Wide Finder 2

[Patrick Logan]

Tim Bray of Sun has created a wiki for his "Wide Finder 2" programming
language shoot-out of sorts.

See his blog announcement...
http://www.tbray.org/ongoing/When/200x/2008/05/01/Wide-Finder-2

[Justin Bailey]

That was an interesting series of posts when he did last. Seems he doesn't have a real idea what to benchmark this time, though. Someone made a good point that coming up with a CPU-bound rather than IO-bound benchmark would be hard when analyzing 43 GB of logs.

Any ideas?

Justin

[M. Edward (Ed) Borasky]

Justin Bailey wrote:

I've been in the "benchmarketing" racket for a number of years. I'll
have to go look at Bray's web page again, but IIRC it suffered from a
number of problems. Still, for any given benchmark, the "optimal"
configuration is one that balances CPU and I/O utilization on the
average with few periods of saturation for either one.

In my not so humble opinion, a far more valuable web site is Dr. Neil
Gunther's blog/website: http://perfdynamics.blogspot.com and
http://perfdynamics.com. Or Google for Guerilla Capacity Planning.

Actually, yes, Google for it, because when you use Google, you are
essentially exploiting a massively parallel process called "map-
reduce" that just works. It always warms my heart to see someone get
rich using a singular value decomposition. ;)