Trouble with compiling on Linux
Lignum opened this issue · comments
Not sure if it's a bug or I'm being silly...
:test
me.apemanzilla.jclminer.tests.TestCLHashing > testHashing_digest FAILED
java.lang.ExceptionInInitializerError
Caused by: com.nativelibs4java.opencl.CLException
java.lang.NullPointerException
me.apemanzilla.jclminer.tests.TestCLConstants > testConstant_K FAILED
java.lang.NoClassDefFoundError
java.lang.NullPointerException
me.apemanzilla.jclminer.tests.TestCLMacros > testMacro_PAD FAILED
java.lang.NoClassDefFoundError
java.lang.NullPointerException
me.apemanzilla.jclminer.tests.TestCLMacros > testMacro_RR FAILED
java.lang.NoClassDefFoundError
java.lang.NullPointerException
me.apemanzilla.jclminer.tests.TestCLMining > testMining_hashToLong FAILED
java.lang.NoClassDefFoundError
java.lang.NullPointerException
Dunno if this is intended, but it doesn't build the CLI, which is a problem.
Unless you're supposed to build it manually, in which case I get this:
Circular dependency between the following tasks:
:classes
\--- :compileJava
\--- :jar
\--- :classes (*)
(*) - details omitted (listed previously)
You can't build the CLI subproject separately as it requires the core code.
Can you attach the full stacktrace of the testHashing_digest test? You can find it by opening build/reports/tests/index.html.
Can you post the stdout logs for testHashing_digest? They are also available from the test reports. (This is why I love Gradle for testing 👍 )
Ah, it's this bullcrap again:
an 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : atiocl64
Jan 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : amdocl64
Jan 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : atiocl32
Jan 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : atiocl
Jan 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : amdocl32
Jan 21, 2016 7:19:59 PM org.bridj.BridJ log
INFO: Library not found : amdocl
I had this error with JCLMiner... eventually it fixed itself for some reason.
Yeah, Linux with the proprietary NVidia drivers is... annoying, at best. I'm assuming you still have the latest drivers from apt installed - 352, right?
Yep, this is a fresh Xubuntu install running nvidia-352. Maybe installing the CUDA SDK will help?
No, that shouldn't be it. I had some success with version 346, so try that out first.
Alternatively, you can try building it with ./gradlew build -x test which will skip the testing entirely, but if the tests fail chances are the miner will have the same problem.
Too late, I'm almost done installing the CUDA SDK... besides, it worked on nvidia-352 last time. But I'll give it a shot if this doesn't work.
Alright. You can also try installing some of the OpenCL packages from apt. Just run apt-cache search opencl and see what packages are available.
Question: If i'm on windows I shouldn't have to install any special OpenCL packages right? Seems weird that last time I got that BSOD... I might have not read instructions right
All you need are your graphics drivers...
Odd.
@JustynC7 You mentioned yesterday that other programs were also causing BSODs, leaving me to believe that it's a problem with your system rather than my code. If you find any evidence otherwise you can open an issue.
I'm going to remove the bug label seeing as this is a driver issue rather than a problem with the code itself.
You don't understand... The other program I knew that I didn't include all the required arguments and the code was touchy in the first place. That is unrelated.
Installing the CUDA SDK fixed it.
Glad to hear it. Have fun mining.
Meh, I'm getting about 20MH/s on Linux... I wouldn't necessarily call that fun, but hey, it works :P
You'll need to set the -w option similar to -d in JCLMiner. Copy the signature from -l, and other than that the format is the same: -w <sig>:<worksize>
Hm, I get ~205MH/s with worksize 999999... my signature is also negative this time?
The signature can be any integer. For the worksize, try 2^21 or so - that gave the best performance on my GTX 760
I went as far as 8097152 and it seems to max out at around 240MH/s...
Alright, on Windows I get 160MH/s?..
Forgetting -a?
And yes, I noticed as well that Linux gets significantly higher hashrates than Windows. Only for NVidia though, on AMD it was exactly the same.
Yep, that was it. I thought it wasn't necessary anymore for some odd reason :P. I get an average of 320 MH/s, though isn't this one supposed to get entire GH/s or is that still in development?