Smartcoin Linux mining administration. [MULTI-MACHINE SUPPORT NOW IN!]

[jondecker76]

dishwara:
the python <#path#> part is normal. Its not asking for a path, its just a tag like the others and part of the launch string (it gets replaced dynamically with the path to the phoenix.py file)
Have you tried a reboot, or deleting all the smartcoin* files in /tmp?  This seems to be a common problem lately, but without using /tmp, then thumb drive installs get hammered too much.  It seems that for some reason, once one of the temporary files is written to /tmp with the wrong permissions, then future attempts fail.  Have you done a full reinstall?  I'm aware that I need to  update the main post to reflect how to install/uninstall, etc - I've just been very busy and haven't had the time yet.


irishmick: 
Sounds like you had the problem with /tmp as well, which is probably why a reboot cured your problem!

Alan - I'll add your suggestion on to the stack - its a good idea, thanks

[1713488069]

1713488069

[1713488069]

1713488069

[1713488069]

1713488069

[jondecker76]

New update available:
r293 fixes a small bug that is the likely cause of people having problems of not being able to overwrite files in /tmp, causing the status screen to show all miners as down. Additionally, the temp files will be purged when smartcoin is started.

[dishwara]

Can you give here complete uninstall & reinstall , step by step procedure?
That may at least help me.
Also, i am not only seeing DOWN, but also my temperatures also low meaning NO mining.

Also, i see strange problem.
While i run phoenix from smartcoin i got 427+ Mhash/s even for my 6970 cards, but if i run the ./phoenix.py i get only 404 Mhash/s maximum for 6970 cards.
What extra adds in smartcoin?

[jondecker76]

This way doesn't seem to be possible, because you don't allow for profile edits yet. Once you specified a worker (1st one with high aggression) it's impossible to get back to specify another worker (low aggression one) and assign GPU's again.

Am I right or did I misunderstood something?

You're right.. The profile system does have the ability to do this, but a forgot that I had locked the ability to specify a different miner for each instance, mainly so people won't try to run different miners on the same card (this may not work, if for example you tried to run phoenix and poclbm on the same card). Perhaps I should bring back the ability to specify a specific miner per instance, just for this reason.. Then this type of failover would be possible.


What is everyone's feelings on this?  We can:
A)  Let you specify the miner to use for each instance.  This gives you more power and flexibility, but can cause problems if you mix different miners per card
B) Let you specify which miner to use for the entire profile. This is how its done currently.  The upside is that it prevents users from using different miners on the same card. The downside is that it limits the flexibility of the system a bit.

[jondecker76]

Here is all you need to do to uninstall and then reinstall: (uninstalling is much easier now that we're using sqlite)

Uninstall:

Code:

cd ~/
sudo rm -r smartcoin (sudo is needed so that it can remove the special .svn folders)
sudo rm /usr/bin/smartcoin (isn't needed if you're installing to the same location, but it doesn't hurt)

Install:

Code:

cd ~/
mkdir smartcoin
cd smartcoin
svn checkout http://smartcoin.googlecode.com/svn/branches/beta1/smartcoin ./
./smartcoin_install.sh

That should be all you need to do


I'll try to make an effort over the weekend to put this information all in one place

[jondecker76]

The first post of this thread, and the beta release post  have both been updated. They now contain current instructions for installing, uninstalling, updating etc. It also has some simple troubleshooting and instructions for making good bug reports.  I will try to keep these up to date in the future as we move forward.

If you look it over, please let me know if there is anything I missed.

[Alan Lupton]

What's the best way to detach from smartcoin? It seems I'm having issues using CTRL + A + D (one worker not working or all workers restarting). And then again, what's the best way to reconnect to the screen? I'm used to use screen -ls and type in the number.

[jondecker76]

Alan,
I just added a usage section to the instructions.

