Subject: some slowness...
From: Craig Chambers (chambers@cs.washington.edu)
Date: Fri Mar 23 2001 - 11:11:47 PST
In running Whirlwind on all the Java benchmarks, I noticed the following.
Here is the time to compile a single method of the toba benchmark:
-- compiling toba_fOpcode_sgenerate_aLjava_fio_fPrintStream_aLtoba_fInstr_rV
## Phase times:
Method compilation: 23095.336 secs
Building CFGs: 1.995 secs
Verifying CFGs: 457.058 secs
Topologically numbering graphs: 47.578 secs
Building DFGs: 225.011 secs
Verifying DFGs: 16.579 secs
Typechecking: 30.041 secs
Verifying CDFGs: 219.459 secs
Composition of Unreachable Code Elimination, Constant
Propagation:22020.539 secs
Unused Value Elimination: 10.805 secs
Phi Elimination: 6.377 secs
Lowering: 16.368 secs
Verifying Lowereds: 0.000 secs
Generating code: 22.088 secs
Compressing: 0.000 secs
Building scopes: 7.258 secs
Resolving names: 2.494 secs
Computing representations: 11.001 secs
Topologically numbering graphs: 0.055 secs
Topologically numbering graphs: 0.169 secs
This 23Ksec compilation accounted for most of the 25Ksec compilation of the
entire benchmark.
-- Craig
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2b25 : Fri Mar 23 2001 - 11:12:04 PST