tryphe
Member
Offline
Activity: 116
Merit: 10
|
|
May 08, 2015, 06:03:23 AM Last edit: May 08, 2015, 06:49:57 AM by tryphe |
|
I'm having an issue here with the error: Cuda error in func 'skein512_cpu_setBlock_80' at line 957 : invalid device symbol. Strangely enough, I'm running a 660ti (Compute 3.0) and have ccminer compiling/running x11 just fine with the same setup. Although I don't think it's quite the latest version(Edit: It's actually quite old, but still uses cuda 6.5), it works with nvidia-331 and nvidia-346. All I have to do is export the cuda lib64 path before running. Here are the juicy infos: nobody@abyss:~$ uname -a Linux abyss 3.13.0-37-generic #64-Ubuntu SMP Mon Sep 22 21:28:38 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
"Ubuntu 14.04" nobody@abyss:~$ lspci 06:00.0 VGA compatible controller: NVIDIA Corporation GK104 [GeForce GTX 660 Ti] (rev a1) (prog-if 00 [VGA controller]) Subsystem: Device 196e:1000 Flags: bus master, fast devsel, latency 0, IRQ 45 Memory at fb000000 (32-bit, non-prefetchable) [size=16M] Memory at d8000000 (64-bit, prefetchable) [size=128M] Memory at e6000000 (64-bit, prefetchable) [size=32M] I/O ports at 6c00 [size=128] [virtual] Expansion ROM at e0000000 [disabled] [size=512K] Capabilities: <access denied> Kernel driver in use: nvidia
nobody@abyss:~$ /usr/local/cuda/bin/nvcc --version nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2014 NVIDIA Corporation Built on Thu_Jul_17_21:41:27_CDT_2014 Cuda compilation tools, release 6.5, V6.5.12 Installed packages(no errors): nobody@abyss:~$ dpkg -l | grep 'ii nvidia' ii nvidia-346 346.46-0ubuntu1 amd64 NVIDIA binary driver - version 346.46 ii nvidia-346-dev 346.46-0ubuntu1 amd64 NVIDIA binary Xorg driver development files ii nvidia-346-uvm 346.46-0ubuntu1 amd64 NVIDIA Unified Memory kernel module ii nvidia-modprobe 346.46-0ubuntu1 amd64 Load the NVIDIA kernel driver and create device files ii nvidia-opencl-icd-346 346.46-0ubuntu1 amd64 NVIDIA OpenCL ICD ii nvidia-prime 0.6.2 amd64 Tools to enable NVIDIA's Prime ii nvidia-settings 346.46-0ubuntu1 amd64 Tool for configuring the NVIDIA graphics driver
nobody@abyss:~$ dpkg -l | grep 'ii cuda' ii cuda-6-5 6.5-14 amd64 CUDA 6.5 meta-package ii cuda-command-line-tools-6-5 6.5-14 amd64 CUDA command-line tools ii cuda-core-6-5 6.5-14 amd64 CUDA core tools ii cuda-cublas-6-5 6.5-14 amd64 CUBLAS native runtime libraries ii cuda-cublas-dev-6-5 6.5-14 amd64 CUBLAS native dev links, headers ii cuda-cudart-6-5 6.5-14 amd64 CUDA Runtime native Libraries ii cuda-cudart-dev-6-5 6.5-14 amd64 CUDA Runtime native dev links, headers ii cuda-cufft-6-5 6.5-14 amd64 CUFFT native runtime libraries ii cuda-cufft-dev-6-5 6.5-14 amd64 CUFFT native dev links, headers ii cuda-curand-6-5 6.5-14 amd64 CURAND native runtime libraries ii cuda-curand-dev-6-5 6.5-14 amd64 CURAND native dev links, headers ii cuda-cusparse-6-5 6.5-14 amd64 CUSPARSE native runtime libraries ii cuda-cusparse-dev-6-5 6.5-14 amd64 CUSPARSE native dev links, headers ii cuda-documentation-6-5 6.5-14 amd64 CUDA documentation ii cuda-driver-dev-6-5 6.5-14 amd64 CUDA Driver native dev stub library ii cuda-drivers 346.46-1 amd64 CUDA Driver meta-package ii cuda-license-6-5 6.5-14 amd64 CUDA licenses ii cuda-misc-headers-6-5 6.5-14 amd64 CUDA misc headers ii cuda-npp-6-5 6.5-14 amd64 NPP native runtime libraries ii cuda-npp-dev-6-5 6.5-14 amd64 NPP native dev links, headers ii cuda-runtime-6-5 6.5-14 amd64 CUDA Runtime 6.5 meta-package ii cuda-samples-6-5 6.5-14 amd64 CUDA example applications ii cuda-toolkit-6-5 6.5-14 amd64 CUDA Toolkit 6.5 meta-package ii cuda-visual-tools-6-5 6.5-14 amd64 CUDA visual tools
Here is the full output of your ccminer branch: nobody@abyss:~$ ./log-mine-nv-suprnova.sh *** ccminer 1.6.3-git for nVidia GPUs by tpruvot@github *** Built with the nVidia CUDA SDK 6.5
Originally based on Christian Buchner and Christian H. project Include some of the work of djm34, sp, tsiv and klausT.
