CCminer(SP-MOD) Modded GPU kernels.

[bathrobehero]

Further, the Windows batch command "taskkill" will not find the imagename "ccminer*" if a user closes the "ccminer.exe" with a "control-c" keystroke, because ccminer has abruptly been closed.  This abrupt closing triggers a Windows error response frequently, and disrupts the loop on the ccminer launch batch file.  If running ccminer in a round-robin batch file, the Windows error response prevents ccminer from reconnecting and mining until a manual intervention occurs.

Closing out of ccminer.exe in a clean fashion under Windows may require some changes to the source code of ccminer itself.

--scryptr

I have the windows error reporting completely disabled on my rigs so crashing apps just silently close without throwing error messages. But I've just tested it on my main rig and it crashes due to this line:

https://github.com/sp-hash/ccminer/blob/windows/ccminer.cpp#L2383

That proper_exit(0); is not so proper I guess. Replacing it simply to exit(0); prevents the crash when you use ctrl+c. It asks if you want to termiate the batch file and pressing N let it continue.

And if taskkill doesn't find it's target it doesn't do anything.

ROUND-ROBIN--

Yaamp adapted the round-robin switch to rotate between the highest-paying algo, and it works until Windows kicks out that error.  I was experimenting with a simple batch file that calls a JSON.conf file.  I'll work your code into my Yaamp round-robin file and see if it will run well.  I suppose that means I will have to compile under Windows.  If I get that far, maybe it will rotate for me.

I did try to disable all the Windows error-reporting.        --scryptr

The error I described should only happen if you use ctrl+c. I don't use yaamp but as I see it they only let you connect to their highest paying algo and disconnect you if you try to connect with any other algo. With -r 0 ccminer won't reconnect so it just closes cleanly then the next line of the batch file gets executed. I've tried the yaamp example loop and it's working flawlessly. So now I'm completely confused as to what is the problem with the loop. 

[scryptr]

Further, the Windows batch command "taskkill" will not find the imagename "ccminer*" if a user closes the "ccminer.exe" with a "control-c" keystroke, because ccminer has abruptly been closed.  This abrupt closing triggers a Windows error response frequently, and disrupts the loop on the ccminer launch batch file.  If running ccminer in a round-robin batch file, the Windows error response prevents ccminer from reconnecting and mining until a manual intervention occurs.

Closing out of ccminer.exe in a clean fashion under Windows may require some changes to the source code of ccminer itself.

--scryptr

I have the windows error reporting completely disabled on my rigs so crashing apps just silently close without throwing error messages. But I've just tested it on my main rig and it crashes due to this line:

https://github.com/sp-hash/ccminer/blob/windows/ccminer.cpp#L2383

That proper_exit(0); is not so proper I guess. Replacing it simply to exit(0); prevents the crash when you use ctrl+c. It asks if you want to termiate the batch file and pressing N let it continue.

And if taskkill doesn't find it's target it doesn't do anything.

ROUND-ROBIN--

Yaamp adapted the round-robin switch to rotate between the highest-paying algo, and it works until Windows kicks out that error.  I was experimenting with a simple batch file that calls a JSON.conf file.  I'll work your code into my Yaamp round-robin file and see if it will run well.  I suppose that means I will have to compile under Windows.  If I get that far, maybe it will rotate for me.

I did try to disable all the Windows error-reporting.        --scryptr

The error I described should only happen if you use ctrl+c. I don't use yaamp but as I see it they only let you connect to their highest paying algo and disconnect you if you try to connect with any other algo. With -r 0 ccminer won't reconnect so it just closes cleanly then the next line of the batch file gets executed. I've tried the yaamp example loop and it's working flawlessly. So now I'm completely confused as to what is the problem with the loop. 

TASKKILL--

I inserted your taskkill lines of code between every launch line in the Windows batch file for Yaamp.  I still get Windows error reports, not every time, but enough to render the batch file useless for a long-running script.  Maybe I haven't completely shutdown the error reports in my system.   I haven't compiled ccminer with your suggested correction.  I need to set up a Windows build toolchain...

--scryptr

[bathrobehero]

TASKKILL--

I inserted your taskkill lines of code between every launch line in the Windows batch file for Yaamp.  I still get Windows error reports, not every time, but enough to render the batch file useless for a long-running script.  Maybe I haven't completely shutdown the error reports in my system.   I haven't compiled ccminer with your suggested correction.  I need to set up a Windows build toolchain...

--scryptr

So it sometimes crashes for you. In that case I doubt the code change will help you. The batch file doesn't call ctrl+c on its own.

[scryptr]

TASKKILL--

I inserted your taskkill lines of code between every launch line in the Windows batch file for Yaamp.  I still get Windows error reports, not every time, but enough to render the batch file useless for a long-running script.  Maybe I haven't completely shutdown the error reports in my system.   I haven't compiled ccminer with your suggested correction.  I need to set up a Windows build toolchain...

