Performance Improvement: Top down or bottom up ?

[MichaelW]

Maybe something like this:

Code Select Expand


    .macro hintbranch
      .byte 0x3e
    .endm

    .macro hintnobranch
      .byte 0x2e
    .endm

    FORP4ONLY = 1

    .ifdef FORP4ONLY
      .print "yes"
      hintbranch
      jz target
    .else
      .print "no"
      jz target
    .endif
  target:


Code Select Expand


  11                  .macro hintbranch
  12                    .byte 0x3e
  13                  .endm
  14
  15                  .macro hintnobranch
  16                    .byte 0x2e
  17                  .endm
  18
  19                  FORP4ONLY = 1
  20
  21                  .ifdef FORP4ONLY
  22                    .print "yes"
  23                    hintbranch
  23 0000 3E        > .byte 0x3e
  24 0001 7400            jz target
  25                  .else
  26                    .print "no"
  27                    jz target
  28                  .endif
  29                target:


[James Ladd]

MichaelW,

If there is ever a poll for who is super helpfull I'll be sure and enter your name !   :bg

Thanks.

[James Ladd]

Ok, I have this im my main include now.

Code Select Expand


.ifdef NOP4
.print "Not Assembling for a Pentium 4 so no branch hints are used."
.macro hintbranch
.endm

.macro hintnobranch
.endm
.else
.macro hintbranch
    .byte 0x3e
.endm

.macro hintnobranch
    .byte 0x2e
.endm
.endif

.ifdef WIN
.include "win.inc"
.else
.include "linux.inc"
.endif


So I will put branch hints in the code, but they wont be output unless compiling for
a P4. By default my makefile assumes compilation for a P4 unless you say:

Code Select Expand


make NOP4=1