Article View: comp.arch
Article #10773Re: Comparison of Alpha, MIPS and PA-RISC-II wanted
From: M Darrin Chaney
Date: Tue, 22 Dec 1992 15:46
Date: Tue, 22 Dec 1992 15:46
52 lines
3005 bytes
3005 bytes
In article <1992Dec21.103432.294@rlgsc.com> gezelter@rlgsc.com writes: >In article <BzKGKB.2Bp@pix.com>, stripes@pix.com (Josh Osborne) writes: >> The we need a way for user level code to tell the OS what file info is >> important to cache, and what isn't. Then you can be almost as efficent >> as VMS/RMS (you will have extra context switch overhead for the extra >> syscalls), but you will still be able to allow any user to create new >> filetypes. >-- >Josh, > >Important point that you bring up tangentially in your post. >While VMS does provide RMS as an operating system service, there >is nothing which prevents you from "folling your own" type of >file structure. However, every program which wants to read from >your special file structure will have to be intimately familiar >with it. Programs which use RMS can process files which were >created using RMS with no problems. > >A misnote about the HSC which appeared in this thread. The HSC >does not really know anything about the content of the disk >files. However, RMS will do things inside of the CPU, such as >maintain index caches on a system wide basis, which are a >signifigant performance improvement. One problem with rolling >yourself is that, invariably, the management and monitoring tools >and utilities are bare-bones. > >- Bob Actually, I've seen alot of people here call RMS "the file system." For those of you who still think that RMS is the file system, I'll clue you in: RMS is not the file system. Let's think of what RMS stands for: "Record Management Services." Hmm, sounds more like a record manager than a file system, doesn't it though. The file system is accessed via QIO calls, and knows nothing about file formats. All it knows about is disks and files, files via file ids. It is actually easy to open a file by id and read virtual blocks from it. If you want to roll your own Unix functions (fread and fwrite), this is the way to do it. It might be a little faster, and you needn't worry about RMS. For stream-lf files, just buffer those blocks, and keep track of where the lf is. It's not difficult, and you can easily get around the 65534 limit that RMS places on files. Have fun... Darrin -- mdchaney@iubacs mdchaney@bronze.ucs.indiana.edu mdchaney@rose.ucs.indiana.edu "I want- I need- to live, to see it all..."
Technical Details
Message-ID:
<Bzo3tJ.KGo@usenet.ucs.indiana.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!usc!sol.ctr.columbia.edu!usenet.ucs.indiana.edu!fractal!mdchaneyReferences:
<BzIxqo.C90@dscomsa.desy.de> <BzKGKB.2Bp@pix.com> <1992Dec21.103432.294@rlgsc.com>Show Full Headers
Complete Article Headers
Newsgroups: comp.arch,comp.sys.dec,comp.sys.sgi,comp.sys.hp Path: arc1-ca.pugleaf.net!sparky!uunet!usc!sol.ctr.columbia.edu!usenet.ucs.indiana.edu!fractal!mdchaney From: mdchaney@fractal.ucs.indiana.edu (M Darrin Chaney) Subject: Re: Comparison of Alpha, MIPS and PA-RISC-II wanted Message-ID: <Bzo3tJ.KGo@usenet.ucs.indiana.edu> Sender: news@usenet.ucs.indiana.edu (USENET News System) Nntp-Posting-Host: fractal.ucs.indiana.edu Organization: Indiana University, Bloomington References: <BzIxqo.C90@dscomsa.desy.de> <BzKGKB.2Bp@pix.com> <1992Dec21.103432.294@rlgsc.com> Date: Tue, 22 Dec 1992 15:46:31 GMT Lines: 52