BTC donation address: 1AJdfCpLWPNoAMDfHF1wD5y8VgKSSTHxPo (tpruvot)
[2015-05-08 02:00:37] Starting on stratum+tcp://wood.suprnova.cc:1158 [2015-05-08 02:00:37] restart_threads [2015-05-08 02:00:37] NVML application clock feature is allowed [2015-05-08 02:00:37] CUDA GPU#0 matches NVML GPU 0 by busId 6 [2015-05-08 02:00:37] NVML GPU monitoring enabled. [2015-05-08 02:00:37] 1 miner thread started, using 'skein2' algorithm. [2015-05-08 02:00:37] Stratum difficulty set to 0.00390625 [2015-05-08 02:00:38] sleeptime: 500 ms [2015-05-08 02:00:38] DEBUG: job_id=54c50fd 193e xnonce2=00000000 time=02:00:29 [2015-05-08 02:00:38] job 54c50fd 193e target change: ffff000000 (1.0) [2015-05-08 02:00:38] GPU #0: start=00000000 end=000fffff range=000fffff [2015-05-08 02:00:38] DEBUG: job_id=54c50fd 193e xnonce2=01000000 time=02:00:29 [2015-05-08 02:00:38] wood.suprnova.cc:1158 skein2 block 136203 [2015-05-08 02:00:38] restart_threads Cuda error in func 'skein512_cpu_setBlock_80' at line 957 : invalid device symbol.
And here is the script to produce the output. export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64 src/ccminer-skein/ccminer -a skein2 -o stratum+tcp://wood.suprnova.cc:1158 -u tryphe.w0 -p x --debug
'ccminer --version' outputs: (x11 miner) ccminer 2014.06.15 libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
(tpruvot miner) ccminer v1.6.3-git libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
It doesn't matter if I use 331 or 346 to compile/run (either or both). It doesn't matter if I set CUDA_paths or the LD_LIB paths to either the 'cuda-6.5' or 'cuda' directories because they point to the same data. Any ideas? Thanks in advance.
|
|
|
|
bathrobehero
Legendary
Offline
Activity: 2002
Merit: 1051
ICO? Not even once.
|
|
May 08, 2015, 10:53:55 AM |
|
Cuda error in func 'skein512_cpu_setBlock_80' at line 957 : invalid device symbol. That usually means the binaries are not compatible with your card's compute version. Probably 3.5 is required for the binaries. Not 100% sure though.
|
Not your keys, not your coins!