Do detach from smartcoin, either use the detach option from the control screen, or ctrl+a d. (I've never seen any miner instance shut down from disconnecting)

To reattach, you can just type smartcoin at the terminal just as you do while launching it.  It will automatically check if there is already a smartcoin instance running, and reattach if there is one instead of starting a new instance.  Another way is to simply type

Code:

screen -r smartcoin
or

Code:

screen -x smartcoin

(depending on whether or not the session is currently attached or not)

[dishwara]

Here is all you need to do to uninstall and then reinstall: (uninstalling is much easier now that we're using sqlite)

Uninstall:

Code:

cd ~/
sudo rm -r smartcoin (sudo is needed so that it can remove the special .svn folders)
sudo rm /usr/bin/smartcoin (isn't needed if you're installing to the same location, but it doesn't hurt)

Install:

Code:

cd ~/
mkdir smartcoin
cd smartcoin
svn checkout http://smartcoin.googlecode.com/svn/branches/beta1/smartcoin ./
./smartcoin_install.sh

That should be all you need to do
I'll try to make an effort over the weekend to put this information all in one place

Thanks.
It really helped me & after uninstall & re-install mining is working as before.
Thanks.

[Alan Lupton]

Again, I got the DOWN phenomenon. Did a reboot already, nothing helped, newest revision. What I changed is the phatk kernel accoring to this post: http://forum.bitcoin.org/index.php?topic=23067.0

Do you check the integrity of ther kernel or is it a coincidence?

[Clipse]

Again, I got the DOWN phenomenon. Did a reboot already, nothing helped, newest revision. What I changed is the phatk kernel accoring to this post: http://forum.bitcoin.org/index.php?topic=23067.0

Do you check the integrity of ther kernel or is it a coincidence?

Im guessing you either didnt copy/paste the code correctly into kernel.cl or you didnt restart the client after making the changes.

If it doesnt start workers after making those changes, something is wrong with the kernel.cl file.

[drgr33n]

Do you mind if I branch off and modify smartcoin to install and run from a location of my choice ? Also would like to add my details to the auto donate feature ?

[m3ta]

I can't believe install.sh just forced an 'updatedb' on my 5-drive, 4+ million files system.
The way I see it, that command is kind of.. 'intrusive' when being ran without any checks or confirmation from the user.

[m3ta]

Code:

enter the miner's path (i.e. /home/you/miner/)

"/home/youR/miner".

Code:

Enter the miner's launch string
Note:use special strings <#user#>, <#path#>, <#server#>

#pass#, not #path".

[Alan Lupton]

Again, I got the DOWN phenomenon. Did a reboot already, nothing helped, newest revision. What I changed is the phatk kernel accoring to this post: http://forum.bitcoin.org/index.php?topic=23067.0

Do you check the integrity of ther kernel or is it a coincidence?

Im guessing you either didnt copy/paste the code correctly into kernel.cl or you didnt restart the client after making the changes.

If it doesnt start workers after making those changes, something is wrong with the kernel.cl file.

Edited the kernel.cl on two other machines. Just copied it over WinSCP.

[plantucha]

I used separate workers because bitcoin.lc has issues with a single worked (username and password) handling all of my GPUs. This is actually the reason I decided to move from Diablo to smartcoin, I needed something to setup a separate worker per GPU and smartcoin is really slick.

Really? I just set up one worker for bitcoins.lc and it runs like a charm. Equally split with the first worker.

I haven't dug in real deep on it yet but with 8 GPUs it's a mess with about 15% rejects, with 4 GPUs I manage 10-12% rejects. One worker per and I'm down to 1.19% rejects, I haven't tried to narrow down exactly what the split is at this point.

PC:Ubuntu64narwhal, Smartcoin r297, ATI6790 x3, AMDX6 1100

bug report:

Smartcoin
with 3GPUs and 4 workers (each worker for different pool)   .........33% rejected
with 3GPUs and 3 workers (each worker for different pool)   .........13% rejected
with 3GPUs and 2 workers (each worker for different pool)   .........5% rejected

with 3GPUs and 2 workers (each worker for different pool) 
+ improved phatk OpenCL kernel 2% increase Hash/s
http://forum.bitcoin.org/index.php?topic=25135.0 and only   .........3% rejected

if One worker per GPU without smartcoin                     .........1%< rejected

it looks like more pools = more rejections

maybe will be good idea to try assign 1poolworker to 1GPU and if pool shut down
reassign whole GPU to another pool (or open some backup pool)
no spreading equally to other pools. if pool opens, redirect GPU back to original pool.

