SSA simplify passes

Matthew Fluet fluet@CS.Cornell.EDU
Mon, 7 Jan 2002 20:18:21 -0500 (EST)


> It's a shame to slow down nucleic, which is now the only benchmark
> where we are significantly worse than SML/NJ.  Oh well.

I'll bet it's the same thing with too much flattening of floating point
tuples.