🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.windows.ms.programmer
Article #2857

Re: Windows on a 286

#2857
From: Risto Lankinen
Date: Sun, 26 May 1991 10:42
57 lines
3316 bytes
jarmolow@samsung.com (Tom Jarmolowski) writes:

>	Forget for the moment flames against the 286, I'm not advocating
>buying one but there are a lot of 286 machines out there.  Does anyone
>KNOW why Windows 3.0 is a dog on a fast 286 ?

Hi!

Here are some reasons:
  - With the same CPU speed, some of the opcodes really execute *slower*
    on a 386 than on a 286.  While this is not the case with all opcodes,
    if the code was originally compiled for 80286 & optimized for speed,
    then the resulting code will not be optimal in speed for 80386 (just
    pretty close to optimal, anyway).
  - Because of better protection in 386, far references to code or data
    (which programs made for Windows are full of) take more time to run.
  - Due to virtual memory management, enhanced mode swaps (ie. writes
    to disk) even such segments, which have been declared 'DISCARDABLE'.
    In standard mode they're discarded (ie. 'forgotten' altogether) to
    make room for new allocations.  The enhanced mode starts discarding
    not before the system memory *plus* the swap area are fully crowded.
    Therefore, with certain memory usage, standard mode just re-reads a
    segment while the enhanced mode unnecessarily writes and reads.  (As
    can be noted, this is not a drawback of using the 386, but of using
    the Windows' default operating mode for the 386 .)

>	Is graphics performance the real bottleneck or is the processor
>& memory managment just so brain dead that the rest of the application
>runs slowly too ?

I'm under the impression, that the best you can do with a 286 system and
Windows, is, first, buying a LOT of memory and second, buying a FAST hard
disk.

>	The real question is: Do the windows accelorator cards I saw
>at Comdex make a 286 usable for windows ?

I've had some bad experience of the accelerator cards with Windows.  I've
tested one, which had a 386 that took over the control from the 286 on the
motherboard, turning it to handle the interrupts & I/O only.

What happened with Windows?  Well, the mouse interrupt is regarded pretty
low in priority - usually.  With the accelerator card, the 286 became the
only source of interrupts for the 386, so the 'true' CPU lost the priority
information of the interrupt - it saw them only coming from one source.
This had the effect, that most of the execution time was spent serving the
mouse, which generates interrupts in rapid succession all of the time it's
being moved.

The moral?  Check the card with Windows before buying.

Terveisin: Risto Lankinen
-- 
Risto Lankinen / product specialist ***************************************
Nokia Data Systems, Technology Dept *  2                              3   *
THIS SPACE INTENTIONALLY LEFT BLANK * 2 +1 is PRIME!  Now working on 2 -1 *
replies: risto@yj.data.nokia.fi     ***************************************
Technical Details
Message-ID: <1186@tuura.UUCP>
Path: arc1-ca.pugleaf.net!utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!sdd.hp.com!elroy.jpl.nasa.gov!decwrl!fernwood!uupsi!sunic!sics.se!fuug!tuura!risto
References: <24462@samsung.samsung.com>

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!sdd.hp.com!elroy.jpl.nasa.gov!decwrl!fernwood!uupsi!sunic!sics.se!fuug!tuura!risto
From: risto@tuura.UUCP (Risto Lankinen)
Newsgroups: comp.windows.ms.programmer
Subject: Re: Windows on a 286
Message-ID: <1186@tuura.UUCP>
Date: 26 May 91 10:42:54 GMT
References: <24462@samsung.samsung.com>
Organization: Nokia Data Systems Oy
Lines: 57