NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/applications/578

Internet Message Format | 1996-08-05 | 3.5 KB

Path: walrus.megabaud.fi!not-for-mail
From: [e-mail] (Petri Nordlund)
Newsgroups: comp.sys.amiga.applications
Subject: Re: Executive Probs with Maple
Date: 13 Jan 1996 20:12:54 +0200
Organization: Megabaud Oy,Helsinki,Finland
Message-ID: <4d8sn6$[e-mail]>
NNTP-Posting-Host: walrus.megabaud.fi

Matt Harrell <[e-mail]> writes:

>:   I received a registration from Amiga Technologies GmbH last week. They
>:   have no plans regarding Executive, but I hope that now, when even
>:   Windows 95 has got better multitasking than Amiga, a better scheduler,
>:   like Executive has, would be implemented in AmigaOS.
>
>    No offense meant here, but where did you hear that Win95 has
>better multitasking than AmigaOS?  That's far from the truth.
>According to _BYTE_, the only things in Win95 that do preemptive
>multitasking are "agents", which are similar to daemons in UNIX and
>AmigaOS.  Other than that, Win95 is still a coopertive multitasking
>system (and requires *much* more RAM and processor power to boot!).

  Windows 95 preemptively multitasks 32-bit Windows applications,
  and it also supports threads. I'm not sure if processes have
  dynamic priorities, but assume it's so. As far as I know, Agents
  are a bit like Amiga commodities, or UNIX's loadable kernel
  modules. I'm not a Windows fan, far from it, but I without Executive,
  I wouldn't boast very much about Amiga's multitasking capabilities.

>    However, I love Executive, and just sent in my registration
>form and money.  It's an excellent program, and I've noticed a very
>significant improvement in system response time when running heavy CPU
>loads, like a rendering in POV-Ray.  One good example that I've
>noticed is the time it takes to open a new AmigaShell window while
>another is doing a CPU intensive task.  There are times, due to memory
>constraints (that I hope to soon alleviate), when I have to stop the
>Executive server.  The difference in response time when starting a new
>Shell window is quite noticeable.  Great program!

  The reason why this happens is that when you have a CPU intensive
  program running at the same priority as the new shell, the CPU
  intensive program gets most of the available CPU time, not 50%
  as you would expect. This happens, because after every time a piece
  of an executable program is loaded to memory, the CPU intensive
  task gets a chance to run for 4/60 seconds. When you're starting a
  new shell, several programs get executed and when each program
  takes about 2-3 seconds to run, starting a new shell can take
  quite a long time.

>Well, there are programs in UNIX that don't like their priorities
>messed with, too.  Many of them simply will not let the priority be
>changed past a certain value (e.g., HP's GlancePlus software).  (There
>are also some which simply choke and/or die if the priorities of
>different program tasks are screwed up--e.g., HP's "omnimver" program
>that's part of their OmnibackII backup software).  However, this is
>something that a piece of software in AmigaOS, an OS in which most
>people do not intentionally change task priorities, and the OS does
>not normally make provisions for doing it automatically, is not likely
>to pay attention to.  Perhaps they should?

  Yeah, why have priorities if everything is always run with priority 0?
--
                                      __
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~///~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
       Petri Nordlund             __///         [e-mail]
 ---------------------------------\XX/----------------------------------