--scryptr

So it sometimes crashes for you. In that case I doubt the code change will help you. The batch file doesn't call ctrl+c on its own.

ERROR REPORTING--

I redid my Error Reporting configuration in Win 7 x64.  I selected "ccminer.exe" as a program to specifically be ignored.  I saved this as an administrator.  I still get "ccminer.exe has stopped working" between every few algo changes in the Yaamp batch file.     If there is another way to shut the error reporting off, I haven't found it yet.         --scryptr

[mendoza1468]

I have a question, if sometimes on yaamp the connection is lost and isn't restarting, i would like to add another pool. How can i do this. (with last ccminer1.5.47)
Am i the only one who lost connection on yaamp?
For now i use :
ccminer.exe -R 5 -T 5 -q -a quark -o stratum+tcp://yaamp.com:4033 -u 5LA1mw7JZKKVf3LK9KLAJSzQZzNwU9Z3Rp -p d=0.002

Can someone explain me what i should add to insert another pool after this one? (round robin, failover or someting else)

P.S sometime if i closed ccminer and restart it im able to log again and mine ...

Thanks for the help

[flipclip]

If flipclip would share his script, perhaps it would be more simple to adapt.

--scryptr

Unfortunately I don't believe my spaghetti code will be simple for anyone to adapt (I'm not a programmer, and I took the idea on as a lark to try out bash).  Let me see if I can clean it up a little so someone else could get it running.

[chrysophylax]

If flipclip would share his script, perhaps it would be more simple to adapt.

--scryptr

Unfortunately I don't believe my spaghetti code will be simple for anyone to adapt (I'm not a programmer, and I took the idea on as a lark to try out bash).  Let me see if I can clean it up a little so someone else could get it running.

that would be awesome ...

as im less of a programmer than most ...

but i know enough to get me into LOADS of trouble ...

...

#crysx

[chrysophylax]

I have a question, if sometimes on yaamp the connection is lost and isn't restarting, i would like to add another pool how can i do it. (with last ccminer1.5.47)
Am i the only one who lost connection on yaamp?
for now i use :
ccminer.exe -R 5 -T 5 -q -a quark -o stratum+tcp://yaamp.com:4033 -u 5LA1mw7JZKKVf3LK9KLAJSzQZzNwU9Z3Rp -p d=0.002
Can someone explain me what i should add to insert another pool after this one? (round robin, failover or someting else)

P.S sometime if i closed ccminer and restart it im able to log again and mine ...

Thanks for the help

ccminer itself doesnt have this facility mate ... unlike sgminer which has that feature built in ... but we are talking two different miners here ...

in the current discussion about scripts and such with flipclip and scryptr - that is pretty much what is on par with what you are asking ...

a way of making ccminer continue mining - albiet different algos and such - but to continue mining in the event of a crash or a timed close then reopen of ccminer ...

there is no real easy way at the moment but it will eventually happen i believe ...

#crysx

[scryptr]

I have a question, if sometimes on yaamp the connection is lost and isn't restarting, i would like to add another pool. How can i do this. (with last ccminer1.5.47)
Am i the only one who lost connection on yaamp?
For now i use :
ccminer.exe -R 5 -T 5 -q -a quark -o stratum+tcp://yaamp.com:4033 -u 5LA1mw7JZKKVf3LK9KLAJSzQZzNwU9Z3Rp -p d=0.002

Can someone explain me what i should add to insert another pool after this one? (round robin, failover or someting else)

P.S sometime if i closed ccminer and restart it im able to log again and mine ...

Thanks for the help

LOOPER--

This is how:

:loop

ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-west01.miningrigrentals.com:3333 -u me -p x  -d 0,1,2

taskkill -t -f /im ccminer*
timeout -t 1

ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-central01.miningrigrentals.com:3333 -u me  -p x  -d 0,1,2

taskkill -t -f /im ccminer*
timeout -t 1

goto loop

This is a slight revision of Bathrobehero's script.  Use the "-R" switch, and the "-r" switch.  The script above puts in a 5 second delay (-R 5) before switching to the lower pool in the batch file, and to try to reconnect once (-r 1).  Yaamp uses it to switch between algos.  As used above, it switches between pools.  Yamp.com has an example script linked on the homepage for algo switching.

If you execute the command "ccminer --help >> help.txt" in a windows command box, you will get a text file explaining all the ccminer commands.

--scryptr

[sp_]

COMMIT 749--
I just cloned git and compiled build 749.  The 2x970 FTW+ rig  is now mining Quark at 30.6Mh/s, an improvement of 1.5Mh/s for the rig.  Each card is now hashing at more than 15Mh/s.  

