Also be aware that it may silently pretend to set the values but not actually do so. Although my 6970s report ram speeds possible of 320-1450, if I set them to anything below 825, it actually just resets them to normal values.
|
|
|
Can't compile this or the latest git on Linux without GPU mining support. I still get this error from make: make[2]: Entering directory `/home/XXX/cgminer-1.5.3' gcc -DHAVE_CONFIG_H -I. -pthread -fno-strict-aliasing -I./compat/jansson -I./lib -I./lib -DHAS_YASM -O3 -Wall -msse2 -march=native -MT cgminer-main.o -MD -MP -MF .deps/cgminer-main.Tpo -c -o cgminer-main.o `test -f 'main.c' || echo './'`main.c main.c: In function ‘reinit_device’: main.c:3420:12: error: ‘reinit_gpu’ undeclared (first use in this function) main.c:3420:12: note: each undeclared identifier is reported only once for each function it appears in
Can't have that now can we! * ckolivas quietly replaces the tarballs with a fixed version Ok, redownload
|
|
|
New release: 1.5.3 Source: http://ck.kolivas.org/apps/cgminer/cgminer-1.5.3.tar.bz2http://ck.kolivas.org/apps/cgminer/cgminer-1.5.3.tar.lrzLinux x86_64 dynamic binary: http://ck.kolivas.org/apps/cgminer/cgminer-1.5.3-x86_64-built.tar.bz2http://ck.kolivas.org/apps/cgminer/cgminer-1.5.3-x86_64-built.tar.lrzChangelog: - Significant work went into attempting to make the thread restart code robust to identify sick threads, tag them SICK after 1 minute, then DEAD after 5 minutes of inactivity and try to restart them. Instead of re-initialising the GPU completely, only a new cl context is created to avoid hanging the rest of the GPUs should the dead GPU be hung irrevocably. Much thanks to sunbreak for testing various modes of failure on his 66 GPU cluster. - Use correct application name in syslog. - Get rid of extra line feeds. - Use pkg-config to check for libcurl version - Implement per-thread getwork count with proper accounting to not over-account queued items when local work replaces it. - Create a command queue from the program created from source which allows us to flush the command queue in the hope it will not generate a zero sized binary any more. - Be more willing to get work from the backup pools if the work is simply being queued faster than it is being retrieved. So I'm leaving tomorrow for 10 days and I still don't have a windows binary post version 1.5.1 since I'm entirely dependent on Ycros for building one for me. If he shows up on these forums and posts a link to a windows build, consider it official. Otherwise, I'm afraid that's all I can do. P.S. If you're wondering what a .lrz file is, that's my extreme compression format which also has extreme encryption capabilities. http://lrzip.kolivas.org
|
|
|
You can't figure out how to run cgminer in the directory you compiled it in? Sucks to be you, then... LOL.... not that m8... in linux I would need to figure out, how to OC my cards! (2x HD 5750, OC'd from 700MHz to 870MHz in windows with MSI Afterburner) amdovdrvctrl works a treat for me and it's a gui app for those trying to escape windows http://sourceforge.net/projects/amdovdrvctrl/
|
|
|
Updated git tree getting ready for last release before I disappear for 10 days.
Significant work went into attempting to make the thread restart code robust to identify sick threads, tag them SICK after 1 minute, then DEAD after 5 minutes of inactivity and try to restart them. Instead of re-initialising the GPU completely, only a new cl context is created to avoid hanging the rest of the GPUs should the dead GPU be hung irrevocably. - Use correct application name in syslog. - Get rid of extra line feeds. - Use pkg-config to check for libcurl version - Implement per-thread getwork count with proper accounting to not over-account queued items when local work replaces it. - Create a command queue from the program created from source which allows us to flush the command queue in the hope it will not generate a zero sized binary any more.
|
|
|
[2011-07-29 22:19:38] Thread 1 idle for more than 60 seconds, GPU 0 declared SICK! [2011-07-29 22:19:38] Attempting to restart thread [2011-07-29 22:19:38] Thread 1 restarted
Restarting the thread didn't seem to work, because it kept SICK and gpu load was still 0%. Trying to "restart" the GPU via the curses interface resulted in a segfault. I seem to have problems running cgminer with gdb attached, even just with text interface (-T). (Still with 947a74bfa3fb1fe731b479b806a84ee6bc82ce0a, trying latest now.) Yeah you caught the code in a state of flux. I've just committed a change which tries to send the device a command if it finds it sick. If the command returns it means the GPU isn't hung (hopefully) and then it will try to restart it itself if it gets to 10 minutes (dead state).
|
|
|
I see some regression since 1.4.1. For 1.4.1 I ran cgminer for multiple days without any problems. With 1.5.2 after a few hours one gpu is always marked as DEAD, and there is no problem because I can just restart cgminer without any problems.
EDIT: I see your latest commit 947a74bfa3fb1fe731b479b806a84ee6bc82ce0a might be associated with this, I will try.
