(OLD) BFGMiner: modular FPGA/GPU, GBT, Stratum, RPC, Avalon/Lnx/OpnWrt/PPA/W64

[hashking]

I'm running ubuntu and I'm trying to mine with more then one modminer.  When I start the program it only picks one up.  I'm using ./bfgminer at the command line.  Isn't the newer version supposed to autodetect the boards?

[1713301802]

1713301802

[1713301802]

1713301802

[1713301802]

1713301802

[1713301802]

1713301802

[hashking]

I'm running ubuntu and I'm trying to mine with more then one modminer.  When I start the program it only picks one up.  I'm using ./bfgminer at the command line.  Isn't the newer version supposed to autodetect the boards?

Is anyone able to answer this question?

[Luke-Jr]

I'm running ubuntu and I'm trying to mine with more then one modminer.  When I start the program it only picks one up.  I'm using ./bfgminer at the command line.  Isn't the newer version supposed to autodetect the boards?

Is anyone able to answer this question?

Are you running the latest firmware? Using BFGMiner from PPA or source?

[hashking]

I'm running ubuntu and I'm trying to mine with more then one modminer.  When I start the program it only picks one up.  I'm using ./bfgminer at the command line.  Isn't the newer version supposed to autodetect the boards?

Is anyone able to answer this question?

Are you running the latest firmware? Using BFGMiner from PPA or source?

It was built from source following the instructions outlined on on the wiki for the mmq.

[purelithium]

If you want to use more than one MMQ, don't you have to specify it with multiple uses of the -S flag and its arguments?

[Luke-Jr]

I'm running ubuntu and I'm trying to mine with more then one modminer.  When I start the program it only picks one up.  I'm using ./bfgminer at the command line.  Isn't the newer version supposed to autodetect the boards?

Is anyone able to answer this question?

Are you running the latest firmware? Using BFGMiner from PPA or source?

It was built from source following the instructions outlined on on the wiki for the mmq.

With older MCU firmwares, you need libudev-dev for autodetect to pick up more than one.

If you want to use more than one MMQ, don't you have to specify it with multiple uses of the -S flag and its arguments?

No, every device except Icarus and Lancelot support autodetection.

[salfter]

I went to check something on my mining rig and found that it was more sluggish than it should be.  top revealed that bfgminer was using an inordinately large amount of memory:

Code:

top - 09:06:31 up 8 days, 15:38,  1 user,  load average: 1.57, 0.51, 0.22
Tasks:  87 total,   1 running,  86 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.3%us,  0.5%sy,  0.0%ni, 99.0%id,  0.0%wa,  0.0%hi,  0.2%si,  0.0%st
Mem:    955744k total,   945944k used,     9800k free,      308k buffers
Swap:  2097148k total,   120000k used,  1977148k free,    10296k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
26545 salfter   20   0 1179m 812m 3172 S    1 87.1  39:04.07 bfgminer
...

It had only been running a couple of days.  I've had it running longer before, but this is the first time I've seen this problem:

Code:

bfgminer version 2.8.0 - Started: [2012-10-02 09:16:12] - [  1 day  23:50:28]
--------------------------------------------------------------------------------
 5s:161.9 avg:133.6 u:135.5 Mh/s | A:5434  R:46  HW:0  E:240%  U:1.9/m
 TQ: 0  ST: 2  SS: 0  DW: 196  NB: 273  GW: 2267  LW: 11715  GF: 91  RF: 1
 Connected to http://us1.eclipsemc.com:8337/ with LP as user salfter_lanbox
 Block: 000005527d3997046e82f85edbb2cc94...  Started: [09:06:35]
--------------------------------------------------------------------------------
 [P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
 OCL 0:  62.0C  39%    | 100.1/133.6/135.5Mh/s | A:5434 R:46 HW:0 U: 1.89/m
--------------------------------------------------------------------------------

 [2012-10-04 08:51:33] Accepted aa6cd27b.2bfb5b6c OCL 0 pool 0
 [2012-10-04 08:52:41] Accepted 2e1a56fa.7e86004e OCL 0 pool 0
...

