luka1002
Newbie
Offline
Activity: 14
Merit: 0
|
|
November 03, 2017, 10:51:43 PM |
|
yep, i dont put it for now because that breaks previous cuda builds... and increase the binary size a lot for nothing on windows
you maybe should add build ONLY for cuda 9, it would be easier for a lot of people.
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
November 03, 2017, 10:52:40 PM |
|
its the purpose of cuda-9 branch (was present before cuda 9 release)
|
|
|
|
The_Immortal
Newbie
Offline
Activity: 55
Merit: 0
|
|
November 05, 2017, 05:08:15 AM Last edit: November 05, 2017, 06:30:26 AM by The_Immortal |
|
Guys... Could you help me please?... What did I make wrong? 1. I installed on my Ubuntu CUDA 9.0.176. 2. I downloaded the source from the following link (ccminer-cuda-9). 3. I edited Makefile.am: ...
#ccminer_CPPFLAGS += -DUSE_LIBSODIUM #ccminer_LDFLAGS += -Lequi/lib #ccminer_LDADD += -lsodium ccminer_LDADD += -lcuda
nvcc_ARCH = -gencode=arch=compute_61,code=\"sm_61,compute_61\"
#nvcc_ARCH += -gencode=arch=compute_52,code=\"sm_52,compute_52\" #nvcc_ARCH += -gencode=arch=compute_35,code=\"sm_35,compute_35\" #nvcc_ARCH += -gencode=arch=compute_30,code=\"sm_30,compute_30\" #nvcc_ARCH += -gencode=arch=compute_20,code=\"sm_21,compute_20\"
nvcc_FLAGS = $(nvcc_ARCH) @CUDA_INCLUDES@ -I. @CUDA_CFLAGS@ nvcc_FLAGS += $(JANSSON_INCLUDES) --ptxas-options="-v"
... 4. I compiled ccminer. And here is strange thing - I got ccminer with the size about 18 MB. But last time (when i used CUDA 7.5) I got it with 34-35 MB... Myabe is it because of my Makefile.am? Ok, screw it, right? 5. ccminer --version showed everything is OK. Now I copied just compiled ccminer and lib directory (with 3 files into it: libcudart.so.9.0 + libcudart.so.9.0 + libcudart.so.9.0.176) to other PC (/home/user/ccminer/) without installed CUDA on it. I did: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/user/ccminer/lib And then: /home/user/ccminer/ccminer --version Illegal instruction Why? I've compiled ccminer on PC with Intel CPU. And I copied it toPC with AMD CPU. But before you've said that CPU doesn't matter, doesn't it?
|
|
|
|
konqueror
Member
Offline
Activity: 99
Merit: 10
|
|
November 05, 2017, 01:24:19 PM |
|
Guys... Could you help me please?... What did I make wrong? 1. I installed on my Ubuntu CUDA 9.0.176. 2. I downloaded the source from the following link (ccminer-cuda-9). 3. I edited Makefile.am: ...
#ccminer_CPPFLAGS += -DUSE_LIBSODIUM #ccminer_LDFLAGS += -Lequi/lib #ccminer_LDADD += -lsodium ccminer_LDADD += -lcuda
nvcc_ARCH = -gencode=arch=compute_61,code=\"sm_61,compute_61\"
#nvcc_ARCH += -gencode=arch=compute_52,code=\"sm_52,compute_52\" #nvcc_ARCH += -gencode=arch=compute_35,code=\"sm_35,compute_35\" #nvcc_ARCH += -gencode=arch=compute_30,code=\"sm_30,compute_30\" #nvcc_ARCH += -gencode=arch=compute_20,code=\"sm_21,compute_20\"
nvcc_FLAGS = $(nvcc_ARCH) @CUDA_INCLUDES@ -I. @CUDA_CFLAGS@ nvcc_FLAGS += $(JANSSON_INCLUDES) --ptxas-options="-v"
... 4. I compiled ccminer. And here is strange thing - I got ccminer with the size about 18 MB. But last time (when i used CUDA 7.5) I got it with 34-35 MB... Myabe is it because of my Makefile.am? Ok, screw it, right? 5. ccminer --version showed everything is OK. Now I copied just compiled ccminer and lib directory (with 3 files into it: libcudart.so.9.0 + libcudart.so.9.0 + libcudart.so.9.0.176) to other PC (/home/user/ccminer/) without installed CUDA on it. I did: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/user/ccminer/lib And then: /home/user/ccminer/ccminer --version Illegal instruction Why? I've compiled ccminer on PC with Intel CPU. And I copied it toPC with AMD CPU. But before you've said that CPU doesn't matter, doesn't it? Did you use
|
|
|
|
The_Immortal
Newbie
Offline
Activity: 55
Merit: 0
|
|
November 05, 2017, 06:01:29 PM |
|
Nope. As I said before all I made is 3. I edited Makefile.am: ...
