Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.os.msdos.programmer > #327
| From | Ross Ridge <rridge@nospicedham.csclub.uwaterloo.ca> |
|---|---|
| Subject | Re: Opcode frequency process/tools? |
| Message-ID | <j89oi2$uqb$1@rumours.uwaterloo.ca> (permalink) |
| Date | 2011-10-26 15:54 -0400 |
| Organization | University of Waterloo |
| References | <f5d9291d-0bb2-4088-ab58-e8b18dd8cfb9@t8g2000yql.googlegroups.com> <ug2mn8-pe3.ln1@ntp6.tmsw.no> |
| Newsgroups | comp.lang.asm.x86, comp.os.msdos.programmer |
Cross-posted to 2 groups.
Jim Leonard wrote: > I suppose that, to be truly close to my input set, I should somehow > log program flow and get the frequency of the actual executed > instructions. I believe the DOSBOX emulator, if compiled in debug > mode, can dump the last 20,000 executed instructions. In fact, I > think that would be more realistic; I will look into that. Terje Mathisen <"terje.mathisen at tmsw.no"@giganews.com> wrote: >Not only more realistic, rather "the only reasonable way to do it", and >the method employed by both Intel and AMD when they try to figure out >which instructions to fast-path in hardware and which can be allowed to >run a little slower. Yah, you really need to count the instructions as they're actually being executed by the processor. A lot of instructions in a program will never be executed or executed only once, while a small tight loop might end up getting the majority of execution time. Ross Ridge -- l/ // Ross Ridge -- The Great HTMU [oo][oo] rridge@csclub.uwaterloo.ca -()-/()/ http://www.csclub.uwaterloo.ca/~rridge/ db //
Back to comp.os.msdos.programmer | Previous | Next — Previous in thread | Next in thread | Find similar
Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-26 11:41 -0700
Re: Opcode frequency process/tools? Terje Mathisen <"terje.mathisen at tmsw.no"@giganews.com> - 2011-10-26 21:09 +0200
Re: Opcode frequency process/tools? Ross Ridge <rridge@nospicedham.csclub.uwaterloo.ca> - 2011-10-26 15:54 -0400
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-28 07:07 -0700
Re: Opcode frequency process/tools? Terje Mathisen <"terje.mathisen at tmsw.no"@giganews.com> - 2011-10-29 02:46 +0200
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-31 07:20 -0700
Re: Opcode frequency process/tools? Terje Mathisen <"terje.mathisen at tmsw.no"@nospicedham.giganews.com> - 2011-10-31 19:22 +0100
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-31 07:41 -0700
Re: Opcode frequency process/tools? Bogus@Embarq.com (Steve) - 2011-11-01 13:13 +0000
Re: Opcode frequency process/tools? "Rod Pemberton" <do_not_have@nospicedham.noavailemail.cmm> - 2011-10-26 19:12 -0400
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-28 07:11 -0700
Re: Opcode frequency process/tools? hopcode <hopcode@nospicedham.nulnulul.de> - 2011-10-27 01:18 +0200
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-28 07:13 -0700
Re: Opcode frequency process/tools? Hugh Aguilar <hughaguilar96@nospicedham.yahoo.com> - 2011-10-26 22:28 -0700
Re: Opcode frequency process/tools? "Rod Pemberton" <do_not_have@nospicedham.noavailemail.cmm> - 2011-10-27 13:37 -0400
Re: Opcode frequency process/tools? Hugh Aguilar <hughaguilar96@nospicedham.yahoo.com> - 2011-10-27 22:49 -0700
Re: Opcode frequency process/tools? Jim Leonard <mobygamer@nospicedham.gmail.com> - 2011-10-28 07:24 -0700
Re: Opcode frequency process/tools? Phil Carmody <thefatphil_demunged@nospicedham.yahoo.co.uk> - 2011-10-27 14:00 +0300
Re: Opcode frequency process/tools? "Rod Pemberton" <do_not_have@nospicedham.noavailemail.cmm> - 2011-10-27 13:38 -0400
Re: Opcode frequency process/tools? Robert Wessel <robertwessel2@nospicedham.yahoo.com> - 2011-10-27 14:51 -0500
Re: Opcode frequency process/tools? Hugh Aguilar <hughaguilar96@nospicedham.yahoo.com> - 2011-10-27 22:40 -0700
csiph-web