determine what the boot differences are between Linuxcoin and that of BAMT.

I'd delete all the existing partitions on the usb drive. Format it. Then try with win32 diskimager. I had a usb drive I was using for ubuntu, then bamt, then bamt again, somewhere the drive got messed up nothing would work. I deleted all the partitions, new format and it was back to new. I used diskpart

http://www.windows-password-reset.com/create%20bootable%20usb%20drive.aspx

[fitty]

Based on the relatively pathetic amount of donations, I really don't think I'll be putting time into BAMT going forward, certainly not enough to justify the time it would take to redesign to a single partition.

Donation sent. That will put you back to work for at least 5 minutes. Maybe 4.

[mmartoccia]

Rebuilding the USB via Terminal on OS X:

Code:

dd if=bamt_v0.4b.img of=/dev/disk1 bs=8192

Same result as doing it with Win32 Disk Imager:

Code:

Loading Operating System......
Boot Error

Need to determine what the boot differences are between Linuxcoin and that of BAMT.

I'd delete all the existing partitions on the usb drive. Format it. Then try with win32 diskimager. I had a usb drive I was using for ubuntu, then bamt, then bamt again, somewhere the drive got messed up nothing would work. I deleted all the partitions, new format and it was back to new. I used diskpart

http://www.windows-password-reset.com/create%20bootable%20usb%20drive.aspx

Done and done.  Same result.  I can rebuild Linuxcoin or a fresh Debian using UNetbootin with success on this USB.  It's only happening with this .IMG.  Its absolutely bizarre.  Calling Gigabyte to find out specifics.

[fitty]

Done and done.  Same result.  I can rebuild Linuxcoin or a fresh Debian using UNetbootin with success on this USB.  It's only happening with this .IMG.  Its absolutely bizarre.  Calling Gigabyte to find out specifics.

So when you write the image you can browse the folders, see the files, all that jazz? It's just when it tries to boot from it that it chokes and dies?

I also assume you updated the bios. I didn't read back.

[mmartoccia]

So when you write the image you can browse the folders, see the files, all that jazz? It's just when it tries to boot from it that it chokes and dies?

I also assume you updated the bios. I didn't read back.

I've tried to browse the folders once the image was written both using dd and Win32 Disk Imager.  On my mac I cannot mount the drive once its been built but on windows I am able to browse everything.

Also, I just got off the phone with Gigabyte TS and I've pointed them to the image file for internal testing.  TS says they are about to release a new BIOS update in the next 2 weeks and hopefully they will be able to slide a solution in provided they can mirror the same issue.  

UPDATE I created the BAMT Flash drive, copied the files from the first partition (I cannot access the 2nd/3rd in Windows), erased and repartitioned, rebuilt the LinuxCoin on the same flash, copied everything out of the dir and replaced it with all the files from BAMT.  It boots and loads!  However, its a base Debian without any of the miners, bitcoin, atitweak, etc... I feel like I am getting closer.  

Not being able to browse the file structure on my mac once the BAMT USB is built is odd.

UPDATE 2  Gave up on making the USB -- way too time consuming and above my technical skills.  Replaced the Gigabyte MB with a new MB that widely accepts the boot loaders/file structures.  Back in action!

[KBundy]

Hello there,

Tried to install this onto my USB drive, Booted it up, don't think it worked, tried to reformat the drive to get rid of it and now my drive says its only 799mb.  Is there a reserved sector of the drive somehow lost or corrupt now?  Sorry, bit of a linux noob probably shouldn't have tried this without more research.

-KBundy

[fitty]

Hello there,

Tried to install this onto my USB drive, Booted it up, don't think it worked, tried to reformat the drive to get rid of it and now my drive says its only 799mb.  Is there a reserved sector of the drive somehow lost or corrupt now?  Sorry, bit of a linux noob probably shouldn't have tried this without more research.

-KBundy

I used diskpart

http://www.windows-password-reset.com/create%20bootable%20usb%20drive.aspx

Just make sure you use diskpart on the right drive (flash drive). you'll be wiping the contents.

You can also google "recover space flash drive partition" which should give you lots of solutions.

[The--Captain]

Uncomment out the line that prints the phoenix cmdline in wrapper.pl, restart mining, then attach using 'screen -d -r gpu0' to see the cmdline that is being used to start phoenix.  Compare that with the command line you have found that works.  You might be forgetting to specify a WORKSIZE, which that kernel seems to need.  Now donate to lodcrappo.

I am specifying a WORKSIZE. And I have donated to lodcrappo the day after I started using BAMT, over 2 weeks ago.
After uncommenting the CMD printing and restarting mining, 'screen -d -r gpu0' returns 'There is no screen to be detached matching gpu0.' Similar for gpu1.
(if I use the same steps with the older kernel, I do get the command line printed before the normal miner output - but with the new kernel it looks it doesn't even get there).

I don't know what you're doing that causes gpumon to fail and never attach, but those kernels worked for me once I specified a WORKSIZE - if I didn't, they threw an error.  Still, it's perl, after all.  It's not that hard to debug...  As lodcrappo says, I can't help you further because I can't replicate your problem.

