[Release] VGA-Mon - Simple GPU management for mining rigs.

[geo.koder]

VGA-Mon

Introduction

VGA-Mon is an absurdly simple node GPU monitoring app specifically tailored towards monitoring mining rigs. See the GitHub for the full markdown README.

I wrote this last night after getting tired of manually restarting my rigs after random OC quirks caused it to fail. This application records when GPU utilization drops below a certain threshold and will restart your mining application, and then if the GPUs are STILL not hashing, will reboot the entire rig. Currently only tested on Windows with nVidia cards but Linux SHOULD work in theory if nvidia-smi has identical output, though the config would need to be slightly modified.

I'll be slowly updating this. Have plenty on the todo, but currently it functions as-is. If AMD has a command line tool that outputs GPU util percentages please drop me a line so I can incorporate it.

With proper configuration (and once I find an AMD CLI application for diagnostics), this application should be entirely platform agnostic, pool agnostic, and miner agnostic.

Donations
ETH: 0x00972cd6a2c6786afbcc24ca592b8c86f33f747a
BTC: 1n4ruYy5QWbTDBbPEyBRWwj1Ni4U4Sz5P

[Alternative]

Looks awesome ! I will talk about that to my miner friends ! However I find that your default values are a bit low, especially considering the rig reboot time. Rebooting every six minutes is, except in case you have a very unstable miner, not necessary. The miner reboot put by default at two minutes is maybe a bit low, don't you think ? With the DAG loading time, you will end up losing a lot on profitability I think.

[geo.koder]

Looks awesome ! I will talk about that to my miner friends ! However I find that your default values are a bit low, especially considering the rig reboot time. Rebooting every six minutes is, except in case you have a very unstable miner, not necessary. The miner reboot put by default at two minutes is maybe a bit low, don't you think ? With the DAG loading time, you will end up losing a lot on profitability I think.

It only reboots the application/machine if it detects a crash (GPU utilization drops below threshold for too long). Ethminer takes less than a minute to start up (including DAG loading) on my Celeron, so I actually thought I might have set my values a bit HIGH because if it reboots the app it has another four minutes to start up before the entire machine gets rebooted.  Feel free to adjust as needed!