usenet@cps3xx.UUCP (Usenet file owner) (10/18/89)
>Also, the extra layer activity will take much more time with MANY windows >open on the screen. (I have had many times when I had 15 to 20 windows of >various types open on the workbench screen...) Is this problem addressed in 1.4? Currently the Layers performance is *terrible* when more than about five windows are open. On the very few occainsions that I have used a Mac, having a lot of windows open did not really slow things down, som its gotta be possible (I know, they don't have to worry about things like SMART_REFRESH) but an Amiga screen with a bunch of SIMPLE_REFRESH windows open (say 10) can be real slow. What about the bug in layers where the RastPort gets all fragmented? > >and will cause a few problems. What I really want is that the screen >front/back gadgets look different from the window front/back gadgets. >(They are, after all, a different beast...) Speaking of which.... I do some programming/consulting work on the side for a couple of people who own amigas. They are not stupid, but generally know little about the inner magic of computers. Even though they have had Amigas for a while, they still are (were, i explained) confused about the difference between screens and windows. Maybe the included docs need to explain this a whole lot better. The included docs should contain a QUICK_REFERENCE card with info about such things as the leftAmiga V, B, N, M switchs. Joe Porkka porkka@frith.egr.msu.edu
nsw@cbnewsm.ATT.COM (Neil Weinstock) (10/19/89)
In article <5028@cps3xx.UUCP> porkka@frith.UUCP (Joe Porkka) writes: [ ... ] >On the very few occainsions that I have used a Mac, having a lot >of windows open did not really slow things down, som its gotta Can't get any slower... (sorry, couldn't resist) ________________ __________________ ____________________________ //// \\// \\// \\\\ \\\\ Neil Weinstock //\\ att!cord!nsw or //\\ "Oh dear, now I shall have //// //// AT&T Bell Labs \\// nsw@cord.att.com \\// to create more Martians." \\\\ \\\\________________//\\__________________//\\____________________________////
mks@cbmvax.UUCP (Michael Sinz - CATS) (10/19/89)
In article <5028@cps3xx.UUCP> porkka@frith.UUCP (Joe Porkka) writes: >>Also, the extra layer activity will take much more time with MANY windows >>open on the screen. (I have had many times when I had 15 to 20 windows of >>various types open on the workbench screen...) > >Is this problem addressed in 1.4? Currently the Layers performance >is *terrible* when more than about five windows are open. > >On the very few occainsions that I have used a Mac, having a lot >of windows open did not really slow things down, som its gotta >be possible (I know, they don't have to worry about things like SMART_REFRESH) >but an Amiga screen with a bunch of SIMPLE_REFRESH windows >open (say 10) can be real slow. > While there may be a few enhancements to the speed of the layers, the real difference between the MAC and the Amiga is that on the MAC, the active window that is doing the input and rendering is forced to be in front. This makes life much simpler from layer's point of view as the rendering is not obscured in any way and can be done directly inot the screen. Then, when the window does get obscured it just forgets what was there and the new window takes its place. Any time a window comes up, it must redraw all parts of itself that were obscured. Part of the reason the Amiga seems faster is that many programs use the SMART_REFRESH windows and don't have that overhead. (Side note: The MAC at its fastest in handling windows is still slower than my Amiga with 15 windows on workbench. IMHO) [rest deleted] > Joe Porkka porkka@frith.egr.msu.edu /----------------------------------------------------------------------\ | /// Michael Sinz -- CATS/Amiga Software Engineer | | /// PHONE 215-431-9422 UUCP ( uunet | rutgers ) !cbmvax!mks | | /// | |\\\/// When people are free to do as they please, | | \XX/ they usually imitate each other. | \----------------------------------------------------------------------/
new@udel.edu (Darren New) (10/19/89)
In article <5682@cbnewsm.ATT.COM> nsw@cbnewsm.ATT.COM (Neil Weinstock) writes: >In article <5028@cps3xx.UUCP> porkka@frith.UUCP (Joe Porkka) writes: >[ ... ] >>On the very few occainsions that I have used a Mac, having a lot >>of windows open did not really slow things down, som its gotta >Can't get any slower... Actually, I think it does get much slower, but the "premptive" (sp?) nature of the Mac "inner loop" keeps you from noticing. Basically, when you get window damage, a simple flag is set in the window structure. Later, if you ask for input (i.e., input events are generated only when you call GetMsg()) and there is no real input, the window list is scanned from front to back looking for damage flags and the "input event" returned is a refresh event. Hence, if you damage a bunch of windows, you can still type right away; you just won't get updated back windows. This becomes most evident in the finder when you close bunches of windows; they don't update before they close. This is also nice because key-repeat does not "run away" if you hold the key down. Basically, if you ask for input, it generates a key repeat only if the key is still down and at least X time has elapsed since last time it returned the key. Hence, if the program does not look at the keyboard, the keys don't repeat. Too bad this is hard to do on the Amiga. -- Darren
limonce@pilot.njin.net (Tom Limoncelli) (10/23/89)
In article <5028@cps3xx.UUCP> usenet@cps3xx.UUCP (Usenet file owner) writes: > Is this problem addressed in 1.4? Currently the Layers performance > is *terrible* when more than about five windows are open. > Joe Porkka porkka@frith.egr.msu.edu It's my understanding that the whole layers concept comes from a paper presented at SIGGRAPH in the late 70's. (exact year & author unknown to me. Does anyone know? I'd like to go about getting the article). I've been told that the layers.library is the pseudo-code from the paper implemented in C almost verbatium. The person that told me this added that there is room for optimization but since it would be a big ROM change, it wouldn't be in 1.2 or 1.3 (this was a while ago). Did C-A have the time/personpower to optimize the routines? Inquiring minds want to know. -- Drew University -- Tom Limoncelli C M Box 1060 -- limonce@pilot.njin.net limonce@pilot.rutgers.edu P O Box 802 -- tlimonce@drunivac.Bitnet <<<-has a "T" in it. Madison, NJ 07940 -- 201-408-5389 ...!rutgers!njin!limonce
rap@peck.ardent.com (Rob Peck) (10/24/89)
In article <8221@cbmvax.UUCP> mks@cbmvax.UUCP (Michael Sinz - CATS) writes: >In article <5028@cps3xx.UUCP> porkka@frith.UUCP (Joe Porkka) writes: >>>Also, the extra layer activity will take much more time with MANY windows >>>open on the screen. (I have had many times when I had 15 to 20 windows of >>>various types open on the workbench screen...) > The use of SMART_REFRESH does, at times, slow things down, but NOWHERE near as much slowdown as when someone uses a GIMMEZEROZERO window. The only time I believe that such a window is necessary is when it is being used to display a SUPERBITMAP window... The nominal idea to GIMMEZEROZERO is that the upper lefthand corner of the window's RastPort is 0,0 for drawing purposes instead of having to offset for drawing to account for the top and left margin, as well as to manually clip the drawing to the other bounds of the visible part of the window. The necessary reason for the availability of a GIMMEZEROZERO window is that it prevents Intuition from rendering its gadgets and borders into the superbitmap. Intuition does this by providing -->5<--- layers totally, one for the drawable area, one for the top, bottom, left, and right edges of the window. THATS when operations on windows can really get bogged down. If a developer does not realize this, and uses a GIMMEZEROZERO when it is not needed, it slows everything down. Talk about extra layer activity... whew!! Rob Peck