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 08, 2015, 01:10:32 PM |
|
read the last messages in the thread (12 Apr)
|
|
|
|
tryphe
Member
Offline
Activity: 116
Merit: 10
|
|
May 08, 2015, 04:07:48 PM Last edit: May 08, 2015, 04:38:12 PM by tryphe |
|
read the last messages in the thread (12 Apr)
Lol, can't believe I missed that in the search But I did that and the build resulted with this: ...[lots more sm_32 errors] ptxas /tmp/tmpxft_00001194_00000000-5_cuda_neoscrypt.ptx, line 10135; error : Instruction 'shf.r' requires .target sm_32 or higher ptxas fatal : Ptx assembly aborted due to errors make[2]: *** [neoscrypt/cuda_neoscrypt.o] Error 255
It seems to compile fine other than a few algos. I tried a few things until I got the compile to work with changing the nvcc arch command to nvcc_ARCH = -gencode=arch=compute_32,code=\"sm_32,compute_32\" Which compiles, but still throws the same error on runtime, obviously because my arch is compute_30. Same thing happens if I specify compute_35/sm_35. So I tried a few other things, like leaving the rest of the Makefile.am alone, and only changing the arch target for skein: skein2.o: skein2.cu $(NVCC) -I. @CUDA_INCLUDES@ @CUDA_CFLAGS@ -gencode=arch=compute_30,code=\"sm_30,compute_30\" --maxrregcount=64 -o $@ -c $< Same thing. Compile runs without a hitch, but the crash still happens with skein512_cpu_setBlock_80. Any other ideas?
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 08, 2015, 05:18:27 PM |
|
hmm right, you need to checkout a previous tagged version, 1.6.3 is still under dev... for this reason (neoscrypt is not compatible with SM 3.0)
|
|
|
|
tryphe
Member
Offline
Activity: 116
Merit: 10
|
|
May 09, 2015, 04:26:58 AM |
|
hmm right, you need to checkout a previous tagged version, 1.6.3 is still under dev... for this reason (neoscrypt is not compatible with SM 3.0)
Aha, well now I don't feel so confused, haha. Surely I can get it working now! Thanks for the help and good work on the project
|
|
|
|
bensam1231
Legendary
Offline
Activity: 1764
Merit: 1024
|
|
May 10, 2015, 07:29:41 AM |
|
Would you consider adding support for Cryptonote? TSIV hasn't updated his miner in ages and Monero is still very lucrative.
|
I buy private Nvidia miners. Send information and/or inquiries to my PM box.
|
|
|
miningpoolhub
Legendary
Offline
Activity: 1456
Merit: 1006
Mining Pool Hub
|
|
May 11, 2015, 02:11:36 PM Last edit: May 11, 2015, 02:28:19 PM by miningpoolhub |
|
We are running http://miningpoolhub.comAs we provide auto switching between multi algo like this batch file. :start ccminer-30-35-50-52.exe -r 0 -a x11 -o stratum+tcp://hub.miningpoolhub.com:12007 -u coincoin.ccminer -p x ccminer-30-35-50-52.exe -r 0 -a x13 -o stratum+tcp://hub.miningpoolhub.com:12008 -u coincoin.ccminer -p x ccminer-30-35-50-52.exe -r 0 -a myr-gr -o stratum+tcp://hub.miningpoolhub.com:12005 -u coincoin.ccminer -p x sleep 3000 timeout 3 goto start ccminer crashes when connection is refused. It works time to time but crashes time to time. Actually https://github.com/cbuchner1/ccminer doesn't crashes but https://github.com/tpruvot/ccminer crashes. Our server send authorize failure message and disconnect when multi algo switching need to work. Please fix this bug. 'coincoin' is our test username. You can test it right away without setting hub or anything on our site.