Request:
Can we assign also specific miner to device?
My CPU AMDX6 is mining 24MHash/s without OC.
And i have to start minerd outside of Smartcoin.
 

[plantucha]

http://forum.bitcoin.org/index.php?topic=25750.0
how about include this script 
than some fan control. 
  And than something what will balancing maximum clock speed based on max set temperature. 

Hi everyone,

I'm an experience software developer who set up a mining rig last month using Ubuntu. I found the overclocking tool available for Linux (AMDOvDrvCtrl) was a bit annoying to use, so I took a look at the source and decided to write my own command-line version. Rather than write one in C or C++, I figured it would be much more useful to wrap the underlying libraries using Python's ctypes package so I could write the tool in Python. This means no compiling, no installing an SDK, and you can read the code yourself to make sure there's no funny business going on.

Anyway, I'm done and the tool is available. You can install it in a few different ways:

From github: https://github.com/mjmvisser/adl3
From PyPi: http://pypi.python.org/pypi/adl3/0.2
With Python's "easy_install": sudo easy_install adl3

If you don't use easy_install, you'll need to run "python setup.py install" to install the package.

Included is the "atitweak" tool. It can list your adapters just like aticonfig:

Code:

$ atitweak -l
0. ATI Radeon HD 5800 Series  (:0.0)
    engine clock range is 80 - 918MHz
    memory clock range is 150 - 1200MHz
    core voltage range is 0.95 - 1.088VDC
    performance level 0: engine clock 800MHz, memory clock 300MHz, core voltage 0.95VDC
    performance level 1: engine clock 918MHz, memory clock 300MHz, core voltage 1.088VDC
    performance level 2: engine clock 918MHz, memory clock 300MHz, core voltage 1.088VDC
1. ATI Radeon HD 5800 Series  (:0.1)
    engine clock range is 80 - 918MHz
    memory clock range is 150 - 1200MHz
    core voltage range is 0.95 - 1.088VDC
    performance level 0: engine clock 800MHz, memory clock 300MHz, core voltage 0.95VDC
    performance level 1: engine clock 918MHz, memory clock 300MHz, core voltage 1.088VDC
    performance level 2: engine clock 918MHz, memory clock 300MHz, core voltage 1.088VDC

It can also change the engine/memory clock speed and core voltage individually for each performance level and adapter:

Code:

$ atitweak --set-memory-clock=300 --performance-level=2 --adapter=0
Setting performance level 2 on adapter 0: memory clock 300MHz

Here's the full command-line help:

Code:

Usage: atitweak [options]

Options:
  -h, --help            show this help message and exit
  -l, --list-adapters   Lists all detected and supported display adapters.
  -e ENGINE_CLOCK, --set-engine-clock=ENGINE_CLOCK
                        Sets engine clock speed (in MHz) for the selected
                        performance levels on the selected adapters.
  -m MEMORY_CLOCK, --set-memory-clock=MEMORY_CLOCK
                        Sets memory clock speed (in MHz) for the selected
                        peformance levels on the selected adapters.
  -v CORE_VOLTAGE, --set-core-voltage=CORE_VOLTAGE
                        Sets core voltage level (in VDC) for the selected
                        performance levels on the selected adapters.
  -A ADAPTERLIST, --adapter=ADAPTERLIST
                        Selects which adapters returned by --list-adapters
                        should be affected by other atitweak options.
                        ADAPTERLIST contains either a comma-seperated sequence
                        of the index numbers of the adapters to be affected or
                        else contains the keyword "all" to select all the
                        adapters. If --adapter is missing, all adapters will
                        be affected.
  -P PERFORMANCELEVELLIST, --performance-level=PERFORMANCELEVELLIST
                        Selects which performance levels returned by --list-
                        adapters should be affected by other atitweak options.
                        PERFORMANCELEVELLIST contains either a comma-separated
                        sequence of the index numbers of the performance
                        levels to be affected or else contains the keyword
                        "all" to select all performance levels. If
                        --performance-level is missing, all performance levels
                        will be affected.

So far, I've only tested this on my own rig (dual 5850s) so it would probably benefit from testing by experienced users.

