Re: JavaMemoryModel: Implementing atomic object creation on Alphas may be cheap

From: Bill Pugh (pugh@cs.umd.edu)
Date: Thu Oct 14 1999 - 14:17:12 EDT


>We are working on redoing the experiments now. It looks like the
>ratios may not be as good for the full data sets, so the technique
>I've described may not that useful. I'll let you know as soon as we
>get the experiments rerun correctly.
>
> Bill
>

Nevermind...

Here are the totals for everything except Jess (which isn't running
correctly for some reason) with -s100, using the benchmarks that have
been tweaked to have more final fields. The numbers aren't nearly as
good, largely because of the much larger proportion of aaload
instructions. Might still be some possible ideas here, but we are
going to have to think about it a while.

# of getfields: 4,960,317,380
# of getfields of refs: 1,293,146,548
# of non-final same: 554,784,608
# of nonfinal, nonnull: 423,345,408
# of instructions: 38,621,111,238
# of bytes allocated: 821,517,507
# of calls to alloc: 22,772,164
# of aaloads: 688,223,478
# of loads of refs: 1,984,221,553
# of nonfinal loads of refs: 1,243,942,160
# of nonfinal nonnull loads: 984,927,321

-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:19 EDT