Skip to content

Latest commit

 

History

History
36 lines (23 loc) · 1.61 KB

README.md

File metadata and controls

36 lines (23 loc) · 1.61 KB
Raw

cgminer2rrd

This repo is a set of helper scripts for pulling data out of the cgminer (or, preferably, sgminer) API and writing it to RRD files or a CSV. It's been developed on Linux, and may or may not work on other platforms.

Requirements

  • python 2.7
  • rrdtool-python
  • Optionally, R and ggplot2
  • sgminer or cgminer - sgminer is preferable as it has more precision on the MHS API endpoint
  • The API enabled for the above (add "api-listen" : true, to your cgminer config file)

Usage

There are two primary usages of the scripts in this repo:

  1. Recording long-term performance and tuning data to RRD files for later graphing
  2. Actively fiddling with overclock settings while writing data to a CSV for later processing with R

RRD Graphing

This is the, ironically, the less compelling usage of cgminer2rrd at the moment. It basically dumps information from the cgminer API to RRD files, ready for later graphing.

To run, use ./poll.py, which will output very little, but poll the cgminer API every 5 seconds

To view data, use ./graph.py, which will render the more useful generated RRD files to graphs, ready for viewing.

Overclock Tuning

To generate a heatmap of your core/mem clock hashrate output:

  1. Be aware of your card's stable range of core and memory clock settings
  2. Customize the top of clocks.py to specify these settings
  3. Run ./clocks.py and wait several days
  4. If ./clocks.py crashes or you have to stop it, just restart it when you can - it won't retry settings that it has already tried
  5. Once ./clocks.py has finished, generate a graph with R --no-save < clocks.R