SSMT SIMULATOR

Authors: Blair Fort (fort@eecg.toronto.edu)
	 Livio Soares (livio@eecg.toronto.edu)

Note that a full pdf report is available in the tarball.


   Here is  the tarball with our  project. In it you  will find the  PDF for our
final report. Additionally, you will find one Makefile and and 4 .c or .h files,
which you should simply copy over a fresh tree of SimpleScalar (version 3.0d).

  Unfortunately,  it  seems   we  have  a  bug  that   manifests  itself  almost
randomly.  This  bug did  not  prevent us  from  executing  many benchmarks  and
configurations, however,  it does  happen from  time to time.  We are  trying to
track it down. The two possible symptons are:

1) A non-speculative fault (address alignment).

2) Infinite loop  in the simulated code  (the benchmark PC  circulates between 4
   addresses).

  The only  way we know how to  guarantee avoiding this issue,  is using perfect
branch predictions ('-bpred perfect' in the command line). Running in this mode,
we have never encountered any problems.