OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.1

[cablepair]

I just want to say in regards to 5870

I consider myself a bit of an expert when it comes to mining with theses cards as I have been doing so for almost a year now and have about 20 of them, they are excellent mining cards very efficient and stable , I have a 5870 that's been mining 11 months straight without any Problems no joke

In any case in regards to drivers and speed I have tested them all fairly extensively and ccc 11.9 with the Sdk 2.5.xx it comes with gives 430-445 mhash depending on the card with 950/180 clocks and the latest cgminer with stock kernel and 9 intensity

If any can prove me wrong (or right) I would love to see it, and would gladly give donation to anyone that provides a detailed benchmark

[1715250318]

1715250318

[JinTu]

when i do that with my 6970, it defaults back to stock settings.

I think this is an issue with the 6xxx series, not the miner software.

All 6000 series cards are restricted to GPU clock -125 for memclock.  For example, if you are running at 900 on GPU, the lowest you can clock the memory is 775.  Anything lower than 125 below the GPU clock will reset the memclock to stock speed.

Windows applications have extra backdoors to the hardware to get around this. We have no cross platform->linux code that can do this, nor do I have access to how it's done on windows. So afterburner and others can work around this limitation.

It is possible to run 6000 series cards at significantly lower memclock rates on Linux, but you need to be willing to reflash your cards. I currently run my dual 6990 rig at a memclock of 150 with a GPU clock of 600 (higher is possible, but cooling/fan noise is an issue with the placement of my rig). The big upside to running at a lower memclock is reduced power consumption, but this would only work if you have a dedicated mining rig.

[P_Shep]

After much headbutting the screen, walls and various other hard objects, I finally got cgminer to cross compile and run under a mipsel processor (for my Asus RT-N16).
My main problems was getting the cross-compiler to work; documentation on dd-wrt application development is very VERY sparse. in the end I discovered that optware (http://www.nslu2-linux.org/wiki/Optware/Platforms) source package includes a self-compiling compiler (though it's ancient).

ANYWAY.

With the options: --disable-opencl --disable-adl --enable-bitforce, I had to edit a couple of things in the code:

1. Added #define _GNU_SOURCE to bitforce.c (was having problems with pthread_rwlock_t)
2. Had to comment out the functional affine_to_cpu, leaving just the empty stub in main.c because my libraries evidently don't support processor affinity (Ancient compiler again, I guess). Perhaps a test for cpu_set_t could be done or something.

So, now cgminer returns with: 'All devices disabled, cannot mine!' Obviously I have no BFL units attached, but should it not scan serial ports or something? Is that the expected result?

Also, I've found cgminer configure expects curses.h to be directly in /include/. The default location however, seems to /include/ncurses/.

[kano]

However ... shares are random so the time to generate 10,000 of them can vary quite a lot on the same card each time you do it.

Please, Kano, I never told him to do "time cgminer --shares 10000 ...", did I? The MHash/s info can be extracted from the logs.
Still, good for you to have brought the potential issue up as I was admittedly sparse with implementation details.

OTOH, with high enough shares even this simplistic approach (that is, "time cgminer...") is acceptable due to the law of large numbers.
Syncing the benchmarking with difficulty change gives you plenty of time.
I'd repeat the benchmark across at least two difficulty changes to rule out the possibility of random packet storms biasing the results.

Yeah its just that using a share count was the option - OK I guess you didn't mean to actually take notice of the share count then
Then an easier way to do it on linux for a 1 hour run ...
 Window 1: cgminer --api-listen blah blah
 Window 2: sleep 3600 ; echo -n quit | nc -4 localhost 4028
OR
 Window 2: sleep 3600 ; java API quit

In windows you need some command for sleep ... not sure what or where to get one.

(however, law of large numbers doesn't come into it - they aren't large numbers)

[rcocchiararo]

im a failure :/

No worries, trial & error is a learning process - personally, I'd rather fail and learn than succeed without understanding why. Try going through this thread and experiment with different methods, then choose the one that works best for you.

I've posted this before in this thread but I guess every now and then it's helpful to refresh as it's a long topic and hard to read everything.