#ccminer_CPPFLAGS += -DUSE_LIBSODIUM #ccminer_LDFLAGS += -Lequi/lib #ccminer_LDADD += -lsodium ccminer_LDADD += -lcuda
nvcc_ARCH = -gencode=arch=compute_61,code=\"sm_61,compute_61\"
#nvcc_ARCH += -gencode=arch=compute_52,code=\"sm_52,compute_52\" #nvcc_ARCH += -gencode=arch=compute_35,code=\"sm_35,compute_35\" #nvcc_ARCH += -gencode=arch=compute_30,code=\"sm_30,compute_30\" #nvcc_ARCH += -gencode=arch=compute_20,code=\"sm_21,compute_20\"
nvcc_FLAGS = $(nvcc_ARCH) @CUDA_INCLUDES@ -I. @CUDA_CFLAGS@ nvcc_FLAGS += $(JANSSON_INCLUDES) --ptxas-options="-v"
... Should I put into Makefile.am as well?
|
|
|
|
joblo
Legendary
Offline
Activity: 1470
Merit: 1114
|
|
November 05, 2017, 07:34:17 PM Last edit: November 05, 2017, 07:55:18 PM by joblo |
|
Should I put into Makefile.am as well? That will just makes it worse. What you want is a generic build but, straight from gcc... There is no -march=generic option because -march indicates the instruction set the compiler can use, and there is no generic instruction set applicable to all processors.There is just too much that is different between AMD and Intel processors. Edit: If you're only compiling for that one PC you can specify -march for that CPU. https://gcc.gnu.org/onlinedocs/gcc/x86-Options.htmlEdit2: It might also be useful to mention what CPUs are inolved. If the AMD newer than Piledriver -march-core2 should work. I have had no success with Piledriver and older.
|
|
|
|
Epsylon3 (OP)
Legendary
Offline
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
|
|
November 05, 2017, 07:37:17 PM |
|
replace march=native by march=core2 in configure.sh then use ./build.sh
|
|
|
|
joblo
Legendary
Offline
Activity: 1470
Merit: 1114
|
|
November 05, 2017, 07:44:52 PM |
|
replace march=native by march=core2 in configure.sh then use ./build.sh
That seems to be the most generic I've found.
|
|
|
|
The_Immortal
Newbie
Offline
Activity: 55
Merit: 0
|
|
November 05, 2017, 11:56:57 PM |
|
replace march=native by march=core2 in configure.sh then use ./build.sh
Finally, it works now! Thank you very much!
|
|
|
|
insaniak
|
|
November 08, 2017, 10:01:46 AM |
|
What would cause this error? GPU #3: aes_cpu_init out of memory
|
|
|
|
chrysophylax
Legendary
Offline
Activity: 2870
Merit: 1091
--- ChainWorks Industries ---
|
|
November 08, 2017, 01:07:39 PM |
|
What would cause this error? GPU #3: aes_cpu_init out of memory
Intensity is too high. #crysx
|
|
|
|
Elange
Newbie
Offline
Activity: 73
Merit: 0
|
|
November 08, 2017, 03:37:14 PM |
|
Hi,
I'm using tpruvot 2.2.2 unde Ubuntu 16.04. I have suddenly found out, some algos are about 15-20% slower on AMD boards than Intel boards.
Two rigs with same cards: 4 pcs 1060 3gb Palit Samsung card + 1 pcs 1050 ti card. All rigs use PCIE splitters.