...

my 5x 750ti + 1x 280x rig went from 41.5 MHASH to 44.5MHASH (standard clocks, no overclock)
520 watt in the wall. Msi gaming z97 motherboard/ celeron cpu/

Release 47 used 510 watt,

Commit 749 is a good one, the gain is even bether on the 980, but it uses more power.
My 5 750ti+980 reports +1MHASH (45.5MHASH) and use 5 watt more  510 -> 515

[chrysophylax]

I have a question, if sometimes on yaamp the connection is lost and isn't restarting, i would like to add another pool. How can i do this. (with last ccminer1.5.47)
Am i the only one who lost connection on yaamp?
For now i use :
ccminer.exe -R 5 -T 5 -q -a quark -o stratum+tcp://yaamp.com:4033 -u 5LA1mw7JZKKVf3LK9KLAJSzQZzNwU9Z3Rp -p d=0.002

Can someone explain me what i should add to insert another pool after this one? (round robin, failover or someting else)

P.S sometime if i closed ccminer and restart it im able to log again and mine ...

Thanks for the help

LOOPER--

This is how:

:loop

ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-west01.miningrigrentals.com:3333 -u me -p x  -d 0,1,2

taskkill -t -f /im ccminer*
timeout -t 1

ccminer.exe -r 1 -R 5 -a quark   -o stratum+tcp://us-central01.miningrigrentals.com:3333 -u me  -p x  -d 0,1,2

taskkill -t -f /im ccminer*
timeout -t 1

goto loop

This is a slight revision of Bathrobehero's script.  Use the "-R" switch, and the "-r" switch.  The script above puts in a 5 second delay (-R 5) before switching to the lower pool in the batch file, and to try to reconnect once (-r 1).  Yaamp uses it to switch between algos.  As used above, it switches between pools.  Yamp.com has an example script linked on the homepage for algo switching.

If you execute the command "ccminer --help >> help.txt" in a windows command box, you will get a text file explaining all the ccminer commands.

--scryptr

nice - windows batch in dos ... been way too long for me ...

#crysx

[Star65]

Nice hashrate on quark & skein btw! Nice work sp_! My hashrate on quark: 2x750ti=3x280x!!! 100W=450W?!  
Cпacибo! I will mine for you! I have 2x750ti atm, but there is a slot for another one (in the daughter comp's).  

[BitcoinRichman]

w  W
how many $$ must i burn to get that speed

[bathrobehero]

TASKKILL--

I inserted your taskkill lines of code between every launch line in the Windows batch file for Yaamp.  I still get Windows error reports, not every time, but enough to render the batch file useless for a long-running script.  Maybe I haven't completely shutdown the error reports in my system.   I haven't compiled ccminer with your suggested correction.  I need to set up a Windows build toolchain...

--scryptr

So it sometimes crashes for you. In that case I doubt the code change will help you. The batch file doesn't call ctrl+c on its own.

ERROR REPORTING--

I redid my Error Reporting configuration in Win 7 x64.  I selected "ccminer.exe" as a program to specifically be ignored.  I saved this as an administrator.  I still get "ccminer.exe has stopped working" between every few algo changes in the Yaamp batch file.     If there is another way to shut the error reporting off, I haven't found it yet.         --scryptr

If I remember correctly I used a reg key like this (Option two):
http://www.sevenforums.com/tutorials/194181-windows-error-reporting-disable-windows.html

[chrysophylax]

Nice hashrate on quark & skein btw! Nice work sp_! My hashrate on quark: 2x750ti=3x280x!!! 100W=450W?!  
Cпacибo! I will mine for you! I have 2x750ti atm, but there is slot for another one (in the daughter comp's).  

we have a proposal ...

we have setup a donation system FOR sp which we are currently testing and mining via yaamp as i write this ...

this has NOT been discussed with sp - nor authorized by sp ... in fact - he knows nothing about it as yet - as we have just finished the setup a few minutes ago ...

currently the test is specifically on x11 - but other algos will follow to sp's btc address ... this can be seen via yaamp currently which will show that it has just started a few minutes ago ...

http://yaamp.com/?address=1CTiNJyoUmbdMRACtteRWXhGqtSETYd6Vd ...

it is the test system we have running with 2 x gigabyte 750ti oc lp cards running fedora 19 x64 and sp's latest ccminer v47 ( with intensity -i 20.6 as its the highest these cards will do ) ...

as this is the usual way we donate - we thought possibly we might not be the only ones that would want to do it this way ...

if you are interested in donating via mining and it is agreeable that a mining system for donation for sp be setup - we shall supply the details of the server and setup of the mining system and you too can join in ...

also - if this is a 'good' way of donating - and everyone can pitch in a small amount of hashpower ( no matter how small ) we will keep this system running for donations AND setup various other ones in different algorithms ... all of which can be publicly checked and seen on yaamp via sp's btc address ...

this is obviously based on a trust system - as donations are the lifeblood of the development process ...

please let us know your thoughts ...

