Matthew Danish wrote: > > On Wed, Sep 25, 2002 at 10:28:58PM +0200, Edi Weitz wrote: > > > > Mario Mommer <[EMAIL PROTECTED]> writes: > > > > > I know that gcc uses lots of "peephole" optimizations. That is,
> There are two mechanisms for introducing changes into this > vector after the VOPs have been called: backpatches and choosers. > Choosers have the ability to change the size of a portion > of the vector, possibly shrinking it if necessary. Sometimes a larger sequence is faster, faster RISC vs slower CISC, sequence, besides the obvious loop unrolling, right? Individual register moves -vs- 68k Link/UnLink ? (I'm stretching my memory here)