Not sure that one's correct yet sorry. Before did you ever get any thread restarts that you can remember?
|
|
|
Here's what my work looks like:
{ "id": 0, "result": { "target": "ffffffffffffffffffffffffffffffffffffffffffffffffffffffff00000000", "midstate": "aca85347af9a2c933be96bdd8a0e9a2b11da0977108243537ea4daa16fb4dd2a", "data": "000000012c0fad8f9157ef574727e1e86cb27760f3914d9b6635d0240000082a000000002ffcdde 6edf23bac3d969036e734e26572061b9b3bd4551b19ccd05a7f928ea54e32b2bc1a09ec04000000 0000 0000800000000000000000000000000000000000000000000000000000000000000000000000000 000000080020000", "hash1": "0000000000000000000000000000000000000000000000000000000000000000000000800000000 0000000000000000000000000000000000000000000010000" }, "error": null }
That's just what work looks like from the pool. I think you just haven't spotted it in 1.4.
If you're having a problem, it's something else.
|
|
|
I'm not sure what you're telling me. That error: null field is coming from your pool, not from cgminer.
|
|
|
The ncurses crash that you fixed in 1.5.0 is back again. This is a really annoying crash and there is no way I can update curses on Ubuntu 10.04 LTS, I was hoping to switch over from Diabolo today :-/ Give a try to libcurl4-gnutls-dev... sudo apt-get install libcurl4-gnutls-dev The bug is in ncurses, not curl. Easy mixup, and they're both libraries used by cgminer...
|
|
|
As for CPU mining in 1.5.x+, I find the rate of accepted shares is the same crap rate as previous versions, as far as CPU mining goes. It's no worse than it ever was, but the efficiency is higher (efficiency being accepted shares / requests, NOT the reject rate).
|
|
|
The code already checks the window size every 2.5 seconds on average and then resizes. Windows curses problem?
|
|
|
Also I noticed after couple hours is CPU mining hashing going slowly down For example on start after 2 minutes is ~21Mh/s after 8 hours ~13Mh/s
if I stop miner and start miner, it is again ~21Mh/s. With all versions versions Cgminer.
That bug was fixed in 1.5.2
|
|
|
Keep the delivery of the data consistent by being in big endian. It should be up to the coders to flip the endianness before feeding it to openCL.
|
|
|
So potentially the 11.7 drivers can fix the 100% CPU usage bug on windows. Well they've done a 180 degree turnaround and now the 11.7 driver uses 100% cpu on linux when mining where previously it was perfectly fine. Perhaps it's awaiting the next SDK version (2.5) to work properly, but with SDK 2.4 the CPU usage is sky high. I don't see anywhere obvious to download this alleged sdk2.5 by itself yet.
|
|
|
The ncurses crash that you fixed in 1.5.0 is back again. This is a really annoying crash and there is no way I can update curses on Ubuntu 10.04 LTS, I was hoping to switch over from Diabolo today :-/
I fixed as much as I could at my end. There will still be a curses bug that I have no control over and only goes away by upgrading curses. See: http://davidben.net/blog/2010/05/03/barnowl-ncurses-and-terminal-resizing/for the story of how the guy found the bug and fixed it. So the only solution for any remaining resizing bugs is to upgrade ncurses or not use the curses interface by using -T mode. On ubuntu you could add the repository for a newer release and just select ncurses from that, or you could install it yourself. Either way the bug is not going to go away entirely since it's in ncurses itself.
|
|
|
New release 1.5.2: Source: http://ck.kolivas.org/apps/cgminer/cgminer-1.5.2.tar.bz2Linux x86_64 binary: http://ck.kolivas.org/apps/cgminer/cgminer-1.5.2-x86_64-built.tar.bz2Windows binary delayed! Changes: - Restarting a hung GPU can hang the rest of the GPUs so just declare it dead and provide the information in the status. - The work length in the miner thread gets smaller but doesn't get bigger if it's under 1 second. This could end up leading to CPU under-utilisation and lower and lower hash rates. Fix it by increasing work length if it drops under 1 second. - Make the "quiet" mode still update the status and display errors, and add a new --real-quiet option which disables all output and can be set once while running. - Update utility and efficiency figures when displaying them. - Some Intel HD graphics support the opencl commands but return errors since they don't support opencl. Don't fail with them, just provide a warning and disable GPU mining. - Add http:// if it's not explicitly set for URL entries. - Log to the output file at any time with warnings and errors, instead of just when verbose mode is on. - Display the correct current hash as per blockexplorer, truncated to 16 characters, with just the time.
|
|
|
is there a solutiion for the problem other than downgrading the driver back to 11-6
I haven't been able to find one. It appears to really be a driver bug.
|
|
|
i have updated my ATI Driver on Ubuntu to the newest Version 11.7. Now I am encountering the Problem, that the miner generates 100% of CPU Usage and is significiantly slower than before the update. Anybody here with experience on that?
Yep, that's one dodgy driver. I've reproduced it.
|
|
|
Make sure you're redirecting STDERR, not stdout. You need to use 2>filename.txt Also you can enable/disable verbose mode on the fly once it's running.
Which is a little bit of a pain to do when used in combination with screen. Being able to specify an actual log file on the command line would be neat. I'm trying to tell you it will work... just redirect 2>filename.txt instead of >filename.txt eg: no log ./cgminer -o blah -u blah -p blah with log ./cgminer -o blah -u blah -p blah 2>filename.txt
|
|
|
|