Assuming Ubuntu, which now uses Upstart the correct way to start cgminer at boot is with a conf file located in /etc/init. This doesn't require any auto-login or other stuff. Just one simple extra file in the right place. Here is a working example that I have used for many months,

/etc/init/miner.conf

Code:

description     "Start BTC Mining"

start on runlevel [2345]
stop on runlevel [016]
kill timeout 30
script
  sleep 15
  cd /home/ubuntu
  exec /usr/bin/screen -dmS Miner su -c /usr/local/bin/startcg ubuntu
end script

Note above it runs as user "ubuntu" but you should replace that with your user name. It is more secure to not run as root user, which would be the default during an init process like this. The "sleep" I have there is to allow time for X to be started at boot. There is a better way but this works fine too. I use "screen" to allow me to remote in with ssh and attach/detach as desired. If you're not familiar with that then google it. You can not bother with screen but then it's harder to login remotely and see realtime status. Also, if it's not already installed then "sudo apt-get install screen" would be needed before using this.

This conf file starts cgminer by running the startcg script. You can put what you need in there. Mine contains:

Code:

#!/bin/bash

export AMDAPPSDKROOT=/home/ubuntu/AMD-APP-SDK-v2.4-lnx32/
export LD_LIBRARY_PATH=${AMDAPPSDKROOT}lib/x86:${LD_LIBRARY_PATH}
export DISPLAY=:0

cgminer 2>>/var/log/cgminer.log

Yours may be a bit different but typically pretty similar to this. I use cgminer.conf but some people like to use the cmd line args in this file instead.

Useful things to know:
To start mining manually: sudo start miner
To attach to cgminer after remote login: sudo screen -r  
(I make an alias for this, alias mm='sudo screen -r' and put it in .bashrc file)
To detach from within cgminer: <ctrl>A D
You can disable starting at boot by renaming the conf file, eg. mv miner.conf miner.conf.down
(that's handy when changing GPU cards so they don't start with wrong settings).

Im runnung Debian 6

With the following script, i understand that i get the same thing you indicated:

Code:

#! /bin/sh
### BEGIN INIT INFO
# Provides:          cgminer
# Required-Start:    $all
# Required-Stop:     $remote_fs $syslog 
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: mining
# Description:       Start BTC Mining
### END INIT INFO

script
  sleep 15
  cd /home/rcocchiararo/bitcoin
  exec /usr/bin/screen -dmS Miner su -c /home/rcocchiararo/bitcoin/startcg.sh rcocchiararo
end script

then

Code:

#!/bin/bash
export AMDAPPSDKROOT=/opt/AMD-APP-SDK-v2.4-lnx32/
export AMDAPPSDKSAMPLESROOT=/opt/AMD-APP-SDK-v2.4-lnx32/
export LD_LIBRARY_PATH=${AMDAPPSDKROOT}lib/x86:${LD_LIBRARY_PATH}

export DISPLAY=:0

cgminer -o http://pit.deepbit.net:8332 -u rcocchiararo@gmail.com_alpha -p 12345 -o http://api2.bitcoin.cz:8332 -u rcocchiararo.alpha -p 12345 --auto-fan --auto-gpu --gpu-engine 950-975 --gpu-memclock 300 --gpu-fan 20-80 --temp-target 65 --temp-overheat 70 --api-listen --api-network -I 9

I originally used your same paths, but after many failures, i ended up changing them (for experimenting, not because they were wrong, in fact, they are perfectly valid for my installation)

First of all, im not familiar with "script", but i noticed a "typescript" file apearing here and there (changing places as i changed paths).

The most i could get (by chance) was to get cgminer to start in screen, but in cpu mining mode

that happened (by chance) automatically (i say by chance, because it was a one time thing, i don't know how i pulled it off, but if, and manually, by running:

Code:

/usr/bin/screen -dmS Miner su -c /home/rcocchiararo/bitcoin/startcg.sh rcocchiararo

exec from terminal just closes terminal.

If i run the last command i mentioned with my user (rcocchiararo) it starts cgminer in screen, with gpu mining.

If i do it with root (in a root terminal) i get cpu mining

Oh, and i use command line arguments, because either on windows or linux, even after saving the config, it ignores me (cgminer always asks for input)

[Red Emerald]

Im runnung Debian 6...

