While running jvm98 on 141b03 vs. 140b92 (sparc solaris), we've found a 3% regression. It appears that the _201_compress subtest has gotten noticably slower (~33%) and probably accounts for pulling the overall test score down. This should be investigated further.

For details:
http://cooljava.central/javaperf/alacrity/1.4.1b03/