Pages:
Author

Topic: BAMT version 0.5 - Easy USB based mining Linux with farm wide management tools - page 53. (Read 324169 times)

donator
Activity: 1218
Merit: 1079
Gerald Davis
Im propably the biggest noob who dares give advice here so you might want to wait for confirmation but to me it seems that your feeding stock voltage to a highly underclocked card. It may or may not be the source of your problems, but either way I'd lower the core voltage.

Core isn't underclocked.  Stock for 5970 is 725.
hero member
Activity: 910
Merit: 1000
Items flashing here available at btctrinkets.com
newbie
Activity: 47
Merit: 0
Hey guys, looking to further optimize my 5970 rig.  I was completely stable prior to the 0.5 release and now having to reboot my box twice daily to keep the cards from faulting.  I was pulling about 330 MHx2 and was hoping someone else who is using a similar card could share an idea or two on what I could to to push up the limit some.   I've also noticed that my second GPU's Mem is clocking at 1000 Mhz when I've told it to stay at 175.  Am I doing something stupid?

Here's my current configuration:

Code:
gpu0:
  disabled: 0
  # optional command to execute prior to overclocking
  pre_oc_cmd: /etc/init.d/munin-node restart

  #overclocking.. all optional 955/830
  fan_speed: 100
  core_speed_0: 730
  core_speed_1: 730
  core_speed_2: 730
  mem_speed_0: 175
  mem_speed_1: 175
  mem_speed_2: 175
  # core_voltage_0: 1.125
  # core_voltage_1: 1.125
  # core_voltage_2: 1.125000

  # optional command to execute after overclocking
  post_oc_cmd:

  # kernel to use.. currently one of:  phatk phatk2 phatk-bamt04 phatk-svn phatk-1.50 poclbm
  kernel: phatk2
  # phoenix style kernel params.  do not include DEVICE=x
  kernel_params: BFI_INT VECTORS FASTLOOP=false AGGRESSION=13 WORKSIZE=128

  # file containing pool URLs
  pool_file: /etc/bamt/pools
  # default time to allow phoenix to run without finding a share.. exceed this and we move to next pool in poolfile
  pool_timeout: 180

  # monitoring values, used by email alerts and gpumon
  monitor_temp_lo: 45
  monitor_temp_hi: 80
  monitor_load_lo: 80
  monitor_hash_lo: 125
  monitor_fan_lo: 2000
  monitor_reject_hi: 2

gpu1:
  disabled: 0
  # optional command to execute prior to overclocking
  pre_oc_cmd:

  #overclocking.. all optional
  # fan_speed: 80
  core_speed_0: 730
  core_speed_1: 730
  core_speed_2: 730
  mem_speed_0: 175
  mem_speed_1: 175
  mem_speed_2: 175
  # core_voltage_0: 1.088
  # core_voltage_1: 1.088
  # core_voltage_2: 1.088

  # optional command to execute after overclocking
  post_oc_cmd:

  # kernel to use.. currently one of:  phatk phatk2 phatk-bamt04 phatk-svn phatk-1.50 poclbm
  kernel: phatk2
  # phoenix style kernel params.  do not include DEVICE=x
  kernel_params: BFI_INT VECTORS FASTLOOP=false AGGRESSION=13 WORKSIZE=128

  # file containing pool URLs
  pool_file: /etc/bamt/pools
  # default time to allow phoenix to run without finding a share.. exceed this and we move to next pool in poolfile
  pool_timeout: 180

  # monitoring values, used by email alerts and gpumon
  monitor_temp_lo: 45
  monitor_temp_hi: 80
  monitor_load_lo: 80
  monitor_hash_lo: 125
  monitor_fan_lo: 2000
  monitor_reject_hi: 2
sr. member
Activity: 309
Merit: 250
Hey Guys,

Trying to get CGI miner to work, Looks like everything is set up and running, however I noticed that the load it usually around 70-80%. Before I switched to CGI, my load on my GPU would always be 99%. Any ideas?

What cards and intensity are you running in cgminer?  You might try setting a higher intensity to see if that increases GPU load.
hero member
Activity: 742
Merit: 500
yes sorry cgminer, and no the hash rate is now lower
Is cgminer setting your clocks?

Posting logs and your config would be helpful.
member
Activity: 66
Merit: 10
yes sorry cgminer, and no the hash rate is now lower
hero member
Activity: 742
Merit: 500
Hey Guys,

Trying to get CGI miner to work, Looks like everything is set up and running, however I noticed that the load it usually around 70-80%. Before I switched to CGI, my load on my GPU would always be 99%. Any ideas?
Do you mean cgminer?

Is your hashrate the same?
member
Activity: 66
Merit: 10
Hey Guys,

Trying to get CGI miner to work, Looks like everything is set up and running, however I noticed that the load it usually around 70-80%. Before I switched to CGI, my load on my GPU would always be 99%. Any ideas?
hero member
Activity: 910
Merit: 1000
Items flashing here available at btctrinkets.com
Quote
Im having this problem now.  Had a troublesome card lock up my rig, now I have it clocked much lower than its comrades.  

I have tried to delete the no oc file, like I have many times in the past. This time though I get "Unable to create trashing info file: Read-only file system".  I tried a few cold reboots, and ejecting the USB properly in windows like you suggested.  Still unable to delete the file.  I suppose I will re image the drive and try again this evening, since BAMT is so damn easy to install Wink  But any ideas would be appreciated.

