Did the graphics drivers update? There have been lots of reports of the recent drivers dropping hash rates. A reboot would cause newly installed drivers to load.
Nope, I'm already at 11.12 (didn't see a performance drop that everyone else reports - interesting). On that note, what are the suggested drivers for mining in Windows? 11.6? 11.6 with sdk 2.4 remains the best for both windows and linux for mining, if not for gaming.
|
|
|
Could you try the git tree now please?
Tried twice and got no segfault - seems to work. Sweet, thanks! I don't know what to make of the other reports of connectivity issues since none of that code was actually changed... Unless something changed indirectly by addition of the API code?
|
|
|
Still missing some of the bounty
|
|
|
Could you try the git tree now please?
|
|
|
Again, died in the ATI driver. Well without hardware to reproduce it here, I'm kinda stuck for now. Unless someone has the knowhow to do a git bisect and find when the problem started happening, then I'm at a loss. My GPU based mining hardware may not be online for weeks again > git bisect bad 59293a37d6abaa879e6f24e26b206a3b318b2f79 is the first bad commit commit 59293a37d6abaa879e6f24e26b206a3b318b2f79 Author: Con Kolivas <kernel@kolivas.org> Date: Mon Dec 26 09:29:16 2011 +1100
Only pthread_join when pthread_cancel does not return an error.
:100644 100644 0ad8b7d89902aed19e0dd2c96333ecb98c0d7e1d 0bf71dc28e700a6481b3250e1ba156a58969a28c M util.c 000;[23:52:41] reini500gb cgminer_build (7.1 Mb)> git show 59293a37d6abaa879e6f24e26b206a3b318b2f79 commit 59293a37d6abaa879e6f24e26b206a3b318b2f79 Author: Con Kolivas <kernel@kolivas.org> Date: Mon Dec 26 09:29:16 2011 +1100
Only pthread_join when pthread_cancel does not return an error.
diff --git a/util.c b/util.c index 0ad8b7d..0bf71dc 100644 --- a/util.c +++ b/util.c @@ -681,7 +681,7 @@ void thr_info_cancel(struct thr_info *thr) if (thr->q) tq_freeze(thr->q); if (thr->pth) { - if (pthread_cancel(thr->pth)) + if (!pthread_cancel(thr->pth)) pthread_join(thr->pth, NULL); thr->pth = 0L; } Fantastic, thanks very much. All I need to do is not pthread_join then.
|
|
|
Again, died in the ATI driver. Well without hardware to reproduce it here, I'm kinda stuck for now. Unless someone has the knowhow to do a git bisect and find when the problem started happening, then I'm at a loss. My GPU based mining hardware may not be online for weeks again
|
|
|
Check the directory, it will be called core.something where something is a number. It will not be called just core.
Hmm? The complaint was: "/home/brec/miners/cgminer/core": not in executable format: File format not recognized rather than not found. And: brec@ubuntu:~/miners/cgminer$ ll core -rw------- 1 brec brec 270467072 2011-12-29 08:06 core
Oh sorry. Try: gdb cgminer core
|
|
|
Can you, in whatever command line window, set "ulimit -c unlimited", then execute a new instance of cgminer and quit that at once (it should print something like "segmentation violation: core dumped"). You can do that without stopping any other running cgminer instances you may have. There should then be a file called "core" in whatever directory you're in. After this, execute "gdb path/to/cgminer core" and in gdb first execute "bt" and then "bt full", and post the results here?
My suspicion is/was that cgminer now triggers some bug in the AMD opencl driver which it didn't before, but if it is only a catalyst problem, it should occur for ckolivas, too (assuming he still uses 11.6).
