search

djm34 / ccminer-msvc2015

ccminer updated to the latest version of visual studio (2015)
GNU General Public License v3.0
33 stars 25 forks source link

invalid device function #15

Closed davezes closed 6 years ago

davezes commented 7 years ago

Ubuntu 16.04 Server Cuda 8.0.61 NVIDIA-SMI 375.66 -- Tesla K80

Everything installed like a charm

Running ./ccminer

yields

[2017-08-11 23:06:20] GPU #0: Tesla K80, 1099.78 MH/s [2017-08-11 23:06:20] GPU #0: invalid device function ...

Thank you!

djm34 commented 7 years ago

you probably need to adjust the intensity...

davezes commented 7 years ago

Hey djm34. I've tried varying the intensity without success. I'll spend some time trying some of the other parameters and check in with results. Thank you for this project and your help!

davezes commented 7 years ago

I got it mining, albeit at a very slow hashrate, by turning on the compute and multiplexer architecture to the earlier 35 (by editing the Makefile). Is that a reasonable thing to do? I'll continue to tune the ccminer parameters, as you suggest, to see if I can up the hashrate. Thanks again, djm34, much appreciated!

matu0 commented 6 years ago

I've got the same problem. What did you change exactly in your Makefile?

djm34 commented 6 years ago

you need to chose compute_XX,sm_XX which is relevant to your cards ie: compute_61,sm_61 (or compute_52,sm_52) for pascal (available with cuda 8+) compute_52,sm_52 (or compute_50,sm_50) for Maxwell (cuda 6.5+) compute_35,sm_35 for 780/780ti/titan (fermi (?)) compute_30,sm_30 for older cards

matu0 commented 6 years ago

Thanks for the reply,

how do I choose the compute?