Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!newstop!sun!amdcad!dvorak.amd.com!proton!tim
From: tim@proton.amd.com (Tim Olson)
Newsgroups: comp.arch
Subject: Re: register save
Message-ID: <1991Mar12.155459.12187@dvorak.amd.com>
Date: 12 Mar 91 15:54:59 GMT
References: <1991Mar11.192116.1974@dgbt.doc.ca>
Sender: usenet@dvorak.amd.com (Usenet News)
Reply-To: tim@amd.com (Tim Olson)
Organization: Advanced Micro Devices, Austin, TX
Lines: 19

In article <1991Mar11.192116.1974@dgbt.doc.ca> don@dgbt.doc.ca (Donald McLachlan) writes:

| 	Now that all the mechanics are out of the way (the way I see them)
| only one question remains. HOW MUCH DOES THIS ACTUALLY SAVE???
| I interpret this as ... What is the ratio of calls to "leaf functions"
| versus calls to "non-leaf functions"?

From a collection of benchmarks we have, we have run dynamic leaf-call
statistics, and see that leaf routines account for 40% - 60% of all
function calls.  The benchmark with the fewest leaf-routine calls was
"compress", with less than 1%; the most leaf-routine calls were in a
kalman filter benchmark (96%).  Because about half of all calls are to
leaf functions, leaf-procedure optimizations are important.


--
	-- Tim Olson
	Advanced Micro Devices
	(tim@amd.com)