I gave up messing with init scripts that start screen sessions and put this in my /etc/rc.local instead

Code:

su - user -c "/path/to/cgminer/start/script.sh"

Oh, and i use command line arguments, because either on windows or linux, even after saving the config, it ignores me (cgminer always asks for input)

This is usually because you have a syntax error in your config.  You can see what the errors are if you specify the config.

Code:

cgminer -c ~/.cgminer/cgminer.conf

I don't know why it is built this way.  IMO, when you fail to parse a config, you should fail out, not act like there was no config.

[rcocchiararo]

Im runnung Debian 6...

I gave up messing with init scripts that start screen sessions and put this in my /etc/rc.local instead

Code:

su - user -c "/path/to/cgminer/start/script.sh"

Oh, and i use command line arguments, because either on windows or linux, even after saving the config, it ignores me (cgminer always asks for input)

This is usually because you have a syntax error in your config.  You can see what the errors are if you specify the config.

Code:

cgminer -c ~/.cgminer/cgminer.conf

I don't know why it is built this way.  IMO, when you fail to parse a config, you should fail out, not act like there was no config.

will try that tomoroy (just tried that in an init script and failed, and now i have to run xD)

weird if the config file is wrong, cause i made cgminer write it out, but oh well, i will see that tomorow also.

thx

EDIT: ok, no, i tried something like that but using screen

[BkkCoins]

Im runnung Debian 6

With the following script, i understand that i get the same thing you indicated:

Code:

#! /bin/sh
### BEGIN INIT INFO
# Provides:          cgminer
# Required-Start:    $all
# Required-Stop:     $remote_fs $syslog 
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: mining
# Description:       Start BTC Mining
### END INIT INFO

script
  sleep 15
  cd /home/rcocchiararo/bitcoin
  exec /usr/bin/screen -dmS Miner su -c /home/rcocchiararo/bitcoin/startcg.sh rcocchiararo
end script

This won't work as a sh script because the script keyword is an Upstart conf file function. You would want to take out both script/end script. I don't know offhand if Debian has Upstart but if it does then the best practice would be to use that by putting this as a conf file in /etc/init. If it doesn't then you could rewrite it as a script by removing the "script / end script" and taking out exec so it just calls screen. I don't think exec is a sh command either. Upstart conf file is not the same as a script.

The cgminer config writer doesn't write a fully working config unfortunately. You have to tweak some values - particularly zero ones. My version here does actually write a working config even with engine/memory values that work. But my version of the code is considered too dangerous by ckolivas to include in the mainline since it reads current ADL values to put in the config. There may be instances/cards where it causes problems - for my 5830s it works ok.

[-ck]

What you say about the config files should not apply with the current 2.2.1 version. It should cope fine with zero values, and redundant entries now.

[ssateneth]

Some people suggested to use CGMiner over phoenix, but I encountered a problem that should probably get fixed, as it's a deciding factor whether or not I use it.

Please support multiple OpenCL platforms. I have 2.1 + 2.5/2.6 on most of my miners, for compatibility reasons. The one computer I tried CMiner on (2.1 + 2.6) defaulted to just 2.6 and didn't even bother showing 2.1 SDK platform + gpu's, which is huge to me because 2.1 is so much faster. Phoenix and Diablominer both support multiple platforms, and as rich as this miner appears to be, it doesn't currently support multiple platforms.

Also does cgminer have phatk 2.2? Or the old slow phatk 1.0?

edit: certain people got their feelings hurt, reworded things a bit.

[kano]

Some people suggested to use CGMiner over phoenix. There is a glaring problem that needs to be fixed before I even consider using cgminer.

Please support multiple OpenCL platforms. I have 2.1 + 2.5/2.6 on most of my miners, for compatability reasons. The one computer I tried CMiner on (2.1 + 2.6) defaulted to just 2.6 and didn't even bother showing 2.1 SDK platform + gpu's, which is huge to me because 2.1 is so much faster.

Also does cgminer have phatk 2.2? Or the old slow phatk 1.0?

I'd suggest you find these people who suggested you use this crappy cgminer and shoot them.
You'll feel so much better

[QuantumFoam]

There's an annoying problem I've run into with multiple versions of cgminer including the latest release 2.2.1. If I quit out of cgminer and re run it later, it will no longer pick up the temperatures or fan rpms on the cards. I have to restart the machine for it to work again. This is on xubuntu 11.04. Any ideas of where to start on trying to fix this, or do I just have to always reboot whenever exiting cgminer before running it again.