I restarted it, and performance returned to normal:

Code:

top - 09:08:27 up 8 days, 15:40,  1 user,  load average: 0.33, 0.42, 0.22
Tasks:  87 total,   1 running,  86 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.2%us,  0.5%sy,  0.0%ni, 99.3%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:    955744k total,   241732k used,   714012k free,     6632k buffers
Swap:  2097148k total,     8216k used,  2088932k free,    86560k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
18706 salfter   20   0  381m  53m  21m S    1  5.7   0:02.37 bfgminer
...

I haven't updated to 2.8.1 yet.  On this machine, I'm GPU-mining with a Radeon 7750.

[Luke-Jr]

If you can, try running it in valgrind for a few minutes with the --leak-check=full valgrind option, --debuglog bfgminer option, and stderr piped to a file. Upload that file somewhere and link it in a new issue.

[FLHippy]

Announcing BFGMiner, the modular FPGA/GPU miner written in C. BFGMiner features overclocking, monitoring, fan speed control and remote interface capabilities.

I just compiled bfgminer on uBuntu to try it out. I have a couple of questions and a wow.

Questions that google didn't help me with...

  What is GCN
  what does GCN do
  why would I want to use it and...
  is it only supported by the diakgcn kernel?
 
WOW!!

   My rig is using 4-5% less power than with cgminer 2.6.5 and I'm getting the exact same hashrate. HOW?

[Luke-Jr]

What is GCN
  what does GCN do
  why would I want to use it and...
  is it only supported by the diakgcn kernel?

GCN is "Graphics Core Next" aka Radeon 7xxx cards. I guess not a big deal anymore, now that we're on FPGAs and soon ASICs...

My rig is using 4-5% less power than with cgminer 2.6.5 and I'm getting the exact same hashrate. HOW?

Can't say there's any one thing that stands out as an obvious power savings (and it may in fact not be any one thing by itself).

[soniq]

Anybody have  step by step instructions for a newb to connect a BFL single to BFGminer? assume I know nothing...I do know some but its not apparent right now

[Luke-Jr]

Anybody have  step by step instructions for a newb to connect a BFL single to BFGminer? assume I know nothing...I do know some but its not apparent right now

Just plug it in and run BFGMiner - it autodetects by default.
If you're using some broken Linux OS (notably Ubuntu), you may need to run a modprobe command (the exact command is in the README file).

[soniq]

Well I hoped it was that easy,  but BFG  detects all my GPU's and is more efficient than CG miner. But not detecting BFL single. Using windows 7.

[kano]

Well I hoped it was that easy,  but BFG  detects all my GPU's and is more efficient than CG miner. But not detecting BFL single. Using windows 7.

lol
Clearly you've made a mistake somewhere.

[Luke-Jr]

Well I hoped it was that easy,  but BFG  detects all my GPU's and is more efficient than CG miner. But not detecting BFL single. Using windows 7.

Can you run bfgminer -D -T -d? and post the output on http://pastebin.com ?

[Luke-Jr]

NEW VERSION 2.8.2, OCTOBER 8 2012

Human readable changelog:

• Numerous fixups for Enterpoint's Cairsmore dynamic clocking; it has actually been tested this time.
• Support for --temp-target and --temp-hysteresis controls on ModMiner FPGA devices.
• Generic failure management for all devices, including automatically attempting to restart dead devices.
• Improved portability to new platforms, partially including Cygwin.
• Various minor error handling improvements and bugfixes.

