This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: loop optimizations, mark 12[a]



  In message <19980627234611.A17278@dot.cygnus.com>you write:
  > 
  > --r5Pyd7+fXNt84Ff3
  > Content-Type: text/plain; charset=us-ascii
  > 
  > Well, it seems I can't operate diff properly.  In addition to
  > adding in bits I didn't intend, I also left out bits I did.
  > 
  > Here is a replacement patch.
Just tried this one the x86.  

integer performance as measured by spec92 is good (improves by a
percent or two).

FP performance got about 6.5% worse, most of that due to 094.fpppp
running more than 2X slower with your patch.

		Runtime		Specmark
Without:	34.5		266.7
With mark12a	88.4		104.1

Something bad is happening.

jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]