Rig using H87 Intel mobo G1840 CPU produces around 96 khs on Lyra2v2, whereas rig on a G43-970 AMD Athlon 245 II CPU mobo produces only around 80 khs. Overclock settings are the same on both rigs: mem +800, core +240. Drivers are 384.66, CUDA 8.0. I compiled ccminer on one rig and copied over to others.
Same goes for groestl, myr-gr, etc. However, Neoscrypt produces around 2500 khs on all rigs no matter what platform.
Any ideas why this could be happening? Would it make sense to compile on AMD rigs individually (maybe some AMD CPU optimizations out there?) I've heard it, ccminer is pretty hard on CPU, could that be the case that AMD CPU is not fast enough to catch up?
Thanks!!
p.s. There is no significant performance difference on ETHash (Claymore) or Equihash (EWBF or ZM).
|
|
|
|
schap
Newbie
Offline
Activity: 14
Merit: 0
|
|
November 08, 2017, 03:50:35 PM |
|
Guys im having a tough time getting this to run with my Nvidia Tesla 2090, can anyone help please. downloaded the Cuda 8 version off git and this is error I am getting..... [2017-11-07 20:02:04] Starting on stratum+tcp://yiimp.eu:3737 [2017-11-07 20:02:04] NVML GPU monitoring enabled. [2017-11-07 20:02:04] 1 miner thread started, using 'x17' algorithm. [2017-11-07 20:02:20] Stratum difficulty set to 0.016 [2017-11-07 20:02:20] x17 block 1630213, diff 1136.048 [2017-11-07 20:02:20] GPU #0: Intensity set to 20, 1048576 cuda threads [2017-11-07 20:02:20] GPU #0: aes_cpu_init invalid device symbol [2017-11-07 20:02:21] GPU #0: aes_cpu_init invalid device symbol Cuda error in func 'x13_hamsi512_cpu_init' at line 685 : invalid device symbol.
I also changed the MakeFile.am to match the correct compatibility which should be nvcc_ARCH = -gencode=arch=compute_20,code="sm_20,compute_20"
thanks for any help
|
|
|
|
chrysophylax
Legendary
Offline
Activity: 2870
Merit: 1091
--- ChainWorks Industries ---
|
|
November 08, 2017, 03:55:48 PM |
|
Guys im having a tough time getting this to run with my Nvidia Tesla 2090, can anyone help please. downloaded the Cuda 8 version off git and this is error I am getting..... [2017-11-07 20:02:04] Starting on stratum+tcp://yiimp.eu:3737 [2017-11-07 20:02:04] NVML GPU monitoring enabled. [2017-11-07 20:02:04] 1 miner thread started, using 'x17' algorithm. [2017-11-07 20:02:20] Stratum difficulty set to 0.016 [2017-11-07 20:02:20] x17 block 1630213, diff 1136.048 [2017-11-07 20:02:20] GPU #0: Intensity set to 20, 1048576 cuda threads [2017-11-07 20:02:20] GPU #0: aes_cpu_init invalid device symbol [2017-11-07 20:02:21] GPU #0: aes_cpu_init invalid device symbol Cuda error in func 'x13_hamsi512_cpu_init' at line 685 : invalid device symbol.
I also changed the MakeFile.am to match the correct compatibility which should be nvcc_ARCH = -gencode=arch=compute_20,code="sm_20,compute_20"
thanks for any help
This model is no longer listed in the nVidia CUDA compute list - https://developer.nvidia.com/cuda-gpus/If it is the next version up from the 2075, then you will find at Compute20, it is too old for even ccminer-tpruvot to compile to. Have you tried adding this compute level in the Makefile.am, then compiling? tpruvot, you would know better as to whether this Compute level is supported or not. #crysx
|
|
|
|
schap
Newbie
Offline
Activity: 14
Merit: 0
|
|
November 08, 2017, 04:05:43 PM |
|
Yes i put in there that I did add nvcc_ARCH = -gencode=arch=compute_20,code="sm_20,compute_20"
|
|
|
|
schap
Newbie
Offline
Activity: 14
Merit: 0
|
|
November 08, 2017, 04:12:36 PM |
|
Guys im having a tough time getting this to run with my Nvidia Tesla 2090, can anyone help please. downloaded the Cuda 8 version off git and this is error I am getting..... [2017-11-07 20:02:04] Starting on stratum+tcp://yiimp.eu:3737 [2017-11-07 20:02:04] NVML GPU monitoring enabled. [2017-11-07 20:02:04] 1 miner thread started, using 'x17' algorithm. [2017-11-07 20:02:20] Stratum difficulty set to 0.016 [2017-11-07 20:02:20] x17 block 1630213, diff 1136.048 [2017-11-07 20:02:20] GPU #0: Intensity set to 20, 1048576 cuda threads [2017-11-07 20:02:20] GPU #0: aes_cpu_init invalid device symbol [2017-11-07 20:02:21] GPU #0: aes_cpu_init invalid device symbol Cuda error in func 'x13_hamsi512_cpu_init' at line 685 : invalid device symbol.
