home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!rutgers!uwvax!uchinews!ellis!sip1
- From: sip1@ellis.uchicago.edu (Timothy F. Sipples)
- Newsgroups: comp.os.os2.misc
- Subject: Re: 32 Bit HPFS
- Keywords: HPFS
- Message-ID: <1993Jan3.033716.6127@midway.uchicago.edu>
- Date: 3 Jan 93 03:37:16 GMT
- References: <1993Jan2.181023.20211@midway.uchicago.edu> <9HTRwB3w165w@tcscs.UUCP>
- Sender: news@uchinews.uchicago.edu (News System)
- Reply-To: sip1@midway.uchicago.edu
- Followup-To: comp.os.os2.advocacy
- Organization: Dept. of Econ., Univ. of Chicago
- Lines: 57
-
- In article <9HTRwB3w165w@tcscs.UUCP> zeta%tcscs@src.honeywell.com writes:
- >> >Will the new 2.1 release have the 32 bit HPFS included?
- >> No. HPFS386 is a part of LAN Server. (It also provides some extra
- >> features necessary for multiuser operation -- it isn't just a 32-bit
- >> HPFS rewrite.)
- >Forgive the obvious stupid question. With OS/2 being such multitasking,
- >there would be several useful features found in networks which would be
- >useful for a stand alone system as well. Of course I dont know much about
- >LAN's and such, the only multiuser OS I'm familier with is VMS. I'm talking
- >about more abilities with file sharing, especially amongst files between
- >several tasks. More so if it makes programs not normally friendly for file
- >access work more reliably in a shared environment.
-
- Unless I'm missing something, that sort of control is already in OS/2.
- (It's rather important in a single user, multitasking operating
- system.)
-
- I would agree that HPFS386's features can be useful in certain
- contexts. One such context is as a LAN server (which is why it comes
- with LAN Server :-)). For multiple serial users (as in a lab
- setting), a security front end with protection for critical files is a
- good idea; in that case, PC/DACS (Pyramid) is probably the way to go.
-
- >> Lest anyone panic, 32-bit purely for the sake of 32-bit isn't such a
- >> hot idea. For one, you have to spend time debugging. (In other
- >> words, why tamper with code that is working so well?) For another, it
- >> isn't all that clear that you'd get much performance boost. (A disk
- >> -- any disk -- is relatively slow, in computer terms. There's usually
- >> very little CPU involved in disk I/O, and CPU intensive operations is
- >> precisely where you get the biggest performance boost with 32-bit
- >> code.)
- >While this is true for the hardware, it still seems that 32-bit disk interface
- >software could still increase performance. I'm not very familier with the
- >two version file system in OS/2, other than that HPFS is very quick, and the
- >FAT version is a lot faster than in DOS systems. it seems that for some things
- >HPFS is slower than a FAT system. If it were to go 32 bit, wouldn't the HPFS
- >be faster all the time?
-
- Not necessarily. HPFS is more powerful in many respects, and in
- certain rare instances FAT may outperform it.
-
- I'm not arguing that there wouldn't be some improvement, other things
- being equal. I am, however, suggesting that it's probably one of the
- last things you'd want to focus your efforts on.
-
- I should note that I'm guessing a bit at IBM's motives here, but it
- seems pretty obvious. This is one particular design decision I can
- more than live with; I don't think it makes sense to get pedantic
- about 32-bit-ness. The current HPFS code works just great, and I
- can't see any overwhelming need to move it to 32-bit at present. When
- they do move it over, bonus points for IBM.
-
- --
- Timothy F. Sipples | Read the OS/2 FAQ List 2.0h, available from
- sip1@ellis.uchicago.edu | 128.123.35.151, anonymous ftp, in /pub/os2/all/info
- Dept. of Econ., Univ. | /faq, or from LISTSERV@BLEKUL11.BITNET (send "HELP")
- of Chicago, 60637 | [Read the List, THEN post to ONE OS/2 newsgroup.]
-