|
|
|
|
|
flipclip
Member
Offline
Activity: 111
Merit: 10
|
|
May 26, 2015, 01:15:33 AM |
|
From a fresh git clone I'm getting a compiling error (Ubuntu 14.04: compiled before, and can compile other current branches). I'm not sure if it is from my wonky machine but: nvcc -I. -I/usr/local/cuda/include -gencode=arch=compute_20,code=\"sm_21,compute_20\" -o scrypt/salsa_kernel.o -c scrypt/salsa_kernel.cu nvcc -I. -I/usr/local/cuda/include -gencode=arch=compute_20,code=\"sm_20,compute_20\" -o scrypt/test_kernel.o -c scrypt/test_kernel.cu In file included from scrypt/salsa_kernel.h:10:0, from scrypt/salsa_kernel.cu:13: ./miner.h:14:21: fatal error: jansson.h: No such file or directory #include <jansson.h> ^ compilation terminated. make[2]: *** [scrypt/salsa_kernel.o] Error 1 make[2]: *** Waiting for unfinished jobs.... If I updated the Makefile.am with: line 130: - $(NVCC) -I. @CUDA_INCLUDES@ @CUDA_CFLAGS@ -gencode=arch=compute_20,code=\"sm_21,compute_20\" -o $@ -c $< + $(NVCC) -gencode=arch=compute_20,code=\"sm_21,compute_20\" @CUDA_INCLUDES@ -I. @CUDA_CFLAGS@ $(JANSSON_INCLUDES) -o $@ -c $< It compiled, up to the next part of the Makefile.am "scrypt/test_kernel.o: scrypt/test_kernel.cu", which failed, but when edited similarily, along with the next three kernels, everything compiled. Again, not sure if it is just my wonky machine but wanted to mention it.
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 26, 2015, 08:55:35 AM |
|
indeed thanks... dont had the problem here but seems logic.
|
|
|
|
george37
Newbie
Offline
Activity: 1
Merit: 0
|
|
May 26, 2015, 12:17:05 PM |
|
In the latest code, v1.6.4, I'm getting the following build error: g++ -DHAVE_CONFIG_H -I. -fopenmp -pthread -fno-strict-aliasing -I/usr/local/cuda/include -DUSE_WRAPNVML -DSCRYPT_KECCAK512 -DSCRYPT_CHACHA -DSCRYPT_CHOOSE_COMPILETIME -O3 -march=native -D_REENTRANT -falign-functions=16 -falign-jumps=16 -falign-labels=16 -MT ccminer-api.o -MD -MP -MF .deps/ccminer-api.Tpo -c -o ccminer-api.o `test -f 'api.cpp' || echo './'`api.cpp ccminer.cpp: In function 'void show_usage_and_exit(int)': ccminer.cpp:2583:15: warning: format not a string literal and no format arguments [-Wformat-security] ccminer.cpp:2585:22: warning: format not a string literal and no format arguments [-Wformat-security] ccminer.cpp: In function 'bool parse_pool_array(json_t*)': ccminer.cpp:3042:32: error: 'json_array_foreach' was not declared in this scope ccminer.cpp:3043:2: error: expected ';' before '{' token ccminer.cpp:3513:1: error: expected '}' at end of input make[2]: *** [ccminer-ccminer.o] Error 1 make[2]: *** Waiting for unfinished jobs.... mv -f .deps/ccminer-hefty1.Tpo .deps/ccminer-hefty1.Po mv -f .deps/ccminer-api.Tpo .deps/ccminer-api.Po mv -f .deps/ccminer-util.Tpo .deps/ccminer-util.Po make[2]: Leaving directory `/home/ubuntu/ccminer' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/ubuntu/ccminer' make: *** [all] Error 2
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 26, 2015, 12:20:58 PM Last edit: May 26, 2015, 12:33:11 PM by Epsylon3 |
|
i guess you have an old version of jansson... there is the 2.6 in compat folder
compat/jansson/jansson.h:151:#define json_array_foreach(array, index, value) \
edit: i build with the 2.5 on my ubuntu 14.04.2 machine : libjansson-dev:amd64 2.5-2
edit2: my debian 7 machine also build fine with the 2.3.1-2
|
|
|
|
antimony121
Newbie
Offline
Activity: 6
Merit: 0
|
|
May 26, 2015, 01:35:11 PM |
|
1.6.4 compiled fine with my machine, ubuntu 15.04 x64 just wondering if the password are parsed in a different way between 1.6.3 and 1.6.4? I am mining with nicehash with the option (multialgo weightings) --pass="f0=0;f2=0;f3=2.7;f4=1.9;f5=99;f6=1.8;f7=7.9;f8=0;f9=0.45;f10=0;f11=4.3;f12=3.9" under ccminer 1.6.3 i get assigned quark with their multialgo, but i get assigned lyre2re under ccminer 1.6.4 the pool option doesn't respect the "algo" parameter set inside the pool, guess it haven't been implemented yet?
