home *** CD-ROM | disk | FTP | other *** search
- Path: tudelft.nl!news
- From: Ejo Schrama <schrama@geo.tudelft.nl>
- Newsgroups: comp.lang.c++
- Subject: Re: Why are 32 bit better than 16 bit pgms?
- Date: 22 Jan 1996 12:13:55 GMT
- Organization: TU Delft, Faculty of Geodetic Engineering
- Message-ID: <4dvv23$mcl@mo6.rc.tudelft.nl>
- References: <4dniuk$lms@dragon.solect.com> <30FBFFE6.1FEB@netcom.com> <4do422$jj1@vixen.cso.uiuc.edu>
- NNTP-Posting-Host: dutgs7.geo.tudelft.nl
- Mime-Version: 1.0
- Content-Type: text/plain; charset=us-ascii
- Content-Transfer-Encoding: 7bit
- X-Mailer: Mozilla 1.12 (X11; I; HP-UX A.09.05 9000/735)
- X-URL: news:4do422$jj1@vixen.cso.uiuc.edu
-
- sjmccaug@prairienet.org (Scott J. McCaughrin) wrote:
- >
- >In a previous article, gallantm@kanservu.ca (Matt Gallant) says:
- >
- >>In article <30FBFFE6.1FEB@netcom.com>, vain@netcom.com says...
- >>>
- >>>I have a simple questions:
- >>>
- >>> What's are 32 bit pgms better than 16 bit programs?
- >>>
- >>> Thanks!
- >>
- >>The answer to this question can be a matter of opinion. The common answers
- >>given are:
- >>
- >>1. For developers, programming becomes easier because you no longer have to
- >>worry about memory segmentation. Because of this, data structures and blocks
- >>are no longer limited to 64K. No more segment:offset addressing. Your
- >>program can "pretend" that memory is a "flat" range, and just use 32-bit pointers
- >>ranging from byte 0 through 4GB. (Of course, not all of this memory area is valid
- >>to the process without making the proper memory allocations)
- >>
- >
- > But with "huge" memory model permits the 64K limit to be exceeded in the
- > real-mode model, and not everyone considers segmented memory a constraint.
-
- Just take a look at the stacksize for the huge memory mode. It is still 64K
- thereby causing a limitation. A 32 C++ compiler like DJGPP doesn't have
- this limitation.
-
- >
- >>2. 32-bit programs are pre-emptively multitasked. A 32-bit application can no
- >>longer hog the process by not yielding on a timely basis. (There is one snag to
- >>this scheme if you're running on Windows 95, and if you develop for Windows 95
- >>you are sure to hear about it. This "snag" is refered to as the Win16Lock or
- >>Win16Mutex.)
- >
- > This is more a feature of Win95 than the memory model -- besides, there are
- > advantages to non-preemptive multitasking. Also, consider CDOS and Concurrent
- > PC-DOS.
- >
-
- IMHO multitasking and 16/32 bit applications are a different thing (in general
- at least). There is no reason why one couldn't develop a multitasking 16 bit
- os. RSX11M on the PDP11-70 was 16 bits and multitasking but is completely outdated
- of course.
-
- >>
- >>3. 32-bit programs are (to a large extent) portable. If written correctly, one set
- >>of source files can be compiled and run on Windows 95 and Windows NT, as well
- >>as being compiled for other non-Intel x86 platforms (i.e. Alpha processors).
- >
- > I believe Win95 supports 16-bit code for backward compatibility as well.
- >
-
- Also this is in general unrelated to the 16/32 bits discussion. Sorry.
-
- >>
- >>4. Also, consider that as new APIs are developed (by Microsoft and 3rd parties)
- >>that they will probably only be ported to the Win32 platforms. For example, I
- >>was just reading an Microsoft Systems Journal article about their new Speech API.
- >> I'm not sure that it's necessarily 32-bit only, but obviously, something as
- >>processing-intensive as speech synthesis and speech recognition will benefit from
- >>32-bits.
- >>
- >>5. Last, but definitely not least, 32-bit programs are allowed to have multiple
- >>threads of execution. This allows you to easily create and execute routines "in the
- >>background" just by creating a new thread. You no longer have to set up your
- >>different tasks to be done and come up with ways of switching between them
- >>yourself, just create a new thread, and Win32 platforms will schedule your
- >>background tasks themselves.
- >>
- > So what? This is the standard for OS/2 as well, which is designed to run in
- > '286 protected-mode (only 24-bit addressing, not 32-bit). Again, this is more
- > a feature of the operating system than the memory model.
- >
- > -- Scott McC.
- >
-
- --
- -------------------------------------------------------------------
- Ernst J.O. (Ejo) Schrama, Faculty of Geodetic Engineering,
- Thijsseweg 11, 2629 JA Delft, Netherlands. schrama@geo.tudelft.nl
- Tel: (31) 15 2784975, Fax: (31) 15 2783711, Page: 065-9762479
- homepage: http://www.geo.tudelft.nl/~fmr/people/schrama.html
- daugther: ftp://www.geo.tudelft.nl/pub/ejo/michelle
- -------------------------------------------------------------------
-
-