This should be obvious, but...
WARNING: This software may severely damage or destroy your graphics card if used incorrectly!

Please try it out, and let me know if you find any problems or bugs.

cheers,
-Mark

[Alan Lupton]

It's me again trying to fight with my system. Did a full uninstall & reinstall, deleted all smartcoin* Files in /tmp (there were not files anyway) and no I'm getting thy "No screen sessions found" syndrome again.

[jondecker76]

Sorry all,  its been a busy weekend and I haven't been around much (family plans, 4th of July, wife's birthday...)
I'll try to catch up on all the questions:


drgr33n:
I don't care what you do with smartcoin locally, but I definitely would not appreciate it if you branched off to add support for installing it anywhere and changed the donation profiles. I literally have hundreds of hours of my time into this.  To be up-front with you (and everybody), smartcoin makes me next to nothing - I've received 4 BTC in donations, 0 BTC from the "pledges" that were made, and about 0.019 BTC (about 2 bit cents) per day via autodonate.  Be patient, support will come to move it to an arbitrary directory pretty soon actually, as things are stabilizing out quite a bit.

m3ta:
I'll modify the installer to make auto-detection optional.  The linux locate command simply is not reliable without running updatedb before hand.
Aside from that,  your install is quite unique in that most people mine on completely stripped down machines, dedicated solely to mining.  Most people wouldn't even dream of running 5 drives packed full of files on their miners, so understand that features that are made are assuming the case of the normal  average user.  But I'll definitely try to keep special cases like this in mind in the future, thanks for the report

"/home/youR/miner".

I can change that, though I commonly see it represented both ways

#pass#, not #path".

Actually, both are valid. The current list of tags includes: <#path#>,<#user#>,<#pass#>,<#server#>,<#port#>,<#device#>

plantucha:

it looks like more pools = more rejections

I don't have that problem here.  I run 4 different workers on each GPU, with 3 GPUs.  My rejections are almost always under 1%.  I have seen other installs over SSH which mimic this behavior, and  have no extra rejection problems.  Additionally,  I see no performance hit running multiple instances eaither. (perhaps it related to an outdated phoenix, what phoenix revision are you running?)

Can we assign also specific miner to device?

I will be adding this feature very very soon (tomorrow most likely), the only thing you will have to watch out for is running 2 different miners on one card (i.e. phoenix and poclbm on the same card), as they could be running different kernels.  However, running different instances of phoenix with  unique parameters is more the type of thing you would be wanting to do.

Alan:
RE: <<<DOWN>>>
I have seen this report a couple of times, and now have the problem figured out.  When things appear to lock up in the down state, its most likely that phoenix is reporting "Work queue empty. Miner Idle".  This is a long standing phoenix connection bug in older phoenix versions. (If it happens again,  look at the miner screen (screen -r miner). You will likely find one or more of the phoenix instances with this error.
I used to get this error all the time in phoenix.  I switched to phoenix r100 about 3 weeks ago, and haven't had the problem since.
Anyone having this problem, I highly recommend getting  updated versions of phoenix installed (I recommend r100, as I know it works well regarding their connection bug, but I'm sure even more recent updates are fine too)
If phoenix was installed via svn, just run:

Code:

cd /path/to/phoenix
svn update -r 100

If you wanted to start fresh with checking out phoenix r100, I would use something like:
svn checkout -r 100 http://svn3.xp-dev.com/svn/phoenix-miner/trunk

also, I have been running the same phatk kernel mod for a while now, with no problems at all.



Also, I ssh'd in and smartcoin is up and running, you accidentally misspelled <#device#> as <#devic#> in your miner launch string

[Alan Lupton]

Thanks man, I really appreciate your constant patience with the <---- n00b

I'll throw at least 4GHash at you for 30 minutes every day. The problem is that dd (live cd and running system) freezes with kernel error at some point and I can't clone the usb flash drives.

fdisk -l shows some like: "physical and logical beginnings differ (not linux?)"

Which could be the problem with flashdrive, because of the controller reassinging the cells all the time. Do you also use a usb flash drive? I didn't have the problem when I cloned my HDD from rig 1 ---> 2.