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 26, 2015, 01:36:57 PM |
|
its limited to 64 characters indeed, i was expecting this problem... will increase it to 128 in the next version
|
|
|
|
zanfr
Newbie
Offline
Activity: 8
Merit: 0
|
|
May 26, 2015, 08:58:27 PM |
|
So I am trying to do some multialgo switching on nicehash, here is my test config file: But, after attempting to connect to x11.eu.nicehash.com et predictably getting dropped, it attempts to mine x11 whilst connected to quark.eu.nicehash.com... Seems it only cares about the first "algo" it finds... Any ideas? { "pools" : [ { "name": "nicehash x11", "algo" : "x11", "url" : "stratum+tcp://x11.eu.nicehash.com:4336", "user" : "1xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx", "pass" : "f0=0;f2=0;f3=3;f12=4", "intensity" : 16 }, { "name": "nicehash quark", "algo" : "quark", "url" : "stratum+tcp://quark.eu.nicehash.com:4345", "user" : "1xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx", "pass" : "f0=0;f2=0;f3=3;f12=4", "intensity" : 18.1 } ], "api-remote": false }
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 26, 2015, 09:12:01 PM |
|
Multi pool doesnt means algo switching... Ccminer is not ready for that yet..
Require changes in all the algos to free the gpu. Hmm in fact maybe not with cudaDeviceReset()
|
|
|
|
antimony121
Newbie
Offline
Activity: 6
Merit: 0
|
|
May 27, 2015, 10:04:41 PM |
|
just compiled the latest git, password is working now, but it doesn't seems to parse the -r (--retries) parameter properly, i set -r 0 (or --retries=0) and it just get ignore with the default value (retry indefinitely) [2015-05-28 07:49:35] Stratum authentication failed [2015-05-28 07:49:35] Stratum connect timeout, failover... [2015-05-28 07:49:35] No other pools to try... [2015-05-28 07:49:35] ...retry after 30 seconds [2015-05-28 07:50:06] Stratum authentication failed [2015-05-28 07:50:06] Stratum connect timeout, failover... [2015-05-28 07:50:06] No other pools to try... [2015-05-28 07:50:06] ...retry after 30 seconds
if i put in two pools it just loops the two pools indefinitely, regardless of the "retries" : "0" option [2015-05-28 07:54:02] Stratum authentication failed [2015-05-28 07:54:02] Stratum connect timeout, failover... [2015-05-28 07:54:02] Switch to stratum pool 1: eu [2015-05-28 07:54:05] Stratum authentication failed [2015-05-28 07:54:05] Stratum connect timeout, failover... [2015-05-28 07:54:05] Switch to stratum pool 0: usa
thanks!
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
May 28, 2015, 04:51:57 AM |
|
i fixed the first case... for the second one... hmm need some more work/brain storm
|
|
|
|
IMweasel
Newbie
Offline
Activity: 29
Merit: 0
|
|
May 29, 2015, 12:24:29 PM |
|
Love your work! But i've tried to solo UTC and nothing happens with CCminer. Good old Cudaminer does the job just fine. Though i'd rather use CC. Tried 1.62,1.63,1.64, all the same. Using the -D flag i immediately see an error...can you fix it? Thanks in advance! JSON-RPC call failed: getblocktemplate [params] Returns data needed to construct a block to work on: "version" : block version "previousblockhash" : hash of current highest block "transactions" : contents of non-coinbase transactions that should be included in the next block "coinbaseaux" : data that should be included in coinbase "coinbasevalue" : maximum allowable input to coinbase transaction, including the generation award and transaction fees "target" : hash target "mintime" : minimum timestamp appropriate for next block "curtime" : current timestamp "mutable" : list of ways the block template may be changed "noncerange" : range of valid nonces "sigoplimit" : limit of sigops in blocks "sizelimit" : limit of block size "bits" : compressed target of next block "height" : height of the next block See https://en.bitcoin.it/wiki/BIP_0022 for full specification.
|
|
|
|
|