BAMT version 0.5 - Easy USB based mining Linux with farm wide management tools

[jamesg]

Sigh... looks very nice and pretty full of awesome. I wish I could get BAMT stable and functioning properly, but it constantly thinks GPUs are failed on every single rig I have, even though they run perfectly fine with CGMiner 24/7 for weeks on end, even using the BAMT version of CGminer.

What is the rough price per GPU?

You might want to talk to gigavps, he is running 30Ghash on bamt/cgminer.  The other large farms I know of are not using cgminer, but every report I've gotten about bamt/cgminer since we moved to 2.3.1 is that it works perfectly.  It seems there is a bug in older versions of cgminer that cause it to have issues if you actually use it's API.  Your problem descriptions support that.

Pricing for BAMT Pro services is based on which services you wish to use and the amount of data you wish to store with us.

Inaba, lodcrappo is correct. Ever since I moved to 0.5 with all of the fixes, my life has been a lot simpler. PM if you want to talk.

[1714054467]

1714054467

[Cobra]

GPUs you would like to monitor.

can i monitor GPU from windows 7 (cgminer)?

Yes, the cgsnoop feature to do that works great if your Win7 machine is using a current version of cgminer. I am getting much better performance with same config using version 2.1.2 instead of current. On a machine with 2 5970's and cgminer 2.1.2 on Win7, it will only show 2 gpu's in mgpumon. If I use the newest version it works great with all 4 gpu's showing up but not worth the drop in performance to monitor them. Curious if anyone else has seen anything similar? I know this is not a BAMT issue, just checking.

[Definit]

im in the same shoes as Inaba, i've tried numerous clock settings, nothing seems to help... i have 12 rigs each one has at least 1 gpu not recognized sometimes 2 gpu's...

maybe i'll have to converse with you as well Gigavps...

also, i'll donate for your help if i can get things figured out... and once things are running correctly i'll also donate towards the BAMT 

[asdlsd]

Yes, the cgsnoop feature to do that works great

Thx.

how reset data for "web" mgpumon or restart it?

[cmg5461]

Just upgraded to 0.5c today, working great!

However, I just stuck my sapphire 6870 in and I can't adjust the clocks unless I go into atitweak.  Also, My voltage is rock solid.  I have no idea how to lower it o.o

[BitMinerN8]

Just upgraded to 0.5c today, working great!

However, I just stuck my sapphire 6870 in and I can't adjust the clocks unless I go into atitweak.  Also, My voltage is rock solid.  I have no idea how to lower it o.o

I believe on 6xxx cards you'd have to mess with the BIOS. I have many 5xxx cards that OC without issue, all settings. My 6xxx ones though I can only change the clock and fan, memory and voltage are locked. YMMV it seems.

[lodcrappo]

Just upgraded to 0.5c today, working great!

However, I just stuck my sapphire 6870 in and I can't adjust the clocks unless I go into atitweak.  Also, My voltage is rock solid.  I have no idea how to lower it o.o

6870s change clocks fine here, but voltage is not possible.

[cmg5461]

Is there any way to check logs to see why my system hung?  Sometimes I need to cycle it, because it stops mining.  Its like it just freezes :/

I also get this:

Code:

root@rig-1:~# mother -v
mother starts (452 seconds since last run)
Another mother is already running. Bail out!

[lodcrappo]

Is there any way to check logs to see why my system hung?  Sometimes I need to cycle it, because it stops mining.  Its like it just freezes :/

reduce your overclocking

[cmg5461]

Is there any way to check logs to see why my system hung?  Sometimes I need to cycle it, because it stops mining.  Its like it just freezes :/

reduce your overclocking

Its crazy underclocked though :/

[lodcrappo]

Is there any way to check logs to see why my system hung?  Sometimes I need to cycle it, because it stops mining.  Its like it just freezes :/

reduce your overclocking

Its crazy underclocked though :/

so reduce the underclocking.  quit breaking it by messing with the clocks, basically.

[lodcrappo]

I also get this:

Code:

root@rig-1:~# mother -v
mother starts (452 seconds since last run)
Another mother is already running. Bail out!

when a GPU is locked up, all processes that attempt to communicate with it tend to hang.

a mother that has been around for 450 seconds is a sure sign of a locked up GPU.

run cards at stock, and problems very likely disappear.  if not, you've got defective hardware (or really, really poor cooling).

[cmg5461]

when a GPU is locked up, all processes that attempt to communicate with it tend to hang.

a mother that has been around for 450 seconds is a sure sign of a locked up GPU.

run cards at stock, and problems very likely disappear.  if not, you've got defective hardware (or really, really poor cooling).

Ok, will try.  I underclock my cards while I sleep for courtesy for my roommate.  Is there any way to find out which card it is?

I clock my 5970 down to 675/300 @ .95v
All temps are kept between 60 and 70

[lodcrappo]

Ok, will try.  I underclock my cards while I sleep for courtesy for my roommate.  Is there any way to find out which card it is?

Unfortunately no.  When a GPU is put outside of it's normal operating conditions and then locks up, all bets are off.  Something hung on GPU 2 might cause GPU 0 not to respond, or vice versa.  One GPU might lock up the entire bus.  There is no definitive way to tell which GPU "caused" the lock up.

In many cases, you will find that the problem is not as simple as one card being set too high or low, but that your system cannot remain stable due to the combined effects of several GPUs being slightly out of tolerance.  Changing the values of any one GPU might change the problem or even make it go away, but that doesn't mean that GPU was necessarily "bad" or the cause of the issue. 

Manually setting clocks is basically begging for strange system behaviors.  Don't be surprised when they happen

[NetworkerZ]

Hiho!

Flashed all my cards to unlocked BIOS versions. BAMT is running fine on 2 of my 3 rigs. The third rig doen't start mining. There is no entry in the messages or bamt.log I'm not sure what to do now, but perhaps BAMT has a problem with the onboard GPU HD 4200?!? I don't want to use it, I haven't configured, etc. But it is recogniced by the driver. Has someone an idea?

THX
NetworkerZ

[yxt]

disable in bios

[NetworkerZ]

Tried that already, but that's not possible. I have three options: UMA, Sideport, UMA+Sideport -> no "disable". Any more ideas?

Greetz
NetworkerZ

[Splirow]

lodcrappo,

Where is the mgpumon.css file? I searched the who thing and can't find it.

[skyhawk]

something is seriously wrong with your image or your usb key.

suggest to download again.

I think my keys are fine, tried 3, of two different brands, and tried burning the img several times. Also, the was no problem with 0.4.

I'll download again from a different source and report once I try.

there is no difference in boot between "0.4" and "0.5".  There is very little difference in anything between the last 0.4 fix and the first 0.5 image.

the numbering sequence change was simply an arbitrary decision designed to try and make a clean start with documentation and other administrative things.
from a code perspective, "version 0.4" and "version 0.5" means basically nothing, I just decided instead of doing fix #37 for "0.4" we would call it 0.5 fix 1.

I had the exact same problem - machine running fine on 0.4, refused to boot 0.5 for love or money. I eventually just wrote the image directly onto the otherwise-unused HDD in the machine and it boots off that. It's an old ASUS board, and one of the many reasons I don't buy ASUS boards anymore.

If you're not familiar with Linux/Unix this can be a bit tricky. Burn something like SystemRescueCD and boot off that, then plug in your flashed USB drive. You need to determine your flashdrive and HDD devices. Run ls -l /dev/disk/by-id This output shows the identifiers of your storage hardware and what devices they link to. I'm going to assume your HDD is /dev/sda and your flashdrive is /dev/sdb.

It should go without saying this operation will make the existing contents of the HDD unusable, without substantial effort to recover them. If you want to properly wipe the HDD first, run shred -n 0 /dev/sda

Run dd if=/dev/sdb of=/dev/sda bs=1m Be sure to ignore partition numbers - no /dev/sda1 or whatever. You want to write the entire contents of the image directly to the first sector of the HDD.

Flashing the image file itself from somewhere is more efficient and faster, but more complicated and hard to explain. I mounted an SMB share from my fileserver and flashed the imagefile with dd if=/mnt/server/bitcoin/BAMT/bamt_v0.5c.img of=/dev/sda bs=1m

[lodcrappo]

something is seriously wrong with your image or your usb key.

suggest to download again.

I think my keys are fine, tried 3, of two different brands, and tried burning the img several times. Also, the was no problem with 0.4.

I'll download again from a different source and report once I try.

there is no difference in boot between "0.4" and "0.5".  There is very little difference in anything between the last 0.4 fix and the first 0.5 image.

the numbering sequence change was simply an arbitrary decision designed to try and make a clean start with documentation and other administrative things.
from a code perspective, "version 0.4" and "version 0.5" means basically nothing, I just decided instead of doing fix #37 for "0.4" we would call it 0.5 fix 1.

I had the exact same problem - machine running fine on 0.4, refused to boot 0.5 for love or money. I eventually just wrote the image directly onto the otherwise-unused HDD in the machine and it boots off that. It's an old ASUS board, and one of the many reasons I don't buy ASUS boards anymore.

If you're not familiar with Linux/Unix this can be a bit tricky. Burn something like SystemRescueCD and boot off that, then plug in your flashed USB drive. You need to determine your flashdrive and HDD devices. Run ls -l /dev/disk/by-id This output shows the identifiers of your storage hardware and what devices they link to. I'm going to assume your HDD is /dev/sda and your flashdrive is /dev/sdb.

It should go without saying this operation will make the existing contents of the HDD unusable, without substantial effort to recover them. If you want to properly wipe the HDD first, run shred -n 0 /dev/sda

Run dd if=/dev/sdb of=/dev/sda bs=1m Be sure to ignore partition numbers - no /dev/sda1 or whatever. You want to write the entire contents of the image directly to the first sector of the HDD.

Flashing the image file itself from somewhere is more efficient and faster, but more complicated and hard to explain. I mounted an SMB share from my fileserver and flashed the imagefile with dd if=/mnt/server/bitcoin/BAMT/bamt_v0.5c.img of=/dev/sda bs=1m

where are you guys getting your image from?  I think one of the mirrors must have a corrupt copy.  there is absolutely no reason a machine would be able to boot the image called 0.4 but not the image called 0.5.  they are the same thing.