Path: news.eternal-september.org!eternal-september.org!feeder3.eternal-september.org!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Frank Slootweg <this@ddress.is.invalid>
Newsgroups: comp.os.linux.advocacy,alt.comp.os.windows-11
Subject: Re: How To Speed Startup Of Microsoft Office? Have It Running All The Time!
Date: 28 Mar 2025 20:39:30 GMT
Organization: NOYB
Lines: 77
Message-ID: <vs74uc.cvg.1@ID-201911.user.individual.net>
References: <vs4g23$15bs1$1@dont-email.me> <6vidujd9u5ruhm8b85o52njfjghgoah65a@4ax.com> <vs6v12.uv8.1@ID-201911.user.individual.net> <vs6v5s$3fj4l$1@dont-email.me>
X-Trace: individual.net sHoIFoAx2zlaa8pzo7RwtQ7zO1ot5C1dE15hY+o6T8ux1xW0kw
X-Orig-Path: not-for-mail
Cancel-Lock: sha1:VjxexOXmGSF+4eVo5JExaiq04zc= sha256:QNkDO8MbEVxttPsOgxFqjhqo8TOpJNZWszE+2PSKwIc=
User-Agent: tin/1.6.2-20030910 ("Pabbay") (UNIX) (CYGWIN_NT-10.0-WOW/2.8.0(0.309/5/3) (i686)) Hamster/2.0.2.2

Paul <nospam@needed.invalid> wrote:
> On Fri, 3/28/2025 2:58 PM, Frank Slootweg wrote:
> > Peter Johnson <peter@parksidewood.nospam> wrote:
> >> On Thu, 27 Mar 2025 21:30:43 -0000 (UTC), Lawrence D'Oliveiro
> >> <ldo@nz.invalid> wrote:
> >>
> >>> Microsoft is trying to reduce the time it takes to start Office on
> >>> Windows, by moving part of the work to the time when you boot your PC
> >>> <https://www.theverge.com/news/637469/microsoft-office-speed-boost-faster-launch>.
> >>>
> >>> What a wonderful idea: make an app start faster by making your machine
> >>> take longer to boot. What if other major Windows apps did the same
> >>> thing? Wouldn?t it be cool to have all these apps lurking in the
> >>> background, already running, chewing up memory and CPU cycles?
> >>
> >> Look under the Startup Apps in the task manager and you'll find a
> >> whole load of things that run at startup. Mine includes the Dymo label
> >> printer app, Copernic desktop search and the app that monitors the
> >> battery backup.
> > 
> >   Exactly, nothing new. But perhaps for Lawrence's - apparently - stone
> > age OS, which doesn't know how to have such 'Startup Boost' (and
> > similar) programs without "chewing up memory and CPU cycles", when
> > they're "lurking in the background, already running" [1]. That problem
> > was already solved at least some four decades ago.
> > 
> > [1] Of course his OS *can* do that. After all, it's Unix-like, isn't it?
> 
> A number of the SVCHOST, don't typically use cycles. You can check
> that with Process Explorer. If elevated as Administrator, it can
> do profiling of processes, and it shows a cycle count for the
> item you're tracing. And many SVCHOST are zero. The ones like
> Windows Update support, would not be zero.

  Exactly. In any sane OS, a suspended/blocked/<whatever> process
doesn't use any CPU cycles, period.

> Quiet processes still use memory. A suspended Metro App could still
> take up memory.

  A lesson (not to you) from the very old days: Memory is there to be
'used'. You didn't buy it for nothing. Memory is allocated to all kinds
of things, but that doesn't mean it's in active use and doesn't mean
that it can not be freed/re-used when needed.

  Ever since BSD Unix, memory could be filled upto 90%  (minfree?) and
that was A Good Thing (TM).

  On my Windows systems, Task Manager normally reports a 'Memory usage'
of some 50%, but I assume/hope it's 'lying' and doesn't include memory
which is allocated but not actually in-use by a process. I.e. I start
and exit an editor. The memory used by that program is not released. Is
it counted in 'Memory usage' or not?

>		  Once it is in the run state, the event loop will be
> running, and any time the OS sends an event, the event loop "eats it"
> and that takes a few cycles at a minimum.

  Of course, if a process has to do something, it uses CPU, but when
it's 'idle', it doesn't.

> The OS has a Memory Compressor (it can only be seen in Process Explorer,
> not in Task Manager). If under extreme memory pressure,
> the MS Office Metro.App could have its actual (occupied) memory
> compressed to half the size.
>
> The OS does have a few tricks, to conserve resources.
> 
> But also at times, is a pig. Nobody is perfect :-)
> There is still lots of room for improvements.

  Yes, when the memory pressure becomes too high, any system will
experience trashing, but an idle process - which is the topic of the
(non-)discussion - should present no problem. Can't fit it? Page/swap it
out. 'Problem' solved.

[...]