Welcome to MilkyWay@home

Posts by Rapalyea

1) Message boards : Number crunching : Perhaps its all a Sham (Message 63971)
Posted 6 Oct 2015 by Rapalyea
Post:
I suspect both BOINC and NVIDIA consider out personal machines perhaps more trouble then we are worth. Why? I considered myself an exemplary neophyte, having spent more then $1,000 on my special build rig. It contributed tens of millions of stones at low power to Einstein@home which was its objective. Recently it was returning about 140,000 stones per day at about 350 watts. Then several things happened.

First, I swapped out a GTX 660 and installed a GTX 750. The entire rig collapsed.

Second, in my attempts to re-engineer the chaos the rig created hundreds of error work units.

The Rig: ASROK seven pcie game board and case. Platinum power supply. Three functioning GPUs [2 x gtx 650 + gtx 660]. The rig would actually run four
GPUs but became unstable. No error work units but stopped production. So I retreated to three GPUs.

Third, these error units seem to have resulted in my machine being denied further Einstein work units. Certainly understandable. Sort of.

Fourth, with much effort I managed to return the rig to its previous operable condition.

Fifth, I tested the rig on MilkyWay with success.

Sixth, I petitioned Einstein@home for permission to receive work units with no success.

This stinks of some sort Post Doc (if even that) having been relegated to gin-up ways to use nerds-on-the-net for specious grant money and little else. Otherwise we would have one simple thing. Specifically, dummy units to use to test our rigs.

It is apparently not even worth that effort, or the effort to simply to reject the eror units and permit a reset. I think the jig is up guys.
____________
2) Message boards : MilkyWay@home Science : GPU v non-gpu tasks (Message 63965)
Posted 4 Oct 2015 by Rapalyea
Post:
My rig runs three separate gpu units [ 2 x gtx 550 + 1 gtx 660] and is built for production efficiency rather then simple production. This means I run the smallest CPU that will keep them fed. Accordingly I would rather not run CPU work units.

Is there a list?




©2021 Astroinformatics Group