I also changed the MakeFile.am to match the correct compatibility which should be nvcc_ARCH = -gencode=arch=compute_20,code="sm_20,compute_20"
thanks for any help
This model is no longer listed in the nVidia CUDA compute list - https://developer.nvidia.com/cuda-gpus/If it is the next version up from the 2075, then you will find at Compute20, it is too old for even ccminer-tpruvot to compile to. Have you tried adding this compute level in the Makefile.am, then compiling? tpruvot, you would know better as to whether this Compute level is supported or not. #crysx Here is what some of the MakeFile.am looks like if HAVE_WINDOWS ccminer_SOURCES += compat/winansi.c endif ccminer_LDFLAGS = $(PTHREAD_FLAGS) @CUDA_LDFLAGS@ ccminer_LDADD = @LIBCURL@ @JANSSON_LIBS@ @PTHREAD_LIBS@ @WS2_LIBS@ @CUDA_LIBS@ @OPENMP_CFLAGS@ @LIBS@ $(nvml_libs) ccminer_CPPFLAGS = @LIBCURL_CPPFLAGS@ @OPENMP_CFLAGS@ $(CPPFLAGS) $(PTHREAD_FLAGS) -fno-strict-aliasing $(JANSSON_INCLUDES) $(DEF_INCLUDES) $(nvml_defs) #nvcc_ARCH = -gencode=arch=compute_50,code=\"sm_50,compute_50\" nvcc_ARCH = -gencode=arch=compute_20,code=\"sm_20,compute_20\" #nvcc_ARCH += -gencode=arch=compute_61,code=\"sm_61,compute_61\" #nvcc_ARCH += -gencode=arch=compute_52,code=\"sm_52,compute_52\" #nvcc_ARCH += -gencode=arch=compute_35,code=\"sm_35,compute_35\" #nvcc_ARCH += -gencode=arch=compute_30,code=\"sm_30,compute_30\" #nvcc_ARCH += -gencode=arch=compute_20,code=\"sm_21,compute_20\" nvcc_FLAGS = $(nvcc_ARCH) @CUDA_INCLUDES@ -I. @CUDA_CFLAGS@ nvcc_FLAGS += $(JANSSON_INCLUDES) --ptxas-options="-v"
|
|
|
|
joblo
Legendary
Offline
Activity: 1470
Merit: 1114
|
|
November 08, 2017, 05:13:25 PM |
|
Guys im having a tough time getting this to run with my Nvidia Tesla 2090, can anyone help please. downloaded the Cuda 8 version off git and this is error I am getting..... [2017-11-07 20:02:04] Starting on stratum+tcp://yiimp.eu:3737 [2017-11-07 20:02:04] NVML GPU monitoring enabled. [2017-11-07 20:02:04] 1 miner thread started, using 'x17' algorithm. [2017-11-07 20:02:20] Stratum difficulty set to 0.016 [2017-11-07 20:02:20] x17 block 1630213, diff 1136.048 [2017-11-07 20:02:20] GPU #0: Intensity set to 20, 1048576 cuda threads [2017-11-07 20:02:20] GPU #0: aes_cpu_init invalid device symbol [2017-11-07 20:02:21] GPU #0: aes_cpu_init invalid device symbol Cuda error in func 'x13_hamsi512_cpu_init' at line 685 : invalid device symbol.
