Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!pasteur!ames!mailrus!iuvax!pur-ee!a.cs.uiuc.edu!uxc.cso.uiuc.edu!urbsdc!aglew From: aglew@urbsdc.Urbana.Gould.COM Newsgroups: comp.arch Subject: Re: Vectorising conditional code. Message-ID: <28200176@urbsdc> Date: 12 Jul 88 15:01:00 GMT References: <893@garth.UUCP> Lines: 13 Nf-ID: #R:garth.UUCP:893:urbsdc:28200176:000:632 Nf-From: urbsdc.Urbana.Gould.COM!aglew Jul 12 10:01:00 1988 ..> weemba playing with Mandelbrot on a Cray-1: ..> .5 Mflops using complex arithmetic, ..> 5 Mflops using real ..> 50 Mflops using assembly This causes me to wonder what the state of the art is in vectorizing complex arithmetic. Of course, the transformations from complex to real, and the appropriate vector optimizations with short circuits, are all well known, but they do no good if they aren't fully implemented. How good are available compilers at vectorizing complex arithmetic? Do people short shrift complex because most "important" programs have already been translated to use reals? Is this a self-fulfilling prophecy?