#crysx

[Epsylon3]

small contribution--

I sent you, KlausT, and tpruvot each a bit.  I have to mow the lawn, I will get back to "git and compile" this evening.  I am still running the release build of v46 on all my rigs.  The release behaved well, the charts on the pool were trending up.

Please don't lower my rates...       --scryptr

Thanks for the 10 mBTC (first gift from the forum since... a long time)

For Cuda 7.0, its not compatible yet, but i submitted a bug report to nvidia which fixed the problem in their dev version. I didnt get details about the fix... but should be in next CUDA release, maybe 7.1, 7.5, 8 (no idea)

[scryptr]

small contribution--

I sent you, KlausT, and tpruvot each a bit.  I have to mow the lawn, I will get back to "git and compile" this evening.  I am still running the release build of v46 on all my rigs.  The release behaved well, the charts on the pool were trending up.

Please don't lower my rates...       --scryptr

Thanks for the 10 mBTC (first gift from the forum since... a long time)

For Cuda 7.0, its not compatible yet, but i submitted a bug report to nvidia which fixed the problem in their dev version. I didnt get details about the fix... but should be in next CUDA release, maybe 7.1, 7.5, 8 (no idea)

CUDA 7.x--

I downloaded and installed CUDA 6.5 on one of my rigs running nVidia driver v349.16.  The Cuda toolkit install reverted some of the driver components to v343.  The system compiles and runs, but my driver is reported as v343 now.       --scryptr

[DougB62]

small contribution--

I sent you, KlausT, and tpruvot each a bit.  I have to mow the lawn, I will get back to "git and compile" this evening.  I am still running the release build of v46 on all my rigs.  The release behaved well, the charts on the pool were trending up.

Please don't lower my rates...       --scryptr

Thanks for the 10 mBTC (first gift from the forum since... a long time)

For Cuda 7.0, its not compatible yet, but i submitted a bug report to nvidia which fixed the problem in their dev version. I didnt get details about the fix... but should be in next CUDA release, maybe 7.1, 7.5, 8 (no idea)

CUDA 7.x--

I downloaded and installed CUDA 6.5 on one of my rigs running nVidia driver v349.16.  The Cuda toolkit install reverted some of the driver components to v343.  The system compiles and runs, but my driver is reported as v343 now.       --scryptr

Yeah - Unfortunately, it doesn't allow you to keep your current driver set. 

[flipclip]

small contribution--

I sent you, KlausT, and tpruvot each a bit.  I have to mow the lawn, I will get back to "git and compile" this evening.  I am still running the release build of v46 on all my rigs.  The release behaved well, the charts on the pool were trending up.

Please don't lower my rates...       --scryptr

Thanks for the 10 mBTC (first gift from the forum since... a long time)

For Cuda 7.0, its not compatible yet, but i submitted a bug report to nvidia which fixed the problem in their dev version. I didnt get details about the fix... but should be in next CUDA release, maybe 7.1, 7.5, 8 (no idea)

CUDA 7.x--

I downloaded and installed CUDA 6.5 on one of my rigs running nVidia driver v349.16.  The Cuda toolkit install reverted some of the driver components to v343.  The system compiles and runs, but my driver is reported as v343 now.       --scryptr

Scryptr I can't remember if you are on linux but:

if you use the "run" installer (from the download page: https://developer.nvidia.com/cuda-toolkit-65) and not the rpm/deb, you are given the option not to install the driver associated with that version of the toolkit, so you can keep your more recent driver. (at least in Ubuntu, though I assume other distros are equivalent.)

[5w00p]

small contribution--

I sent you, KlausT, and tpruvot each a bit.  I have to mow the lawn, I will get back to "git and compile" this evening.  I am still running the release build of v46 on all my rigs.  The release behaved well, the charts on the pool were trending up.

Please don't lower my rates...       --scryptr

Thanks for the 10 mBTC (first gift from the forum since... a long time)

For Cuda 7.0, its not compatible yet, but i submitted a bug report to nvidia which fixed the problem in their dev version. I didnt get details about the fix... but should be in next CUDA release, maybe 7.1, 7.5, 8 (no idea)

CUDA 7.x--

I downloaded and installed CUDA 6.5 on one of my rigs running nVidia driver v349.16.  The Cuda toolkit install reverted some of the driver components to v343.  The system compiles and runs, but my driver is reported as v343 now.       --scryptr

Yeah - Unfortunately, it doesn't allow you to keep your current driver set. 

Actually, the Windows CUDA 6.5 Toolkit does allow you to keep your current driver set, if you just uncheck the options after having chosen a Custom Install.





It worked for me.