Posts by Stephen Uitti

1) Message boards : News : Use at your own risk (Message 637)
Posted 6 Jul 2018 by Stephen Uitti
I get it - use at your own risk. But i run test programs for all sorts of projects. I'd like to see this project succeed.
I run Linux machines with GPUs. Around the start of June, i started seeing other Linux boxes with new Collatz credit, so i thought i'd give it a shot. I turned on just the GPUs (my normal), and got no units at all. I also turned on CPU units, and only got GPU units, and on 12 June 2018, i got credit. (!) Later, i started getting CPU units also. They said they'd run long, so i let them run a couple days. My estimates were also way past the deadline (time into the unit run / fraction done = total time). On 24 June 2018, i aborted two units in progress, and whatever was in the queue. I turned off CPU units. I continued getting GPU units. On 5 July 2018, i got zero new credit, with 52 units in error (Validate Error). That's about 2 days (52 hours is pretty close) of GPU time. I had gotten lower credit on July 4th, but not zero. Curiously, though these units report as taking about an hour each, WUProp reported only 14 hours for the most recent 24 hours. I don't know which to believe. The system is run 24/7. There's nothing else on the GPU. And yet, i'd gotten about half credit on July 4th, which may or may not be about right, given that i don't know exactly when boincstats samples the data. These units may be running full time but not validating - as reported.

An invalid unit: task 4135440 work unit 3727239 computer 146594

GTX 950 with AMD Phenom(tm) II X4 965 Processor and 16 GB RAM on Linux Mint 13 (out of support for a year - but i tracked down which disk drive was bad, and have taken it from service, so an upgrade is once again in my near future).


©2018 Jon Sonntag; All rights reserved