I also changed the MakeFile.am to match the correct compatibility which should be nvcc_ARCH = -gencode=arch=compute_20,code="sm_20,compute_20"
thanks for any help
This model is no longer listed in the nVidia CUDA compute list - https://developer.nvidia.com/cuda-gpus/It's listed on the legacy page as compute 2.0. https://developer.nvidia.com/cuda-legacy-gpusSince it's not the compute version perhaps it's the drivers or the cuda version. As a legacy product it's probably no longer supported in the latest drivers and cuda. You need to go back in time with all the SW.
|
|
|
|
pallas
Legendary
Offline
Activity: 2716
Merit: 1094
Black Belt Developer
|
|
November 08, 2017, 09:48:47 PM |
|
Cuda 9 does not support compute 2.
|
|
|
|
schap
Newbie
Offline
Activity: 14
Merit: 0
|
|
November 09, 2017, 12:55:48 AM |
|
Cuda 9 does not support compute 2.
What about 8? I have 8 installed
|
|
|
|
thecheapone
Newbie
Offline
Activity: 5
Merit: 0
|
|
November 09, 2017, 01:43:01 PM |
|
I am a micro miner, one GPU card in my desktop computer, and no programmer, just an "end consumer". I am using the precompiled windows 64 version of ccminer 2.2.2 from GitHub. The miner works, but I am unable to increase the intensity level when scrypt mining. No matter what intensity I set in the command line using the -i function, no matter where in the command line I put the -i function, and even though it initially assigns more threads, it then reverts to a different (?default) intensity a few lines down. My gpu doesn't even warm up, and the hash rate is low. Rather than overclocking my GPU, I would prefer to increase the intensity so that it works harder. How do I get it to actually run with the intensity settings I want? This is what I see:
D:\ccminer-x64-2.2.2-cuda9>ccminer-x64 -a scrypt -o stratum+tcp://us.multipool.us:XXX -u theXXX.1 -p XXX -i 14.99
*** ccminer 2.2.2 for nVidia GPUs by tpruvot@github *** Built with VC++ 2013 and nVidia CUDA SDK 9.0 64-bits Originally based on Christian Buchner and Christian H. project Include some kernels from alexis78, djm34, djEzo, tsiv and krnlx.
BTC donation address: 1AJdfCpLWPNoAMDfHF1wD5y8VgKSSTHxPo (tpruvot)
[2017-11-08 06:34:23] Adding 16128 threads to intensity 14, 32512 cuda threads [2017-11-08 06:34:23] Starting on stratum+tcp://us.multipool.us:XXX [2017-11-08 06:34:23] NVML GPU monitoring enabled. [2017-11-08 06:34:23] NVAPI GPU monitoring enabled. [2017-11-08 06:34:23] 1 miner thread started, using 'scrypt' algorithm. [2017-11-08 06:34:24] Stratum difficulty set to 131072 (2.00000) [2017-11-08 06:34:24] scrypt block 1631827, diff 10898.083 [2017-11-08 06:34:25] GPU #0: 32 hashes / 4.0 MB per warp. [2017-11-08 06:34:26] GPU #0: Performing auto-tuning, please wait 2 minutes... [2017-11-08 06:34:26] GPU #0: maximum total warps (BxW): 1484 [2017-11-08 06:34:45] Stratum difficulty set to 256 (0.00391) [2017-11-08 06:34:45] scrypt block 1631828, diff 8720.082 [2017-11-08 06:35:09] scrypt block 1631829, diff 8720.082 [2017-11-08 06:36:47] GPU #0: 589791.16 hash/s with configuration T12x11 [2017-11-08 06:39:34] scrypt block 1631830, diff 8720.082 [2017-11-08 06:39:34] GPU #0: using launch configuration T12x11 [2017-11-08 06:39:34] scrypt block 1631831, diff 8720.082 [2017-11-08 06:39:34] GPU #0: Intensity set to 12.0313, 4224 cuda threads [2017-11-08 06:39:34] scrypt factor set to 9 (1024)
I XXX'd out the actual pool, worker name, password, but you can see where I attempted to set the intensity to 14.99 in the command line, was assigned extra threads, then lower down the intensity was changed to 12.0313 with only a fraction of the threads. I have also tried with setting the T levels myself in advance, so that the auto-tuning does not occur, same result. Any help would be appreciated. Thank you in advance.
|
|
|
|
|