random coalescion

Stephen Weeks sweeks@research.nj.nec.com
Fri, 5 Mar 1999 21:16:34 -0500


Here are the running times comparing the benchmarks with no coalescion
and with random coalescion.  Things look pretty good.  Only life and
logic slowed down.

		none	random
barnes-hut	13.6	12.5
count-graphs	12.7	10.4
fft		39.1	34.9
knuth-bendix	26.1	18.8
lexgen		48.9	37.2
life		56.9	58.3
logic		49.7	56.0
mandelbrot	21.1	21.1
matrix-multiply	13.1	13.0
mlyacc		23.0	23.2
nucleic		22.6	20.9
ratio-regions	25.0	20.2
simple		20.6	18.7
tsp		29.9	29.7
vliw		21.5	21.3
zern		46.2	44.6