Full changelog

• Update to libblkmaker 0.1.2
• Bugfix: --temp-target no longer has a simple default (fixes build without OpenCL support)
• Bugfix: icarus: Silence false epoll error
• Bugfix: icarus: Set firstrun for errors starting next job, so the current one finishes properly
• Bugfix: icarus: Restore generic failure management for write errors
• Use strtod not strtol for bitforce temp backup.
• Cope with broken drivers returning nonsense values for bitforce temperatures.
• Minor warning fixes.
• Fix unused warnings on ming build.
• Fix sign warning in ocl.c
• fds need to be zeroed before set in modminer.
• Put scrypt warning on separate line to avoid 0 being shown on windows as bufsize.
• Prevent corrupt values returned from the opencl code from trying to read beyond the end of the buffer by masking the value to a max of 15.
• Icarus USB write failure is also a comms error
• api.c DEBUG message has no paramter
• Icarus catch more USB errors and close/reopen the port
• API-README update cgminer verison number
• hashmeter fix stats kh/s on 32bit windows
• cairnsmore: Increase maximum clock frequency to 210 Mhz
• icarus: Hashrate estimates really don't need the attention of a warning, demote them to debug
• cairnsmore: Automatically "downgrade" default FPGA-per-device to 1 for dynclock devices
• Bugfix: cairnsmore: Get autodetection of dynclock to work consistently
• cairnsmore: Adjust dynclock usage to react in proper time
• dynclock: Document function usage
• cairnsmore: Fix race on dynclock detection
• icarus: Detect attempts to send commands via work and neuter them
• cairnsmore: Glasswalker has a minimum multiplier of 20
• cairnsmore: Detect frequency changing support despite hashing of commands
• modminer: Allow clocks down to 2 Mhz just in case
• Allow device drivers and users to properly change target temperatures for non-GPUs
• Check that ncurses*-config installs actually work before deciding to use them
• Bugfix: Fix multiple bugs in autogen.sh
  • Don't use readlink -f unneccesarily (it's not portable)
  • Always run autoreconf within the real source directory
  • Run configure from PWD, *not* the real source directory
• Bugfix: Include nonce in data buffer for debugging
• Bugfix: swap32* wants count of 32-bit blocks, not bytes
• Initial Cygwin port
• Revert "Remove needless roundl define.", since it is needed for Cygwin and OpenWRT
• Bugfix: Deal with various compiler warnings
• modminer: Implement --temp-hysteresis logic
• Support for maximum frequency being below the default, eg when the maximum is temporarily reduced to deal with temperature
• Bugfix: modminer: Reduce dynclock max frequency as needed to keep temperature below cutoff
• Bugfix: Restore disabled label, needed to skip over hashrate calculations (which mess up otherwise)
• Bugfix: bitforce: Count actual throttling as hardware errors
• icarus: Allow failure in case of reopen failure, now that the miner core will retry on its own
• If a device dies, attempt to reinitialize it occasionally
• Bugfix: The REST flag is now preferred over WAIT, since the former might trigger the latter
• Bugfix: modminer: Update temperature readings when disabled (fixes thermal cutoff recovery)
• Bugfix: Move thermal cutoff to general watchdog code (fixes bitforce recovery)
• Rename enable_device to register_device, since it only works for setting it up at startup
• Move targettemp from ADL to cgpu_info, so all devices can readily use it
• Bugfix: "REST" flag had too much padding
• Bugfix: adl: Only warn and disable GPU due to thermal cutoff, if it's actually enabled
• Bugfix: bitforce: Only warn and disable bitforce due to thermal cutoff, if it's actually enabled

[bitpop]

Is it possible to support stratum? I'm currently using a proxy.

[crazyates]

Is it possible to support stratum? I'm currently using a proxy.

Cgminer supports stratum, but I don't know if Luke has any plans to support it or not. He is the author of the competing standard, GBT.

[bitpop]

Oh i see, i pretty much have to use stratum with bitcoincz for now.

[Luke-Jr]

Is it possible to support stratum? I'm currently using a proxy.

Better to use a pool that supports standard GBT, but Con wrote Stratum support for cgminer, and it seems like it isn't too much trouble to merge it in for some future BFGMiner version.