100% Stale Shares

[fortif78]

I'm running cgminer 2.11.3 on linux and every time I start up and connect to a pool, I get 100% rejected shares. Can anyone give me a little advice on how to fix this?  thanks

[MaGNeT]

I'm running cgminer 2.11.3 on linux and every time I start up and connect to a pool, I get 100% rejected shares. Can anyone give me a little advice on how to fix this?   thanks

What settings do you use? What pool? What GPU?

Do you think we are telepathic?

[fortif78]

I'm running cgminer 2.11.3 on linux and every time I start up and connect to a pool, I get 100% rejected shares. Can anyone give me a little advice on how to fix this?   thanks

What settings do you use? What pool? What GPU?

Do you think we are telepathic?

Sorry wasn't sure if it was just a generic issue that had been caused by missing something simple. Anyway, I've got a 5850 and 5870 running on this rig atm. Both are running at stock settings; no overclocking. I've tried switching the intesnitys on the cards to see if I'd get any response, but still nothing

[ssateneth]

sounds to me like you're mining ltc and you forgot to tell it to switch to scrypt mode.

[MaGNeT]

sounds to me like you're mining ltc and you forgot to tell it to switch to scrypt mode.

He still doesn't give much info, so keeps us guessing...

[fortif78]

I'm currently using 2 5850s on this rig, running cgminer with the scrypt option through terminal with the command:

Code:

./cgminer --scrypt -o <pooladdress:port> -u <username> -p <password>

returns "Too many values for temp cutoff".

Doing some research I've noticed it could possibly be the cgminer.conf file. I tinkered with that a little based off of the post here:
http://forum.litecoin.net/index.php?topic=639.0

my file looks something like (with the url, user, and pass filled in accordingly):

Code:

{
"pools" : [
	{
		"url" : "http://www.ulr.com:port",
		"user" : "user",
		"pass" : "pass"
	}
],

"intensity" : "10, 10",
"gpu-engine" : "850, 850",
"gpu-thread" : "2, 2",
"scrypt" : true,
"api-listen" : true,
"shaders" : ", 1440, 1440",
"thread-concurrency" : "1600, 1600",
"lookup-gap" : "2, 2"
}

I proceeded to run cgminer using the code:

Code:

./cgminer --config <filepathofconfigfile>

Running it that way I get an extremely low hashrate, but will get accepted shares. An additional problem revealed itself as the computer locks as soon as cgminer accepts its first share, but I think that correcting the overall issue would correct that as while.

--fort

[Rawted]

Can you adjust your thread concurrency to 8000 and report back?

[fortif78]

Ok, so I figured out for some reason cgminer just wasn't reading the scrypt command. Did a fresh install and now I am accepting shares.
The problem that I'm encountering now is that I can't get my cards to do over 150 kh/s. I'm running Xunbuntu 12.04,  11.11 drivers, and cgminer 2.11.3. Both cards in this rig are currently 5870's. If I push either of them over an intensity of 11 I get dumped with an extremely large amount of hardware errors.
Adjusting the thread concurrency, memclock, and engine are either making no difference, or adding to the amount of hardware errors. So essentially I'm stuck with

Code:

./cgminer --scrypt -o http://pooladdress:port -u username -p password -I 11 --gpu-fan 50

This will get me ~150 kh/s and a noteable portion of rejected shares. I've tried adjusting about everything I can in different ways with no luck at all. I've even tried different combinations of drivers + cgminer. Maybe I just haven't found the right combo?
Has anyone run into an issue like this?


--fort

[crazyearner]

Ok, so I figured out for some reason cgminer just wasn't reading the scrypt command. Did a fresh install and now I am accepting shares.
The problem that I'm encountering now is that I can't get my cards to do over 150 kh/s. I'm running Xunbuntu 12.04,  11.11 drivers, and cgminer 2.11.3. Both cards in this rig are currently 5870's. If I push either of them over an intensity of 11 I get dumped with an extremely large amount of hardware errors.
Adjusting the thread concurrency, memclock, and engine are either making no difference, or adding to the amount of hardware errors. So essentially I'm stuck with

Code:

./cgminer --scrypt -o http://pooladdress:port -u username -p password -I 11 --gpu-fan 50

This will get me ~150 kh/s and a noteable portion of rejected shares. I've tried adjusting about everything I can in different ways with no luck at all. I've even tried different combinations of drivers + cgminer. Maybe I just haven't found the right combo?
Has anyone run into an issue like this?


--fort

5870s can easily generate 400kh/s And I currently have one 5870 doing that.

