home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.misc
- Path: sparky!uunet!munnari.oz.au!newsroom.utas.edu.au!cam!esk!scarnie
- From: scarnie@esk.compserv.utas.edu.au (Stuart George Carnie)
- Subject: Re: Can software be 486-specific?
- Message-ID: <1992Nov16.122543.28011@cam.compserv.utas.edu.au>
- Sender: news@cam.compserv.utas.edu.au
- Nntp-Posting-Host: esk
- Organization: University of Tasmania at Launceston
- References: <1992Nov13.153056.7950@tamsun.tamu.edu>
- Date: Mon, 16 Nov 92 12:25:43 GMT
- Lines: 28
-
- In article <1992Nov13.153056.7950@tamsun.tamu.edu>, tpradeep@cs.tamu.edu (Pradeep K Tapadiya) writes:
- >
- >From what I know, 486 has six more instructions compared to a 386.
-
- Well, you are certainly right, but I've just looked at the asm mnemonics, and
- _none_ of them would benefit the operation of something such as Windows NT so
- greatly, as to completely wipe the 386 area of the market. I see that programs
- which say "requires" a 486 wouldn't even use these instructions, and would
- only say this if speed was a major requirement.
- I have also seen ratings of 386-40's
- of 9.74 mips, and it is common for 486-33DX's to have a rating of 10.7mips,
- although I won't forget that 486's have internal cache, which _does_ help.
-
- >
- >My question is, can't each of these instructions be simulated by
- >a set of 386 instructions such that the functionality is not lost?
- >This way we just need to write a parser which will go through
- >the 486 specific executable file looking for these unique
- >instructions and replace it with the equivalent set of 386
- >instructions, thus producing a 386 specific executable file.
-
- I've looked at all the instructions, and they would certainly be easy to
- implement on a 386 system, without CPU time loss.CM2J;N
- --
- scarnie@esk.compserv.utas.edu.au
-
- "On a distant planet, a creature strums his guitar..." {:-)
- ------------------------------------------------------------------------------
-