Topic: ethminer-0.9.41-genoil-1.1 - page 51. (Read 397387 times)

Hopfully quick question about genoil ethminer...

I just pulled the latest genoil ethminer from github and built it without any problems.  I am trying to stratum mine at Dwarfpool directly with command:

  ethminer -G -S exp-us.dwarfpool.com:8018 -O .rig3:

which according to everything Ive read here should work fine.  However, I can get a stratum connection but I never get a work package:

Genoil's ethminer 0.9.41-genoil-1.0.8
=====================================================================
Forked from github.com/ethereum/cpp-ethereum
CUDA kernel ported from Tim Hughes' OpenCL kernel
With contributions from nerdralph, RoBiK, tpruvot and sp_

Please consider a donation to:
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d

  ?  10:11:59|ethminer  Connecting to stratum server exp-us.dwarfpool.com:8018
  ?  10:11:59|stratum  Connected to stratum server exp-us.dwarfpool.com : 8018
  ?  10:11:59|stratum  Starting farm
  ?  10:11:59|stratum  Subscribed to stratum server
  ?  10:11:59|stratum  Authorized worker .rig3
  ?  10:11:59|stratum  Discarding incomplete response
  m  10:12:01|ethminer  Waiting for work package...
  m  10:12:03|ethminer  Waiting for work package...
  m  10:12:05|ethminer  Waiting for work package...
  m  10:12:07|ethminer  Waiting for work package...
  m  10:12:09|ethminer  Waiting for work package...
  m  10:12:11|ethminer  Waiting for work package...
  m  10:12:13|ethminer  Waiting for work package...
  m  10:12:15|ethminer  Waiting for work package...
  m  10:12:17|ethminer  Waiting for work package...
  m  10:12:19|ethminer  Waiting for work package...
  m  10:12:21|ethminer  Waiting for work package...
  m  10:12:23|ethminer  Waiting for work package...

It is happening with both Ethereum and Expanse pools. Trying to figure out if its my issue or is Dwarfpool just weird.  Is anyone else having this issue and if so have you been able to work around it and how?

Ok, went and downloaded the latest Nvidia driver(361.45.11) .run file, managed to get it installed and not trash the X Server.  Then looped back and went through the install steps again for ethminer.  This time a different error on the cmake:

CMake Error at /usr/share/cmake-3.2/Modules/FindCUDA.cmake:586 (message):
  Specify CUDA_TOOLKIT_ROOT_DIR
Call Stack (most recent call first):
  libethash-cuda/CMakeLists.txt:3 (FIND_PACKAGE)

cmake command:
cmake -DBUNDLE=cudaminer -DCOMPUTE=50 ..

thx . MSVC didn't need those. I blindly applied your patch, don't have time to spin up a Linux instance now.

So the driver directly from Nvidia... Dang, I tried to do that yesterday and it jacked up X Server and I had to go back to square one... GRRRRRRR.... 

Thanks for the feedback, let me see if I can figure out how to get that Nvidia driver installed without trashing X again.

The 352 driver is from august 2015. The cuda 7.5 toolkit is from september 2015. I think You need a driver that has a releasedate after sept. 2015

It is the 750ti.  I loaded the 352 driver I believe from the Ubuntu repo.  This is running on Ubuntu 14.04.

What card are you testing? If you are on a old driver you need to upgrade to the latest

Here is the output from using a -U
Genoil's ethminer 0.9.41-genoil-1.1.3
=====================================================================
Forked from github.com/ethereum/cpp-ethereum
CUDA kernel ported from Tim Hughes' OpenCL kernel
With contributions from nerdralph, RoBiK, tpruvot and sp_

Please consider a donation to:
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d

CUDA error in func 'getNumDevices' at line 112 : no CUDA-capable device is detected.
terminate called after throwing an instance of 'std::runtime_error'
  what():  no CUDA-capable device is detected
Aborted (core dumped)

Add -U to the commandline

Thanks for the pointer on the template line fix.  It compiled, but now looks like I am back to:

CUDA error in func 'getNumDevices' at line 112 : no CUDA-capable device is detected.
terminate called after throwing an instance of 'std::runtime_error'
  what():  no CUDA-capable device is detected
Aborted (core dumped)


Here is my cmake:
cmake -DBUNDLE=cudaminer -DCOMPUTE=50 ..

During the make it output the following:
In file included from /home/buchanaw/Documents/cpp-ethereum/ethminer/main.cpp:39:0:
/home/buchanaw/Documents/cpp-ethereum/ethminer/MinerAux.h: In member function ‘void MinerCLI::doStratum()’:
/home/buchanaw/Documents/cpp-ethereum/ethminer/MinerAux.h:1023:14: warning: suggest explicit braces to avoid ambiguous ‘else’ [-Wparentheses]
      else if (client.waitState() == MINER_WAIT_STATE_WORK)
              ^
/home/buchanaw/Documents/cpp-ethereum/ethminer/MinerAux.h:1058:14: warning: suggest explicit braces to avoid ambiguous ‘else’ [-Wparentheses]
      else if (client.waitState() == MINER_WAIT_STATE_WORK)
              ^

However it did appear to complete.

