i386 performance problem due to regmove.c change

Jeffrey A Law law@cygnus.com
Sun Jun 28 00:51:00 GMT 1998


  In message < 199806241723.NAA13994@jwlab.FEITH.COM >you write:
  > I seem to be seeing around a 3 percent performance hit when
  > running the crafty chess benchmark and the Brl-Cad raytracing
  > benchmarks on an Intel Pentium II with:
  > 
  > Thu Jun 11 14:50:02 1998  Michael Meissner  <meissner@cygnus.com>
  > 
  > 	* regmove.c (regmove_optimize): If we can't replace the
  > 	destination in an insn that sets the source, generate an explicit
  > 	move of the source to the destination.
  > 	(copy_src_to_dest): New function.
  > 
  > installed.  If I remove this patch then life gets better (not to
  > mention faster :-).
I would be very curious to know if this performance regression went
away after my changes to recompute register set/reference counts.

jeff



More information about the Gcc-patches mailing list