[Prev][Next][Index][Thread]
Can someone verify E/D 1.99q11 speedups?
Many people have reported that E/D 1.99q10 and 1.99q11 are much faster
than previous Executors. There are two big factors which should
indeed have sped up Executor, and they probably explain everything.
However, since I had to change our timer code around for those
releases I am a little nervous that part of the speedup could be a
newly introduced timer bug. Since people are reporting that *all*
programs seem faster, that's unlikely, but ARDI would look really bad
if a bug caused Executor's benchmark numbers to appear to have been
artificially inflated.
Could someone who noticed a big speedup please time Speedometer with a
stopwatch under 1.99q9 and 1.99q11 and verify that it really did get
faster, independent of what Executor thinks?
Thanks!
-Mat
Follow-Ups: