Show Posts
I decided to start mining some alt coins and thought I'd fire up a device I had lying around. I'm using a Tesla S870 system, but only 1 of the 2 connections for now (2 cards). I'm having an issue finding a stable mining configuration. Essentially, the system appears to be busy and accepting work, but not coins are ever mined. I've done a lot of searching and trail & error with the -l command-line option. As I understand it, the -l option should be the multiprocessors x CUDA cores, correct? I.e. (from deviceQuery):
Code:
(16) Multiprocessors x ( 8) CUDA Cores/MP: 128 CUDA Cores
Therefore -l L16x8 ?
If I pump up the -l setting (i.e. L128x64, but can be much lower), it will often say it's getting 300+kh/s, which I believe is completely off. I noticed this in another post in this thread, and it seemed to be a one-off. I know that this is not rockstar hardware, but I would like to use it for some light mining. My questions are:
(1) What's the _right_ way to determine the -l settings? I have tried many options as well as 'auto' with -D and even -P (I am a web guy, after all
), which often leads to L0x0 and crashes.(2) Is there anything I can do to help with support for this hardware?
Here's my configuration:
OS:
Code:
$ uname -a
Linux hypercoil 3.5.0-23-generic #35~precise1-Ubuntu SMP Fri Jan 25 17:13:26 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
NVIDIA Driver:
Code:
$ cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module 304.54 Sat Sep 29 00:05:49 PDT 2012
GCC version: gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu5)
NVCC/Cuda Tools:
Code:
$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2012 NVIDIA Corporation
Built on Fri_Sep_21_17:28:58_PDT_2012
Cuda compilation tools, release 5.0, V0.2.1221
CudaMiner:
Code:
$ ./cudaminer
*** CudaMiner for nVidia GPUs by Christian Buchner ***
This is version 2013-12-10 (beta)
based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
Cuda additions Copyright 2013 Christian Buchner
My donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm
Kind regards,
DW
