Cluster slack time?

Does anybody have a Linux cluster burning idle time? Does anybody
have a Linux cluster with unlocated processor faults? Post 'em if
you've got 'em! We can fill every core 99+% for as long as it amuses
you.

Uncle Al has a nasty problem calculating moments of inertia (and stuff
thereform) of expanding radius crystal lattice. Run time/point
increases as (radius)^2.

Output data looks like

radius atoms CHI
49943.000 29915206780110 0.999999997820489774
49956.000 29938573338492 0.999999998020169402
49969.000 29961952191080 0.999999995536169206
49982.000 29985343007260 0.999999998865988518
49995.000 30008746083630 0.999999999066498653

Graphed output looks like

http://www.mazepath.com/uncleal/qzdense.png
A decent graph; 150,000 CPU-hrs
http://www.mazepath.com/uncleal/bzdense.png
50K radius = 21 min/point in an AMD FX-55
That's about 7 GFLOPS sustained, folks.
Needs work.

I supply the Linux static executable or you can compile (*no*
Microsoft compilers - fatal bug) and optimize C++ source code
yourself. Every 1000 CPU-hrs helps. The results support a benchtop
test of General Relativity,

http://www.mazepath.com/uncleal/qz3.pdf

A chemist thinks 1000 atoms are a whole lot of atoms. A physicist
sees data for 30 trillion atoms and asks why we stopped. Calculation
would be 100K-500K radius or more. Even an FX-55 can't handle that...
and an FX-60 will be obsolete by June.

Thank you!

--
Uncle Al
"Best efforts will not substitute for knowledge," W. Edwards Deming