Start with drivers keep any back up of the old version you have, upgrade to latest ones and go from their.... You could manually adjust the clocks using overclocking software and only set threads and shades, then use software to adjust clock speeds then start up configuration for that gpu


What temps are the gpu hitting what hardware errors are you getting what is your cgminer config. Try this in your config file as this is taken from my cgminer...

cgminer --scrypt -w 256 -g 1 -I 15 --thread-concurrency 8000, --shaders 1600

Also start --thread-concurrency at 1600 to begin with and then if fine at --thread-concurrency upgrade to 3200 if fine then 3 x 4x max would be 8000 for the card if can get it that high..

where I 15 is start at lower to begin with start at 9 with all clocks at default Core Clock 850MHz and Memory clock 1200 providing you have this version

Once it is running fine on i9 leave running for 10 to 15 min and see if any HW errors or anything wrong going on

[fortif78]

U

Ok, so I figured out for some reason cgminer just wasn't reading the scrypt command. Did a fresh install and now I am accepting shares.
The problem that I'm encountering now is that I can't get my cards to do over 150 kh/s. I'm running Xunbuntu 12.04,  11.11 drivers, and cgminer 2.11.3. Both cards in this rig are currently 5870's. If I push either of them over an intensity of 11 I get dumped with an extremely large amount of hardware errors.
Adjusting the thread concurrency, memclock, and engine are either making no difference, or adding to the amount of hardware errors. So essentially I'm stuck with

Code:

./cgminer --scrypt -o http://pooladdress:port -u username -p password -I 11 --gpu-fan 50

This will get me ~150 kh/s and a noteable portion of rejected shares. I've tried adjusting about everything I can in different ways with no luck at all. I've even tried different combinations of drivers + cgminer. Maybe I just haven't found the right combo?
Has anyone run into an issue like this?


--fort

5870s can easily generate 400kh/s And I currently have one 5870 doing that.

Start with drivers keep any back up of the old version you have, upgrade to latest ones and go from their.... You could manually adjust the clocks using overclocking software and only set threads and shades, then use software to adjust clock speeds then start up configuration for that gpu


What temps are the gpu hitting what hardware errors are you getting what is your cgminer config. Try this in your config file as this is taken from my cgminer...

cgminer --scrypt -w 256 -g 1 -I 15 --thread-concurrency 8000, --shaders 1600

Also start --thread-concurrency at 1600 to begin with and then if fine at --thread-concurrency upgrade to 3200 if fine then 3 x 4x max would be 8000 for the card if can get it that high..

where I 15 is start at lower to begin with start at 9 with all clocks at default Core Clock 850MHz and Memory clock 1200 providing you have this version

Once it is running fine on i9 leave running for 10 to 15 min and see if any HW errors or anything wrong going on

I tried running it with your config and found that if I mess with the thread concurrency at all, cgminer fails to start. Intensity over 10 brings my hashrate up immensely but drives my HW count through the roof. For example, after running at -I 15 for 5 minutes, my hashrate is ~301 kh/s, but I've only got 2 accepted shares between two cards, and a HW count of 23 and 26 for each card respectively. 

[-ck]

 Update your SDK to 2.6 or 2.7 and delete any .bin files.

[fortif78]

Update your SDK to 2.6 or 2.7 and delete any .bin files.

Updated to 2.6. Upon running

Code:

./Install-AMD-APP.sh

everything looks to go smoothly until the end when i receive:

Code:

/sbin/ldconfig.real: Path `/lib/x86_64-linux-gnu' given more than once
/sbin/ldconfig.real: Path `/usr/lib/x86_64-linux-gnu' given more than once
/sbin/ldconfig.real: Cannot stat /usr/lib/x86_64-linux-gnu/libnss_db.so: No such file or directory
/sbin/ldconfig.real: Cannot stat /usr/lib/libfglrx_dm.so: No such file or directory

Nevertheless, at the end it reads AMD APP installation completed. So I am not sure if those files had any bearing on the ability for the SDK to install and function properly. I then continued to run cgminer as follows:

Code:

./cgminer --scrypt -o http://server:port -u username -p password -I 10 --gpu-fan 50

(server,port,username, and password all specified when I ran it obviously)
Cgminer starts and opens, but zero improvement. HW actually turned to 1 on my second card pretty much as soon as cgminer started running. I tried to increase the intensity just to see what happened; same thing as before. HW count just rises and rises. I'm so lost on what the issue is still

--fort

[gnarcoinguy]

You ever figure this out?  I am having a very similar problem.  100% hardware errors with cgminer on linux.  On the same machine in windows though it works fine... I've tried 6 or 7 different ati drivers now... I'm running out of ideas.

[dE_logics]

You ever figure this out?  I am having a very similar problem.  100% hardware errors with cgminer on linux.  On the same machine in windows though it works fine... I've tried 6 or 7 different ati drivers now... I'm running out of ideas.

+1