Performace question

From: Zaher S Andraus (zandrawi_at_eecs.umich.edu)
Date: Fri Apr 09 2004 - 14:51:39 MDT


Folks,

I am trying to prove equivalence of 2 implementation
of 2 FIFOs, via VIS. The design is attached.

I am encountering a very weird phenomenon. I vary
the number of entries in the FIFOs (via changing the
LAST parameter in the design). I get a non regular
"jump" of performance for a certain value.

The table of run time is:

LAST run-time (seconds)
1 0.17
2 0.26
3 62.17
4 0.73
5 2.83
6 17.93
7 122.60
8 304.74

When LAST=3 we get a very slow invariant checking. The
rest of the table makes perfect sense. I repeated the
run sessions and I got the same result!!

Clearly I can't answer this question because it depends
on the internals of VIS. Can anyone help in that ?

-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x
Zaher S. Andraus
Advanced Computer Architecture Laboratory in EECS
University of Michigan - Ann Arbor
http://www.eecs.umich.edu/~zandrawi/academic.htm
-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x-x



This archive was generated by hypermail 2.1.7 : Fri Apr 09 2004 - 14:52:52 MDT