Oh, and I meant donate *again* ;-)

[fitty]

Uncomment out the line that prints the phoenix cmdline in wrapper.pl, restart mining, then attach using 'screen -d -r gpu0' to see the cmdline that is being used to start phoenix.  Compare that with the command line you have found that works.  You might be forgetting to specify a WORKSIZE, which that kernel seems to need.  Now donate to lodcrappo.

I am specifying a WORKSIZE. And I have donated to lodcrappo the day after I started using BAMT, over 2 weeks ago.
After uncommenting the CMD printing and restarting mining, 'screen -d -r gpu0' returns 'There is no screen to be detached matching gpu0.' Similar for gpu1.
(if I use the same steps with the older kernel, I do get the command line printed before the normal miner output - but with the new kernel it looks it doesn't even get there).

Can you run phoenix manually from the command line with the new kernel? I assume that works fine it just won't work with gpumon?

[lodcrappo]

Hello there,

Tried to install this onto my USB drive, Booted it up, don't think it worked, tried to reformat the drive to get rid of it and now my drive says its only 799mb.  Is there a reserved sector of the drive somehow lost or corrupt now?  Sorry, bit of a linux noob probably shouldn't have tried this without more research.

-KBundy

https://github.com/aaronwolfe/Big-A-Miner-Thing/wiki/Faq

[jh1523]

Can you run phoenix manually from the command line with the new kernel? I assume that works fine it just won't work with gpumon?

Yes I can run phoenix from the command line with the new kernels just fine. I don't care what gpumon does. It's just that the BAMT mining scripts don't start the phoenix miner correctly with the new kernel.

Don't worry, I'm done looking for help here. I'll fix it myself even though I've never used perl before (perl isn't the end of all things you know, some people use other languages).

[lodcrappo]

Can you run phoenix manually from the command line with the new kernel? I assume that works fine it just won't work with gpumon?

Yes I can run phoenix from the command line with the new kernels just fine. I don't care what gpumon does. It's just that the BAMT mining scripts don't start the phoenix miner correctly with the new kernel.

are you sure you didn't overwrite phoenix itself, instead of just creating a new directory under kernels and putting the kernel you want to use there? all your symptoms really sound like you did.   the captain is reporting that if the correct procedure is followed, the kernel you want to use does work fine in bamt (although it seems to give poor hash rates compared to the provided customized kernels).

 if you overwrote any part of the phoenix program supplied with bamt, you will have no joy with bamt.  the supplied phoenix is *not* a stock version and has many modifications which are expected by the wrapper script and required if you're going to use bamt at all.

Don't worry, I'm done looking for help here. I'll fix it myself even though I've never used perl before (perl isn't the end of all things you know, some people use other languages).

I'm not really sure what that means.

[oo-oo]

hi, i have a rig 5 x 5850 with BATM, but i have a big problem... if i overclock the 5850s more than 800/900, the rig fails, stop working and i have to reboot the system... Its any solution for this ?? Becouse im only getting 1,5 Ghas with a 5 x 5850s rig, wich is realy realy a low average and the last thing i wanna do is change this exelent OS...

[lodcrappo]

hi, i have a rig 5 x 5850 with BATM, but i have a big problem... if i overclock the 5850s more than 800/900, the rig fails, stop working and i have to reboot the system... Its any solution for this ?? Becouse im only getting 1,5 Ghas with a 5 x 5850s rig, wich is realy realy a low average and the last thing i wanna do is change this exelent OS...

the operating system has very little, if anything, to do with the maximum clock rates your cards will be stable at.  with 5 gpus in one machine, you are very likely to have heat and/or power problems.  run it with 1 card, see how stable it is and what clock rates/hash rates are possible.   once one card is working properly, worry about adding the rest.

also, older cards like 5xxx series will probably get slightly better hashrates using the phatk-bamt04 kernel instead of default phatk.

[oo-oo]

Thanks, i will  try  one by one to see what the problem is. I know that it isn't heat a power problems.

BTW, the web interface to monitoring the rig sometimes load correctly and sometimes it doesn't, why is this ?

[lodcrappo]

Thanks, i will  try  one by one to see what the problem is. I know that it isn't heat a power problems.

Have a look back through this thread, you're hardly the first who "knows" it isn't their rig that is the problem... so far, it basically always turns out that it is the rig in situations like this.  Maybe you are the first to break this trend, but I wouldn't bet on it.

BTW, the web interface to monitoring the rig sometimes load correctly and sometimes it doesn't, why is this ?

Am I supposed to just guess?  Cosmic rays released from the sun are causing interference with your processor.

If you provide details about exactly what happens when it doesn't work for you, including logs and any errors messages, we might be able to give a better answer.

[lodcrappo]

Fix #14

Phoenix updated to v1.6.1 (with special BAMT modifications)

Recommend all miners try the new 'phatk2' kernel provided in this version.  Seems to provide a bit better hash rates than any of the kernels included previously on a wide range of cards.

Just edit bamt.conf and set:

  kernel: phatk2

in each GPU you want to try it on.

[gnar1ta$]

Thanks, phatk2 +3 on my 6950's.  Would it be a lot of work to add clocks to gpumon?