Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!sun-barr!decwrl!shelby!neon!carcoar!wilson
From: wilson@carcoar.Stanford.EDU (Paul Wilson)
Newsgroups: comp.lang.lisp
Subject: float boxing/unboxing vs. usage statistics?
Summary: any stats on how much a good compiler can reduce overhead?
Keywords: floating point, boxing, compilers, optimization, numerical programs
Message-ID: <1990Mar12.054909.5054@Neon.Stanford.EDU>
Date: 12 Mar 90 05:49:09 GMT
Sender: root@Neon.Stanford.EDU (System PRIVILEGED Account)
Reply-To: wilson@carcoar.Stanford.EDU (Paul Wilson)
Organization: U. of Illinois at Chicago (UIC, *not* UofC or UIUC)
Lines: 26

I'm looking for information on how good a Lisp compiler can be at keeping
floating point values in registers, so that they don't have to be "boxed"
into a heap format that keeps the gc happy.

Perhaps some of the MacLisp or S-1 Lisp people have some idea about this.
Roughly, the question is this:  in floating-point intensive code,
how frequent are loads and stores of heap floats relative to adds,
multiplies, etc.?

I figure I can squash a 32-bit IEEE float into a tagged word in about
7 cycles, and un-squash one into a standard float in about 5 cycles.
I want to know how often I'll have to do this, compared to the actual
floating-point work at hand.

Thanks prematurely,

    Paul

Paul R. Wilson                         
Software Systems Laboratory               lab ph.: (312) 996-9216
U. of Illin. at C. EECS Dept. (M/C 154)   wilson@bert.eecs.uic.edu
Box 4348   Chicago,IL 60680 
Paul R. Wilson                         
Software Systems Laboratory               lab ph.: (312) 996-9216
U. of Illin. at C. EECS Dept. (M/C 154)   wilson@bert.eecs.uic.edu
Box 4348   Chicago,IL 60680