[Proofer]

This reflects a post-2.2.1 version that I happened to download and which failed to run; the problem is fixed by reverting to the 2.2.1 release.  If this report is not useful, just ignore it

The problem version is:
 53c1e9a Allow the OpenCL platform ID to be chosen with --gpu-platform.

On starting cgminer (using p2pool):

Code:

[2012-02-03 21:29:46] Failed to init GPU thread 0, disabling device 0
[2012-02-03 21:29:46] Restarting the GPU from the menu will not fix this.
[2012-02-03 21:29:46] Try restarting cgminer.
Press enter to continue:
[2012-02-03 21:29:46] No long-poll found on any pool server

after hitting enter:

Code:

 cgminer version 2.2.1 - Started: [2012-02-03 21:29:46]
--------------------------------------------------------------------------------
 (5s):0.0 (avg):0.0 Mh/s | Q:1  A:0  R:0  HW:0  E:0%  U:0.00/m
 TQ: 1  ST: 1  SS: 0  DW: 0  NB: 1  LW: 0  GF: 0  RF: 0
 Connected to http://127.0.0.1:9332 without LP as user Proofer
 Block: 00000c85b0e184ef6717d1354c307e89...  Started: [21:29:46]
--------------------------------------------------------------------------------
 [P]ool management [S]ettings [D]isplay options [Q]uit
 GPU 0:  35.5C 4577RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
 GPU 1:  38.5C 4577RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
 GPU 2:  35.5C 4604RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
 GPU 3:  31.0C 4615RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
 GPU 4:  33.5C 4504RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
 GPU 5:  35.0C 4504RPM | OFF  /  0.0Mh/s | A:0 R:0 HW:0 U:0.00/m I: 0
--------------------------------------------------------------------------------./start.sh: line 6:  1395 Segmentation fault      DISPLAY=:0 cgminer -c cgminer.conf 2> logs/$now.log

The above was from the second run with the post 2.2.1 version, after deleting the .bin after the first run; same output as from the first attempt and the .bin is absent.

[Proofer]

I'm going to start here, but feel free to push me over to the Pools/p2pool topic.

I have some conceptual confusion/ignorance about using cgminer with p2pool.

(1) What metrics should I be looking at to tune parameters?

(2) What shares are cgminer reporting as accepted?  As I (mis?)understand it, cgminer shares are traditional difficulty=1 pool shares, but p2pool shares are different, having a higher difficulty.  What is "accepting" the shares that cgminer reports as accepted?

(3) I have been running with default intensity.  Now with 2.2.1 that means a default of 1 thread per GPU rather than the former 2.  With p2pool, would this be good? bad? don't know?

(4) I have assumed that cgminer's donation feature won't work with p2pool and that donations to CK must therefore be manual.  Is that right?

Edit: (1) and (2) are related.  With traditional pools, cgminer's U statistic (accepted/minute) seems to be the gold standard for tuning.  But with p2pool...?

[-ck]

This reflects a post-2.2.1 version that I happened to download and which failed to run; the problem is fixed by reverting to the 2.2.1 release.  If this report is not useful, just ignore it

The problem version is:
 53c1e9a Allow the OpenCL platform ID to be chosen with --gpu-platform.

Did you do ./autogen.sh before reconfiguring and rebuilding?

What does "./cgminer -n" show?

[BkkCoins]

I'm going to start here, but feel free to push me over to the Pools/p2pool topic.

I have some conceptual confusion/ignorance about using cgminer with p2pool.


(2) What shares are cgminer reporting as accepted?  As I (mis?)understand it, cgminer shares are traditional difficulty=1 pool shares, but p2pool shares are different, having a higher difficulty.  What is "accepting" the shares that cgminer reports as accepted?

Isn't this the same as solo mining with cgminer? The shares are accepted by bitcoind but that would happen rarely?

But I'm unclear myself on what the P2pool client does in the mix. Does it tell bitcoind that the difficulty is lower than normal and alter how work is handled?

[-ck]

I'm going to start here, but feel free to push me over to the Pools/p2pool topic.

I have some conceptual confusion/ignorance about using cgminer with p2pool.

(1) What metrics should I be looking at to tune parameters?

