[Date Prev][Date Next][Subject Prev][Subject Next][
Date Index][
Subject Index]
Re: overcoming the XP lag
- Subject: Re: overcoming the XP lag
- From: "Robert Holmgren" holmgren@xxxxxxxx
- Date: Fri, 5 Nov 2004 03:48:33 -0500
** Reply to message from Michael Norman on Thu, 04 Nov
2004 21:07:25 -0500
>> But when I manually
>> command D KS=0,0 on the CMline, and then do the same spacebar tricks again, I
>> still see the tiniest jerkiness. No change in behavior that I can see.
>
> Hm. I think this cursor movement, movement across either a line of type or
> space with the cursor key, is more affected by the mode con: settings. For
> me it's the same with or without ks=0,0. Not a smooth as my W98 desktop,
> but nothing I'd complain about. The real difference -- which is to say the
> noticeable affect -- is when you type. Without the ks command, I get a lag,
> or a bit of a leap, when you type, as if the cursor is sticking for just a
> millisecond on one character before it moves and creates the next. I can
> type roughly 30 wpm, fast enough to notice the effect.
OK, well that right there is news to me. When Martin first reported this, I
understood, perhaps wrongly, that he was referring to horizontal and vertical
movement of the cursor _without_ typing, i.e. using the arrow keys. So you're
specifically irritated by a lag while typing new text? You're a touch typist
who stares at the screen! I stare at the keyboard or mostly at the ceiling or
out the window at the trees -- I was a pianist -- hardly ever look at the
screen (or the music). Big difference! Do you work in Draft mode and see the
formatting? I work in eXPanded so that I'm not distracted by the formatting,
which I view as a separate process (and as minimal as possible). Anyway...
interesting. Let me do some looking for a change...
Hmmm. I just tried it on two XP Pro machines. I was typing as fast as I
could, and then I gave that up and just typed garbage at a huge clip, four or
five lines per minute, maybe more. The cursor always kept right up with the
characters. Never fell behind. Never jerked even once that I could see.
Tried a more moderate pace, with little pauses, to give the CPU time to do some
time slicing (pay some attention to other processes that are running).
Nothing. Have you got a lot of concurrent processes running? Just a glance at
Taskman indicates about 30 or 40 on my machines, maybe more. Any big apps
running? (In 9x there were very few.) I just tried raising XyWrite's
priority, and the cursor seemed almost ahead of my fingers, it was so fast.
Not that XyWrite isn't stealing most of the CPU in any case! But I'd try
raising the priority of the VDM in which XyWrite is running -- don't go higher
than "high", or use a DOS tool to raise it about 20% (that's a lot). (Do you
know how to set the priority of a process? One way is via Taskman, goto View
==> Options first, Select enough Columns to make the display meaningless, i.e.
at least grab CPU time, base priority, that sort of thing. Then highlight
XyWrite's NTVDM, right-click on it for a context menu, and Set Priority.)
> Question please: is there some clue in the delay between the
> error message and the *done*? Or is that meaningless?
Probably meaningless. I don't know, honestly. But I can issue completely
illegal Default commands like D BG=255,255,-725678 -- and I still get a "Done",
even though the values don't take (no audible protest either) -- so I wouldn't
read too much into the "Done". You issue KS=0,0 _once_, and that's good for
the whole remainder of the session? What's this about the NumLock? You toggle
it *once*, Off-On-Off, and that achieves the same improvement? Somebody asked
what they had in common: they both touch the keyboard handler, but beyond
that... Two things occur to me: 1) are you sure that it is KS, or the 0,0
values of KS, that are achieving this improvement? Could it be the incurring
of an *error* that is doing it? Something in the operating system may be
interfering with Editor in a way that was never
contemplated/anticipated/prepared for by Editor, and consequently something is
being destabilized. Or maybe the keyboard module in Editor is reinitialized in
defensive response to the error. What happens if you manually LOAD MY.KBD
(whatever MY should be)? Or you use the old LDKBD command instead of LOAD?
Or, you try to LOAD a non-existent KBD file?
Have you tried starting XyWrite in FullScreen, and then Alt-Enter into a window?
R
-----------------------------
Robert Holmgren
holmgren@xxxxxxxx
-----------------------------