This is not what you wanted; gdb complains about the core file; something's missing: brec@ubuntu:~/miners/cgminer$ ulimit -c unlimited brec@ubuntu:~/miners/cgminer$ ./start.sh [2011-12-29 08:05:59] Started cgminer 2.1.0 [2011-12-29 08:06:01] Long-polling activated for http://us.eclipsemc.com:9009/LP [2011-12-29 08:06:05] Accepted 00000000.2d99f6fd.75adf0e9 GPU 2 thread 2 pool 2 GPU2 39.5C 4499RPM | (5s):331.7 (avg):7717.5 Mh/s | A:1 R:0 HW:0 U:600.00/m I:9 [2011-12-29 08:06:05] Accepted 00000000.dc0e6fa8.92e05d8a GPU 0 thread 0 pool 0 GPU0 39.5C 4530RPM | (5s):333.8 (avg):11744.1 Mh/s | A:1 R:0 HW:0 U:600.00/m I:9 [2011-12-29 08:06:06] Accepted 00000000.0571fbb8.8357d774 GPU 3 thread 3 pool 0 GPU3 36.5C | (5s):331.9 (avg):7717.5 Mh/s | A:1 R:0 HW:0 U:600.00/m I:9 q[2011-12-29 08:06:08] Accepted 00000000.d380e8b8.604fe453 GPU 3 thread 3 pool 0 GPU3 37.0C | (5s):337.0 (avg):292.9 Mh/s | A:2 R:0 HW:0 U:22.77/m I:9 [2011-12-29 08:06:08] Accepted 00000000.f779248b.8d5528c7 GPU 2 thread 2 pool 2 GPU2 41.5C 4483RPM | (5s):464.4 (avg):394.7 Mh/s | A:2 R:0 HW:0 U:22.77/m I:9 [2011-12-29 08:06:10] Accepted 00000000.d086e605.b674ab68 GPU 2 thread 2 pool 2 GPU2 42.5C 4499RPM | (5s):385.0 (avg):471.1 Mh/s | A:3 R:0 HW:0 U:34.15/m I:9 ./start.sh: line 5: 23833 Segmentation fault (core dumped) DISPLAY=:0 cgminer -c cgminer.conf 2> logs/$now.log brec@ubuntu:~/miners/cgminer$ gdb core GNU gdb (Ubuntu/Linaro 7.2-1ubuntu11) 7.2 Copyright (C) 2010 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-linux-gnu". For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>... "/home/brec/miners/cgminer/core": not in executable format: File format not recognized (gdb) bt No stack. (gdb) bt full No stack. (gdb) q brec@ubuntu:~/miners/cgminer$
Check the directory, it will be called core.something where something is a number. It will not be called just core.
|
|
|
How are rejects for everyone now with the new code? They should be the lowest ever.
Looks the same for me but I always had low rejects using Bitminter pool & cgminer (<0.1%). 0.1% sounds good. There will only be a difference in this release if multipool is set up, and more so if you have switched from your primary pool.
|
|
|
How are rejects for everyone now with the new code? They should be the lowest ever.
|
|
|
It's in the newer fglrx drivers; details if you make AMD open-source their drivers... > LANG=C gdb cgminer core GNU gdb (GDB) 7.3-debian Copyright (C) 2011 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-linux-gnu". For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>... Reading symbols from /home/reini/root/bitcoin/cgminer_build/cgminer...done. [New LWP 14321] [New LWP 14317] [New LWP 14241] [New LWP 14326]
warning: Can't read pathname for load map: Input/output error. [Thread debugging using libthread_db enabled] Core was generated by `cgminer -c /home/reini/.cgminer/bitcoin.conf'. Program terminated with signal 11, Segmentation fault. #0 0x00007f7ef1c09729 in ?? () from /usr/lib/libamdocl64.so Traceback (most recent call last): File "/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.16-gdb.py", line 59, in <module> from libstdcxx.v6.printers import register_libstdcxx_printers ImportError: No module named libstdcxx.v6.printers (gdb) bt full #0 0x00007f7ef1c09729 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #1 0x00007f7ef1c09778 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #2 0x00007f7ef1bff6c1 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #3 0x00007f7ef1bf6088 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #4 0x00007f7ef1c005b0 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #5 0x00007f7ef1c01260 in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #6 0x00007f7ef1c0a3cc in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #7 0x00007f7ef1c084bd in ?? () from /usr/lib/libamdocl64.so No symbol table info available. #8 0x00007f7ef69dab50 in start_thread (arg=<optimized out>) at pthread_create.c:304 __res = <optimized out> pd = 0x7f7eec178700 unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140183103571712, 4430654472375514548, 140737349963280, 140183103572416, 140183284547648, 7, -4359201957654547020, -4359143591589956172}, mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0, cleanup = 0x0, canceltype = 0}}} not_first_call = <optimized out> freesize = <optimized out> __PRETTY_FUNCTION__ = "start_thread" #9 0x00007f7ef5c743bd in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112 No locals. #10 0x0000000000000000 in ?? () No symbol table info available. (gdb) Well at least I know it's not my code at fault. Shame though...
|
|
|
Hi Con, It looks like api-example.c, api-example.php, API.java/API.class and miner.php didn't get packaged up in the 2.1.0 source tarball. I see they are there in GIT but the README implies they are included in the main distribution.