(2) What shares are cgminer reporting as accepted?  As I (mis?)understand it, cgminer shares are traditional difficulty=1 pool shares, but p2pool shares are different, having a higher difficulty.  What is "accepting" the shares that cgminer reports as accepted?

(3) I have been running with default intensity.  Now with 2.2.1 that means a default of 1 thread per GPU rather than the former 2.  With p2pool, would this be good? bad? don't know?

(4) I have assumed that cgminer's donation feature won't work with p2pool and that donations to CK must therefore be manual.  Is that right?

Edit: (1) and (2) are related.  With traditional pools, cgminer's U statistic (accepted/minute) seems to be the gold standard for tuning.  But with p2pool...?

1. MH/s as reported in the average column is still the only objective measure. Provided you don't have any problem communicating with your pool/server it should tell you ultimately how well it's tuned.
2. No idea with p2pool, you'll have to ask a p2pool expert.
3. Default intensity is "dynamic" meaning it's meant to be used on a desktop with a GUI. If this does not describe your setup, use a static intensity, usually 7-9 provides the best, depending on hardware. Try and find the largest value that does not fluctuate too much. Dynamic works fine regardless of what pool you're using it on, provided that's what you want.
4. It should still work fine.

[Proofer]

This reflects a post-2.2.1 version that I happened to download and which failed to run; the problem is fixed by reverting to the 2.2.1 release.  If this report is not useful, just ignore it

The problem version is:
 53c1e9a Allow the OpenCL platform ID to be chosen with --gpu-platform.

Did you do ./autogen.sh before reconfiguring and rebuilding?

Yes.

What does "./cgminer -n" show?

Too late for that as I reverted to the first commit that had the 2.2.1 version number.  But if you think it would be helpful, I can re-pull/build the latest commit and then try the -n.  Oh, if you just want to know what's installed: with the first 2.2.1 release (what I'm currently running) it outputs:

Code:

[2012-02-03 22:48:38] CL Platform vendor: Advanced Micro Devices, Inc.
[2012-02-03 22:48:38] CL Platform name: AMD Accelerated Parallel Processing
[2012-02-03 22:48:38] CL Platform version: OpenCL 1.1 AMD-APP-SDK-v2.4 (595.10)
[2012-02-03 22:48:38] GPU 0 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] GPU 1 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:48:38] GPU 2 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] GPU 3 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:48:38] GPU 4 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] GPU 5 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:48:38] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:48:38] Dual GPUs detected: 1 and 0
[2012-02-03 22:48:38] Dual GPUs detected: 3 and 2
[2012-02-03 22:48:38] Dual GPUs detected: 5 and 4
[2012-02-03 22:48:38] 6 GPU devices detected

[-ck]

No, I need -n from the last commit to see if/why it fails. I can't reproduce it... No offence, but are you sure you did ./autogen.sh first?

[Proofer]

No, I need -n from the last commit to see if/why it fails. I can't reproduce it... No offence, but are you sure you did ./autogen.sh first?

No offense taken; I'm sure.  After that I did:
CFLAGS="-O2 -Wall -march=native" ./configure
make clean
make

Here's the -n output from the latest commit, the one that fails to run:

Code:

[2012-02-03 22:56:14] CL Platform 0 vendor: Advanced Micro Devices, Inc.
[2012-02-03 22:56:14] CL Platform 0 name: AMD Accelerated Parallel Processing
[2012-02-03 22:56:14] CL Platform 0 version: OpenCL 1.1 AMD-APP-SDK-v2.4 (595.10)
[2012-02-03 22:56:14] Platform 0 devices: 6
[2012-02-03 22:56:14] GPU 0 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] GPU 1 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:56:14] GPU 2 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] GPU 3 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:56:14] GPU 4 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] GPU 5 ATI Radeon HD 5900 Series hardware monitoring enabled
[2012-02-03 22:56:14] Failed to ADL_Overdrive5_FanSpeedInfo_Get
[2012-02-03 22:56:14] Dual GPUs detected: 1 and 0
[2012-02-03 22:56:14] Dual GPUs detected: 3 and 2
[2012-02-03 22:56:14] Dual GPUs detected: 5 and 4
[2012-02-03 22:56:14] 6 GPU devices max detected

...which (if memory serves) is the same as what -n produces from the 2.1.2 version number commit, which is running fine.