I'm running the tpruvot fork of ccminer, compiled from here: https://github.com/tpruvot/ccminer on ubuntu systems. in general runs fine, for a large number of algorthims. OTOH, for equihash, I'm regularly getting the following style of errors:
[2017-10-29 18:02:47] GPU #0: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:47] GPU #4: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:47] GPU #3: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:47] GPU #2: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:47] GPU #1: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:47] GPU #5: solver: CUDA error 'unspecified launch failure' in func 'solve' line
[2017-10-29 18:02:52] GPU #0: MSI GTX 1070, 353.70 Sol/s
[2017-10-29 18:02:52] GPU #0: unspecified launch failure
[2017-10-29 18:02:52] GPU #0: scanhash_equihash:192 all CUDA-capable devices are busy or unavailable
[2017-10-29 18:02:52] GPU #0: init: CUDA: failed to alloc memory
this is the version that ccminer returns:
*** ccminer 2.2.2 for nVidia GPUs by tpruvot@github ***
Built with the nVidia CUDA Toolkit 8.0 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)
ccminer v2.2.2
libcurl/7.52.1 OpenSSL/1.0.2g zlib/1.2.11 libidn2/0.16 libpsl/0.17.0 (+libidn2/0.16) librtmp/2.3
this is how I'm invoking ccminer:
and also for all other algorithms
I wonder how this can be resolved.