After been given the tip about using a windows machine to "safeley eject " the usb-drive, I have not had to rewrite a single Bamt image. However I do suspect a rewrite would solve the roblem your having, do make sure you backup your bamt.conf and pools.conf to save time. I've wasted some time along the road hunting down typos when not doing this.

I'd like to also point out that I have since stopped locking up my rig and at current I have had it stable for a week, which might not sound that impressive to most of you guys untill you hear where it is: *drumroll*
I've moved it outdoors. The weater conditions seen so far range fron -8 to 15 celcius and everyting is looking good.
member
Activity: 546
Merit: 10
Mother went ahead on wrote me a noOCGPU0 file, happened before, will happen again untill my risers get here, the card in question is sitting over a rather hot northbridge of the mobo. Somehow the file came out as a read only, I couldnt delete it. chmod 777 noOCGPU0 did not work, after a reboot the file was deleteable tough, wonder what caused this ?

This happens when the filesystem isn't closed properly (and on really bad crashes like the ones where mother has to reboot the rig, its not uncommon).  On next boot it is mounted read-only. 

Sometimes it fixes itself on the boot after that.  Sometimes you have to stick the key in a Windows machine, "Safely Eject" it, and then it will be fine.

Best solution: quit locking up your rig Smiley



Im having this problem now.  Had a troublesome card lock up my rig, now I have it clocked much lower than its comrades. 

I have tried to delete the no oc file, like I have many times in the past. This time though I get "Unable to create trashing info file: Read-only file system".  I tried a few cold reboots, and ejecting the USB properly in windows like you suggested.  Still unable to delete the file.  I suppose I will re image the drive and try again this evening, since BAMT is so damn easy to install Wink  But any ideas would be appreciated.
hero member
Activity: 616
Merit: 506
 Weird, the card had no issues running at that speed for 4+ days straight since I picked it up.  

That is not weird at all.  It is extremely common for cards that are overclocked too much to appear stable for several days, even a couple weeks, and then lock up.
newbie
Activity: 59
Merit: 0
Pair of 5770s consuming 104 watts.  For what it is, the second card is a beast.


Why is GPU 0 showing no fan?

Not sure, but it doesn't report in windows either.  Likely a bad sensor, but it still scales the RPM fine.  I picked up the pair of 5770s for a total of $80 so I can't complain.

BTW total system power on that rig is ~148 watts, it's an atom board with a PCIE 16x and PCIE 1X slot, so I filled them both up, giving me 2.64 MH per watt, not too bad at all.
hero member
Activity: 742
Merit: 500
Pair of 5770s consuming 104 watts.  For what it is, the second card is a beast.


Why is GPU 0 showing no fan?
newbie
Activity: 59
Merit: 0
BAMT stopped undervolting one of my cards for some reason.  I cold restarted the computer and verified the configs had not changed.  Any ideas. I'm glad I noticed it relatively quickly, card was running at 84C!
 
GPU0 used to run at 1v 860 MHz, now no matter what it will only run at stock 1.2v 850 MHz and full RAM speed, any ideas?


EDIT: figured it out, apparently GPU 0 crashed and left some type of noOC file in /live/image/....

Deleted it and now it's working again, lowered the OC on that card from 865 to 850, see if that resolves the issue.  Weird, the card had no issues running at that speed for 4+ days straight since I picked it up.  



Pair of 5770s consuming 104 watts.  For what it is, the second card is a beast.

rjk
sr. member
Activity: 448
Merit: 250
1ngldh
Funny had same problem.   Is anyone here Litecoin mining on their BAMT rigs?  I am not well versed with linux and would like to get some of my quad cores LTC mining. Thanks for any help!
BAMT is 32-bit so you won't have much luck.

Why does it have to be 64-bit. I have an XP 32 mining. The miner runs a scrypt through each thread.
Because 64-bit is faster for scrypt calculations.

May be true but I would like to run it just to see on the couple BAMT rigs I got with Quad cores.
It should still run fine on 32-bit, just slower.
hero member
Activity: 497
Merit: 500
Funny had same problem.   Is anyone here Litecoin mining on their BAMT rigs?  I am not well versed with linux and would like to get some of my quad cores LTC mining. Thanks for any help!
BAMT is 32-bit so you won't have much luck.

Why does it have to be 64-bit. I have an XP 32 mining. The miner runs a scrypt through each thread.
Because 64-bit is faster for scrypt calculations.

May be true but I would like to run it just to see on the couple BAMT rigs I got with Quad cores.
rjk
sr. member
Activity: 448
Merit: 250
1ngldh
Funny had same problem.   Is anyone here Litecoin mining on their BAMT rigs?  I am not well versed with linux and would like to get some of my quad cores LTC mining. Thanks for any help!
BAMT is 32-bit so you won't have much luck.

Why does it have to be 64-bit. I have an XP 32 mining. The miner runs a scrypt through each thread.
Because 64-bit is faster for scrypt calculations.
hero member
Activity: 497
Merit: 500
Funny had same problem.   Is anyone here Litecoin mining on their BAMT rigs?  I am not well versed with linux and would like to get some of my quad cores LTC mining. Thanks for any help!
BAMT is 32-bit so you won't have much luck.

Why does it have to be 64-bit. I have an XP 32 mining. The miner runs a scrypt through each thread.
hero member
Activity: 837
Merit: 1000
Using the config standard file of BAMT , can i set a max temp like 80C for overheating ? And if any core reach this,  to stop mining ?
sr. member
Activity: 349
Merit: 250
lodcrappo,

Can a BAMT rig with mgpumon read out the stats of a non-BAMT rig with cgminer but API enabled?
Pages:
Jump to: