Topic: BAMT - Easy persistent USB key based linux for dedicated miners/mining farms - page 7. (Read 167479 times)

Anyone getting BAMT to work on GA-990XA-UD3 bios version F8 version 1.1? I can't seem to boot from the USB, downloading linuxcoin but I would rather have BAMT working, there is a BIOS update that supports
Support Quick Boost function which is version F10G but I need a workable USB drive to update the BIOS? can anyone confirm that this will fix my issue to get BAMT to boot?

Update on my situation:

Found out the 5850 is using 50% cpu (two cores)

My box is 100% updated.  Has anyone had this bug before?

I gave up trying to get my two wireless cards working and went with a wireless bridge. Works great and I don't have to worry about changing anything with BAMT to make it work.

I don't think BAMT would be a good choice for this situation.  There is some wireless support but it requires linux experience to make use of it.

On that note, if anyone knows of a way to make wireless stuff easier for non linux folks, I'd be happy to include it in the upcoming 0.5 image if it doesn't cause problems.

If you get it working let me know.  I believe I followed it correctly.  Compiled without errors and it now can see the cards but still unable to change any values.

BAMT includes it and the source code so it is a matter of just editing and compile ... but it didn't work.

Ok, thanks for the fast reply. I stumbled upon this webpage http://allmybase.com/2012/01/25/overvolting-the-ati-radeon-5970-in-linux-with-radeonvolt/ a while back where it seems someone managed to use radeonvolt to overclock the 5970, so I guess I'll have to give it a try when I can spare some time.

I wonder how (or if) windows tools which allow overvolting would work under wine.

I don't believe there is any linux tool (not just limitation of BAMT) which enables overvolting.

There is RadeonVolt but it doesn't work with 5900 series.  If you want to overvolt in Linux you are looking at a custom BIOS flash. 

Hi,

I'm trying to overvolt a reference 5970 in BAMT, but cannot get it to work.
I have tired every possible variation in bamt.conf I can think of, but nothing seems to work.
Using atitweak I can get the card to undervolt, but overvolting just seems impossible...
Is there some sort of built in lock that stops setting vcore higher than 1.05?

shutdown -h worked.  I will keep in mind shutdown -h -p for next time.

The pump did fail but GPUs look fine, and computer POSTs.  Can't check mining capabilities until I get a replacement pump.

Still I feel stupid.  I have for months been considering getting a backup pump and a flow meter (which plugs into CPU fan header so BIOS will see low flow as a fan failure and power did system).

So I got a new 5850 today, and my load is staying at ~80%

I copied the config over and removed all overclocking besides memclock.  What am I doing wrong? o.o

If that does not work for you, I use: shutdown -h -P now
I have a couple rigs that just hang at the end of shutdown procedure and don't actually turn off if they don't have the -P option in there.

Its not hung (and no hung processes).  temps are only 78C right now.  At idle cards are only putting out 40W so it takes some time to raise temps.

Will try shutdown -h  thanks.

i use "shutdown -h now" to shutdown a rig. If the rig is hung tho, that may not work.

urgent:
command to power off (not reboot) BAMT rig?

The reason I ask is I know BAMT needs to use coldreboot instead of reboot.

I need to power off a BAMT rig (I believe watercooling pump failed).
cgminer killed the GPU when their temp went >60C but with no pump even idle they will eventually overheat.

can "shutdown" be used?  I don't want to take a chance and have it hang.  BIOS should kill the board when CPU reaches overheat temp but I would prefer to power it off for sure (BIOS could hang).

And an explanation for anyone who uses BAMT and wants 'quit' to work:

The next version (or current git) will be API version 1.2

With the new version you must include "--api-listen --api-allow W:127.0.0.1" when you start cgminer if you want the local machine to be able to send cgminer the 'quit' command
(The 'W:' means allow 127.0.0.1 to have write/priviledged access to cgminer)

The next verison will still allow all read access to cgminer as any current version does without changing the options you start cgminer with.

As mentioned above in my quote, the 'quit' command will also be fixed in the next release.

Also for lodcrappo:
And what is an unavoidable bug in probably every piece of API usage software, that you can fix with 1.2:
At the moment you probably look for "MHS 5s" to find the 5s average.
That will actually fail if anyone starts cgminer with the --log option with some value other than 5.
You can find the correct value for '5' with the 'config' command in the new 1.2 "Log Interval" - which of course will be '5' unless someone runs cgminer with the --log option.
(The --log option has been in cgminer for a long time)
So if "Log Interval" is "10" then you should look for "MHS 10s" not "MHS 5s"

Edit: and in case you didn't spot the obvious:
"quit\n" didn't (and never will) work.
"quit" always has been accepted by the API to mean quit cgminer - but stopped working due to a problem with the api.c doquit() function - however, the command "quit" was being accepted and attempting to do what was requested.
You should not be adding extra characters on the end - as your normal reporting interface shouldn't do that either (if it does)

I've never used anything but windows on my dedicated miners. My mining rigs are all on wireless adapters, will I be able to install drivers to get them working? I know they support ubuntu.

I've put a pull request to ckolivas to fix 'quit'
I guess it will be there once he accepts the pull (or if anyone for some unknown reason needs it urgently ... it's in my git or look at the pull request)
It seems it hasn't worked for a while

Edit: and it's now in ckolivas' git.

it just doesn't.  no one knows why.  part of bamt's mystery

Could someone with some BAMT knowledge answer this:

https://bitcointalksearch.org/topic/m.760428

and the reply:

https://bitcointalksearch.org/topic/m.760487