Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 519. (Read 5805546 times)

I only noticed the difference with rollNtime. The result was cgMiner ran through the full roll of the work on the rollNtime work before trying a different work from a different pool.
As far as correcting the shares possibly switch every N minutes will work for you. I would set it to maybe every 1 minute then the pool gets work from both for 1 minute switces and repeats. There will likely be some variation because of luck and because of new blocks whatever pool is being used at the time of the most new blocks would likely be a little slow.

Have run 2.6.4 since its release only changing to 2.6.5 last night and I have to say with Win7 64bit and 1 BFL single I have had 0 issues. Failover setting not load balance. The software works great for me!

I have noticed a similar behavior too, but with me pool 3 gets almost all of the shares and pools 0, 1 and 2 get starved.
Sam

New version - 2.6.5, 15th August 2012

seems to work great on win7 x64 

thanks allot for all your great work 

Regards Testit

When using the load balance feature between 2 pools, I always have more accepted shares on pool 0 than pool 1. If I switch the pools on my .conf file I get the same result. Pool 0 always has a considerably higher amount of accepted shares than pool 1.

I have a cgminer instance running 2 gpu's of which I can see each gpu is working both pools and the gpu's are both identical as well as the same settings. They both run within 1MH/s of each other. One gpu runs about 4c higher than the other and the hotter one usually always has a slightly higher U value than th other.

On a separate box I run just a single gpu and I get the same result from it in that pool 0 always has a higher amount of shares. Currently I just run one .conf will pools listed one way and on the other .conf I am listing the pools opposite the first. Any idea how I may be able to get a better balance. Also at wat cost is my decision to run the load balance feature possibly negating the efficiency versus just using the failover feature? I should also say my decision to use the load balance is an attempt to lower variance I may see from a single pool.

Thanks for your time

EDIT: Is there a better alternative to the load balancing option to distributing work between 2 pools in attempt to avoid variance of a single pool?

May be a little quick to confirm wat I stated about
"On a separate box I run just a single gpu and I get the same result from it in that pool 0 always has a higher amount of shares."

Have different results on very small sample sizes once I reversed the pools on that miner. Going to check with the pool and see if they are running difficulty 1 shares or something different. I guess this could cause issue with the load balance feature.

Oh well ...

Doesn't help; I can't get it to mine because it's compiling without OpenCL support.

cgminer

You can only be getting that error if you're downloading something that needs a library that isn't on your machine when you build it, so I assume you're download my x86_64 built binary.

Build what part? The driver, cgminer or...?

If you're on linux, it really is very little effort to build it yourself. Don't be afraid. Long term you're much better off.

I don't know, I used to have the driver and SDK 2.5 installed separately, but I don't remember how I did it. That's how I had originally gotten my phatk kernel that stopped working after cgminer 2.6. Now I just have the 12.1 driver installed and whichever SDK that comes with (2.6 I assume). Any attempts to install the SDK separately are met with segfaults.

Honestly I dunno wtf is wrong with your setup. The opencl library is required and unless you fix the way it's installed on your machine, then you have to compile it yourself.

cut/paste

2.6.5
An Xubuntu 11.04 x86_64 executable is in my github downloads called cgminer-2.6.5a
https://github.com/kanoi/cgminer/downloads

For anyone who didn't realise, it's just the executable file to put in place of 'cgminer'
Nothing else needs changing
First get and extract the full binary release from ckolivas and then copy my file in place of 'cgminer'

No Problems so far on my GPU+2xIcarus or BFL (I'm now only running 2 instances of cgminer)

The same configure options as cvolivas' binary version
In case anyone was wondering:
CFLAGS="-O2 -W -Wall" ./configure --enable-icarus --enable-bitforce --enable-ztex --enable-modminer --enable-scrypt

I have also added a WinXP cgminer-2.6.5a.exe

This ONLY has BFL + ICA (as below) thus it doesn't need a computer with OpenCL on it
CFLAGS="-O2 -W -Wall" ./configure --enable-icarus --enable-bitforce

You will most likely also need the windowsdlls.zip file there in my downloads since some of my *.dll might be slightly different to ckolivas

maybe 5%. it's worth it. 5870 I could give you good settings

So far working great on Pi with multiple BFLs.

Awesome, thanks. That did it. I am now running v 2.6.5

ScarecrowMagick

I always do a "make clean", before "make"

Not quite, you need to execute "./autogen.sh" before the configure step. (Also, you probably don't need the sudo there.)

Still kind of new at this GIT stuff.

I stopped cgminer, (on linux) then did the following:

git pull
./configure --enable-cpuminer
make
sudo make install

when it was done, I started cgminer again, and it still said it was v 2.6.4

It's working, tho.  Did I do it correctly?

scarecrowmagick

Damn, still requires that random file. Back to 2.6.2.

New version - 2.6.5, 15th August 2012

Bugfixes yay

Human readable changelog
Found the longstanding problem with dynamic mode not really working on windows - the windows timer resolution is only 15ms and we're trying to sample much smaller than that. This was leading to the time taken to do GPU work appearing as zero for many samples. I used oversampling of results to find a mean time.
gpu-memdiff should now take effect when you change gpu engine clock from the menu as well
Hopefully the ADL gpu-map feature should work now when you have more adl devices than opencl (more ATI cards that dont support opencl).
More tweaks to the queueing mechanism to increase efficiency and keep minirigs fully work laden.
failover-only is available via the RPC API now.
More tweaks to bitforce devices to decrease comms errors.
Fixes to writing json config files.
Updated miner.php


Full changelog
- Don't try to get bitforce temperature if we're polling for a result to
minimise the chance of interleaved responses.
- Set memory clock based on memdiff if present from with engine changes,
allowing it to parallel manual changes from the menu as well.
- Increase the timeout on bitforce as per Paul Sheppard's suggestion to account
for throttling + work time + excess.
- Fix ADL gpu-map not working when there are more ADL devices than openCL.
Initial patch supplied by Nite69. Modified to suit.
- Windows' timer resolution is limited to 15ms accuracy. This was breaking
dynamic intensity since it tries to measure below this. Since we are repeatedly
sampling similar timeframes, we can average the gpu_us result over 5 different
values to get very fine precision.
- Fix harmless unused warnings in scrypt.h.
- API allow display/change failover-only setting
- Check we are not lagging as well as there is enough work in getwork.
- Minimise locking and unlocking when getting counts by reusing shared mutex
lock functions.
- Avoid getting more work if by the time the getwork thread is spawned we find
ourselves with enough work.
- The bitforce buffer is cleared and hw error count incremented on return from a
failed send_work already so no need to do it within the send_work function.
- miner.php allow a custom page section to select all fields with '*' - e.g. to
create a STATS section on a custom page
- Escape " and \ when writing json config file
- miner.php optional single rig totals (on by default)