Who has a bored cluster? REV 2.0

[UncleAl]

You can try it but I doubt it is that simple if you want speed. My UK programmer worked almost four months transforming serial BigCHI into parallel CHIpir. Each value report is devastating to process throughput. Subtleties. He gave up, then had a dream with the solution. Source code length about doubles. Speed is everything.

For a cluster run you need the big gun primed to fire, with documentation. Tell me how many processors can be grabbed and I'll mount a ZIP including a script file for execution.

[Morally.Corrupted]

Originally Posted by UncleAl

You can try it but I doubt it is that simple if you want speed. My UK programmer worked almost four months transforming serial BigCHI into parallel CHIpir. Each value report is devastating to process throughput. Subtleties. He gave up, then had a dream with the solution. Source code length about doubles. Speed is everything.

For a cluster run you need the big gun primed to fire, with documentation. Tell me how many processors can be grabbed and I'll mount a ZIP including a script file for execution.

Well Currently, I have 4 - Alexander has graciously said he could commit 2-3 to the project provided I can run a suitable OS for PPC, etc.. but that's not an issue as BCCD has support in Beta 2.2.2 for PPC's - I can also try to enlist a few more so realistically within about a weeks time could have a 10 node cluster up and runnning.

[UncleAl]

Running serial BigCHI through MPI probably won't do it. Parallel-executing CHIpir source code is twice as large as that for BigCHI. Speed is important. We do everything to make the executable fast. Tell me how many CPUs are in the cluster and I'll mount a ZIP archive with a script file, the source code, and documentation. Parallel source code should be compiled for the cluster in which it will reside.

Remember - compile long_double_precision - NO Microsoft compilers!

[alexander]

lol uncle, i can't even conceive a cluster of windows boxes.... it sounds so funny and wasteful, it would be like hiring an 85 metric ton John Deere excavator, to plant a year old oak tree in your back yard.

[alexander]

not to say that a load of windows machines can't make it into a cluster, but only in a distributed computing project... a dedicated cluster, that's a totally different beast, on it's own...

[alexander]

ok, i have 3 machines which i can donate for a little while to the project:

i have
800mhz PPC box
a Netra T1
my AMD64 3400+ box (desktop, but i would not mind, for a little while )
i may have a 1.0 ghz p3 sitting upstairs (my dad's PC, and i know he is not using that)

Inte3rc3pt (aka morally corrupted) has
800 Mhz box
another slow box
P4HT
his crazy AMD box

I used to have a quad 933 p3 box (4 penguins was real nice)... but that's been given as a gift to someone :|

let me see what i have kicking around the garage, i may have another machine or two that may need to be put together

I may have another friend who may be able to donate his machine to the project, for a while anyways

i also may have another friend who may be able to just give me his old box, but i have to call him

ooh i wonder if squelch has anything....

I'll know for sure as it gets closer to monday...

[alexander]

Uncle, you guys ever think of buying time on the amazon cluster?

[alexander]

or applying for the UCAL or MIT cluster usage?

[UncleAl]

4-CPU cluster timing run set up for 12 points, 3 points/CPU. The first command line should write results within a fractional second. You have the correct long_double_precision CHI values from the serial runs. It verifies operation and sets the first time stamp. The second line is the timing run. If the first line is OK it is only a matter of waiting. The serial timing runs gave you the correct long_double_precision values for CHI for 30,000, 30,001, and 30,002 angstrom radii. If you compiled double_precision (Microsoft crap compilers) obtained CHI values will be different toward the right side.

http://www.mazepath.com/uncleal/uncleal3.zip

chipirbz.cpp - C++ source file, CHIpir loaded for full benzil. It is in UNIX format - linefeeds only. It must be compiled for the cluster OS.
readme.pir - Instructions and documentation.
timer.pir - Timing script file for 4 CPUs in cluster. 12 points calculated.

mpirun -np 5 ./chipirbz 3.35 0.15 5.0 1 1 >> time1.pir
mpirun -np 5 ./chipirbz 29999.0 1.0 30011.0 1 1 >> time2.pir

If it bombs for "-np 5" try "-np 4" (number of CPUs used plus the master)

buying time on the amazon cluster applying for the UCAL or MIT cluster usage

The short form: Resources won't be allocated unless justified by, for instance, a scholarly publication. No journal will publish except for experimental results. Professional management is about process not product.

I don't criticize the System, I turn my back on it. This comes out of my own wallet, with volunteers as available. It's a hobby. If it overturns all of physics on a tiny technicality... that's as much fun as anybody can have wthout being naked. If it validates physics as is, it was also fun.

"It doesn't matter how beautiful your theory is, it doesn't matter how smart you are. If it doesn't agree with experiment, it's wrong," Richard Feynman. Sombody should look; me, for instance. You if it pleases you. If it stops being interesting, stop. No harm, no foul.

[UncleAl]

Alexander,

800mhz PPC box
a Netra T1
my AMD64 3400+ box (desktop, but i would not mind, for a little while )
i may have a 1.0 ghz p3 sitting upstairs (my dad's PC, and i know he is not using that)

Too slow as single CPUs! Boot Linux (Knoppix LIVE! DvD is wonderful), try the timing run.

http://www.mazepath.com/uncleal/uncleal2.zip

An AMD FX-55 does it (difference of output file timestamps) in 36 min 4 sec. You are looking at hours! Calculating pi to 2.1 million decimal places is 5.95 sec in the FX-55. The next points in the CHI crunch are 1.7 hrs (6100 secs) each and then longer - at least 1000 points. It is an ugly probem.

http://www.mazepath.com/uncleal/uncleal2.zip
timing run

A "little time" won't do it. I ran the FX-55 for 80 days at 99.9% CPU while the rest of the house ran off a 200 MHz Pentium II. It needs a cluster now.

Inte3rc3pt (aka morally corrupted) has
800 Mhz box
another slow box
P4HT
his crazy AMD box

We went looking for sour CPUs in a 168-CPU Pentium cluster in Europe over six weekends (found 3 unreliable cores). It outperformed a 16-cluster of Opteron-848s... but not by much. Intel Pentiums are crap CPUs for doing heavy maths.

let me see what i have kicking around the garage, i may have another machine or two that may need to be put together

I would not be looking for large caliber capability if it were an easy problem. The number of atoms crunched increases as (radius)^3. We've got compute time down to (radius)^2. The problem starts at 107 trillion atoms and goes to 1 quadrillion atoms, 10^15. It's got to be a cluster or a single CPU well beyond an AMD FX-55 - a 3 GHz FX-74 or a 2.5 GHz Phenom X4 9850 would be interesting. A single Pentium box is unsuitable - but you know that! Wintel is a bad joke, Vista is its punchline.

What will Intel 45 nm architecture, hafnium oxide, quad-core CPUs accomplish? They'll boot Vista in 4 GB of RAM and leave you a MB of memory free in which to run DOS. Slowly.