record deep dive

[Ulf Wiger (AL/EAB)]

I took R10B-2 for a test drive using my ring benchmark
on my new Blade1500 Sun workstation (1 GHz, 1 GB RAM)

I was able to push the ring benchmark to 630,000 processes
before I got into lots of iowait. Beam used about 800 MB
of RAM.

Timings in microseconds:

N procs  spawn   send
630000   6.47106 1.96776

I also tried a ring benchmark with increasing payload,
using the hybrid heap, and it performed excellently,
with message passing cost of abot 1 us regardless of 
message size. Apparently, the compiler is capable of
realizing that the output of lists:duplicate/2 should
be put on the shared heap, since it is later passed 
as an argument in spawn_link()? Impressive.

/Uffe