UPDATE:  Just re-ran cmake with just -DBUNDLE=cudaminer and did the make and now it outputs the following:

Genoil's ethminer 0.9.41-genoil-1.1.3
=====================================================================
Forked from github.com/ethereum/cpp-ethereum
CUDA kernel ported from Tim Hughes' OpenCL kernel
With contributions from nerdralph, RoBiK, tpruvot and sp_

Please consider a donation to:
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d

[OPENCL]:No OpenCL platforms found
No GPU device with sufficient memory was found. Can't GPU mine. Remove the -G argument

Here is my command:
./ethminer -F http://us2.ethermine.org:4444/

. -X --farm-recheck 200 --cl-local-work 128 --cl-global-work 4096 --cuda-devices 0

I fixed it by running -DCOMPUTE=50 for my 750ti.

Yep, same

You can fix it by adding before each of those two lines from

to
just adding template <> before each line.... well, it compiled. didn't test it

Ok, just downloaded a fresh copy, and tried to do the make and this is the error being produced during the make:

[ 73%] Built target BuildInfo.h
/home/buchanaw/Documents/cpp-ethereum/libethcore/Miner.cpp:7:10: error: specializing member ‘dev::eth::GenericMiner::s_dagLoadMode’ requires ‘template<>’ syntax
 unsigned dev::eth::GenericMiner::s_dagLoadMode = 0;
          ^
/home/buchanaw/Documents/cpp-ethereum/libethcore/Miner.cpp:8:10: error: specializing member ‘dev::eth::GenericMiner::s_dagLoadIndex’ requires ‘template<>’ syntax
 unsigned dev::eth::GenericMiner::s_dagLoadIndex = 0;
          ^
make[2]: *** [libethcore/CMakeFiles/ethcore.dir/Miner.cpp.o] Error 1
make[1]: *** [libethcore/CMakeFiles/ethcore.dir/all] Error 2

Yes all GPUs in the rig do exactly the same thing. You could indeed have each GPU in the rig generate only a part, but that would mean having to haul all these parts over the PCI bus to complete all DAGs on each GPU, which would mostly complicate things. Since it's a one-off job that takes only about 10 seconds, it's much easier to have each GPU generate their own. The CUDA implementation is optimized for speed but I haven't bothered tweaking the OpenCL one. It's fast enough

Curiously does streaming DAG actually use all the GPUs to make the first file then transfer it amongst themselves (like bit torrent) or do they each generate their own? Food for thought if you wanted to do something to improve it.

I pushed a new 1.1.3 binary to the 110 branch. https://github.com/Genoil/cpp-ethereum/tree/110/releases

Added sequential DAG loading in case your rig has problems loading them in parallel. Add -L sequential to the command line to do this.

I've successfully tested this accross two different GPU brands using the -X flag, so I'm quite confident it works with same brand GPUs as well. If you didn't have any problems with previous 1.1 releases, there's no need to add this flag.

Note on releases: I changed my release process a little bit. The master branch will hold the latest minor release, in this case 1.0. Then for a new minor release, 1.1 currently, I will have a branch on github, 110 in this case. All bugfixes and minor changes to that release will go into that branch and the 3rd number of the version (the revision) will then increase). When I feel the new feature is complete and stable enough, I will merge the branch into master and create a new branch for a next feature. So that would be 120 for the next feature.

The current 1.1 new feature is on-GPU DAG generation. When the sequential loading solves some issues that were reported by people with large rigs, I consider it complete enough to move on to the next feature. Will most likely be some stuff I'll get paid for 

Hi Genoil.Your miner is excellent but no GPU overheating protection and if you is not a big deal could be inserted?

No worries, my linux skills are not that great, but I try. These are the steps I have used on my Ubuntu 14.04 rigs. (I think I got from a readme or faq from a mining pool) Feel free to try this it if it works for you great. Again, I'm no expert, just collect stuff that works for me.
From a fresh install:
IIRC you need to reboot, then continue:
On the 110 branch or remove the -b 110 once it goes main.
This is where you then need to edit ~/ccp-ethereum/libethash-cuda/CMakeLists.txt to remove references to 61.
And then compile.
If it builds with no errors, the compiled ethminer will be in the ~/cpp-ethereum/build/ethminer folder. I just copy that to my home folder and run it from there. (cp ethminer ~ && cd ~)

Hope it works for you.

@BitminerN8 Thanks for the input.  I had seen the note from Geniol to you and I had removed those references in that file.  Still had the same issue so I suspected drivers.  Made the mistake of trying to install the .run file then couldn't get X Win to let me log back in with Ubuntu.  Found some references to messed up perms on .XAuthority and a few other things.  Nothing worked. 

So back to the drawing board, I just blew the box away and starting from scratch.  Hopefully I can get some working drivers and then try to install Genoil's miner again...  I swear, sometimes you cannot win for losing... It has been a nonstop battle with this rig... It is a new build and I did not have it running on 1.0.8.

Thanks again for the feedback.

https://github.com/Genoil/cpp-ethereum/tree/110/releases