home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!charnel!rat!usc!sdd.hp.com!saimiri.primate.wisc.edu!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!att!cbnews!shurr
- From: shurr@cbnews.cb.att.com (larry.a.shurr)
- Newsgroups: comp.os.os2.misc
- Subject: Re: Service Pack Results...
- Message-ID: <1992Nov5.155029.20839@cbnews.cb.att.com>
- Date: 5 Nov 92 15:50:29 GMT
- References: <1992Nov3.144905.2229@cbnews.cb.att.com> <1992Nov5.102042.16261@monu6.cc.monash.edu.au>
- Organization: AT&T Bell Laboratories, Columbus, OH
- Lines: 46
-
- In article <1992Nov5.102042.16261@monu6.cc.monash.edu.au> parry@yoyo.cc.monash.edu.au (Tom J Parry) writes:
- }larry.a.shurr (shurr@cbnews.cb.att.com) -- i.e., me -- wrote:
- }> This is my impression, as well. PM Windows appear faster, but windowed
- }> VIO & VDM sessions are slower than before. The VIO & VDM windows appeared
- }> to be updated in interleaved "zones" -- watch carefully during a full
- }> screen paint or a scroll, though maybe it's less visible if you have a
-
- I was not entirely correct when I said this. Full screen paints appear to
- use this interleaved method, but scrolls don't. Instead, it looks like a
- scroll is performed by redrawing the window from the bottom up. Not a
- bitblt to be seen! I find this unsightly as well as horribly slow. I've
- stopped running TE/2 in a window because it's just too slow.
-
- }I have been playing with Linux and X386 on a 486-33 recently and the speed
- }of X (running in 4MB with about 2MB free!) is awesome compared to the wps
- }especially in xterm windows. One of the reasons for this seems to be the
- }way xterm does screen updates. It seems to update the screen to what is
- }happening at the time of the update. So a long multi-page directory listing
- }may be done in 2 or 3 updates. Compared to OS/2's 50 updates for the same
- }action, it makes xterm much faster.
-
- I thought that was something that X Windows did routinely. Maybe it's
- implementation-dependent and the implementations I've seen all do it.
- I certainly agree that the technique avoids "wasted motion," so to
- speak.
-
- }When using full screen OS/2 sessions, I can get 128lines scrolled by in
- }0.78 seconds. I certainly can't see everything that is displayed. If I want
- }to I will pipe the output somewhere. In a windowed session, it scrolls each
- }line. If it just displayed the whole screen as it should appear at the
- }_time_ each update starts then the screen would scroll maybe 25 lines at a
- }time and things wouldn't be so slow and we'd be able to see as much as we
- }can from a full screen session. It's rediculous to redraw the screen for
- }every line of output if it's coming that fast.
-
- Well... I don't want to argue the point since I agree fundamentally that
- when data are coming into the window fast enough, then line-by-line update
- is "wasted motion." However, when the data aren't coming in very fast, at
- 2400 baud say, then line-by-line may be the way to go. Notwithstanding our
- possible disagreement on this point, OS/2 windowed VIO & VDM sessions are
- so slow that it's impractical, even at low data rates.
-
- Larry
- --
- Larry A. Shurr (las@cbnmva.att.com or att!cbnmva!las) speaking only for myself.
- EOR (end-of-ramble)
-