home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:10506 comp.lang.misc:3542
- Newsgroups: comp.arch,comp.lang.misc
- Path: sparky!uunet!spool.mu.edu!news.nd.edu!mentor.cc.purdue.edu!hrubin
- From: hrubin@mentor.cc.purdue.edu (Herman Rubin)
- Subject: gotos in hardware
- Message-ID: <BxCLAu.5AF@mentor.cc.purdue.edu>
- Organization: Purdue University Statistics Department
- References: <1992Oct29.153514.22927@yrloc.ipsa.reuter.COM> <BwxsF6.3DF@mentor.cc.purdue.edu> <1992Nov6.111616.109@turtle.fisher.com>
- Date: Sat, 7 Nov 1992 13:25:41 GMT
- Lines: 26
-
- In article <1992Nov6.111616.109@turtle.fisher.com> ferris@turtle.fisher.com writes:
- >Since this discussion is posted in comp.architecture shouldn't we be
- >discussing the possibility of a goto-less object code?
-
- >I believe that it is possible at the micro-code level but not at the
- >machine level. However, I can envision something like:
-
- .................................
-
- Now I raise the question as to why anyone would want to do this? The only
- answer I can think of is that it would eliminate the possibilities of coding
- for speed, and lighten the load on compilers.
-
- Since computing relative to memory access was speeded up about 30 years ago,
- there have been various amounts of look-ahead and scheduling. A goto, with
- no intervening possible exits, should be handled in such a way that that
- the hardware starts getting the code at the new location early, so that
- there are no unnecessary delays. It could even issue instructions from
- that "new" code in the same way it does from current code. The "structured"
- approach, where unconditional transfers are not the rule, prevents this
- type of optimization.
- --
- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
- Phone: (317)494-6054
- hrubin@snap.stat.purdue.edu (Internet, bitnet)
- {purdue,pur-ee}!snap.stat!hrubin(UUCP)
-