You're right, thanks. Naughty Kano... anyway I'll fix it in the git tree.
|
|
|
(though q for quitting is broken, I can affirm that)
If your using Windoze that's kind of normal operation for most of us as its a known Windoze bug. There are several older posts mentioning this if you want more details. Sam Looks like people are getting it on linux too now. Anyone want to get a core dump and get a backtrace? (If that's within your capabilities of course). Of course I'd debug it but it doesn't happen here... then again I don't have any GPU mining to test it on right now
|
|
|
-------------------------------------------------------------------------------- [P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit GPU 0: 72.5C 4403RPM | 359.9/361.5Mh/s | A:5435 R:3 HW:0 U:4.99/m I:9 GPU 1: 76.5C | 359.8/361.8Mh/s | A:5413 R:10 HW:0 U:4.97/m I:9 GPU 2: 72.5C 3617RPM | 364.6/361.3Mh/s | A:5284 R:7 HW:0 U:4.85/m I:9 GPU 3: 67.5C | 365.7/361.5Mh/s | A:5349 R:5 HW:0 U:4.91/m I:9 GPU 4: 73.0C 3255RPM | 362.8/361.5Mh/s | A:5391 R:7 HW:0 U:4.95/m I:9 GPU 5: 76.0C | 368.0/361.7Mh/s | A:5417 R:10 HW:0 U:4.97/m I:9 --------------------------------------------------------------------------------
The spacing and misalignment problem is still with us, in this instance due to double-digit rejects for a couple of GPUs. But as indicated by the hyphen borders, LOOK AT ALL THAT WASTED SPACE ON THE RIGHT!!11!!1 Some field widths could be increased. My plan is to fix this; might take a few days (due to my schedule, not coding time). If someone else wants to do it, please let me know. Here we go again The idea was to leave those since they're ever increasing and you'll never be able to guarantee enough column width and alignment.
|
|
|
if not yet done, try: sudo aticonfig --install --adapter=all sudo reboot
Did that after installing fglrx. I know I'm missing something simple. Have you started Xorg?
|
|
|
[2011-12-27 14:19:26] Started cgminer 2.1.0 [2011-12-27 14:19:42] Pool 32 slow/down or URL or credentials invalid [2011-12-27 14:19:42] Long-polling activated for http://us.eclipsemc.com:9009/LP [2011-12-27 14:19:50] Accepted 00000000.65d7e293.179988b8 GPU 4 thread 4 pool 0 ...
As noted yesterday I'm consistently seeing a pool-problem message at startup. My request to have the problem pool identified was granted -- thank you! Alas, now I don't know what the ID means: 32? I have three pools specified in my config. At least 32 is a power of two, which is always comforting. 32 is the donation pool. Thanks for offering a donation
|
|
|
Q: Can you change the autofan/autogpu to change speeds in a different manner? A: The defaults are sane and safe. I'm not interested in changing them further. The starting fan speed is set to 85% in auto-fan mode as a safety precaution, but if a specific fan speed has been set, it will use that first before adjusting automatically. This reminded me that there's something I didn't understand about the fairly recent posts on this subject: If I want to set a specific starting fan speed, I would use the gpu-fan option, right? But if I also use auto-fan, what would cgminer use as a permissible range for the fan speed percentage? E.g., suppose I want a range of 0-85, starting speed of 50, and auto-fan. Can that be done? Perhaps I don't understand what "has been set" refers to in the quoted README excerpt. No, that's a bridge too far... You either set a single value which is the starting speed for auto-fan or it's a static value without auto-fan. Or you set a range which is used by auto-fan.
|
|
|
|