how to delete win386.swp? [Archive]

July 27th 04, 06:23 PM

I think this will be easier for you:

1-Go to Control Panel / System / . Then set the virtual memory by yourself. Use any max value (256 MB could be OK)
2-Reboot. You will find that a new and smaller swp file is in C directory instead of Windows
3-Since no one like to set the virtual memory, you can go to Control Panel / System and reset virtual memory by letting Windows manage your virtual memory
4-Reboot, you will find a new swp file in the WINDOWS directory

"Don Phillipson" wrote:

> " Bill in Verona, WI" > wrote in message
> ...
>
> > I'm getting KERNEL32.DLL errors and it was suggested that I delete
> > win386.swp but I get an "access denied" message.
>
> 1. WIN386.SWP is the essential Windows swap file i.e.
> no normal software ever proposes deleting it. This is why
> Windows will not let you do so (although DOS is not
> thus limited, see BB's post.)
> 2. Have you checked details? There is a notorious
> virus that calls itself KERNAL32 (not the legitimate
> Windows KERNEL32.)
> 3. If you truly have some problem with WIN386.SWP
> you should do two things first:
> -- search to see if there are two or more on your
> hard drive(s). There should be only one.
> -- / Start / Settings / Control Panel / System / Performance
> and click on button Virtual Memory.
> If in doubt, select first option, Let Windows manage it.
> You may assign the swap file to another drive if you want.
>
> --
> Don Phillipson
> Carlsbad Springs (Ottawa, Canada)
>
>
>

PCR

July 27th 04, 09:53 PM

A very respectable name-- it was Harper, MVP (& not Martell, MVP)--
has said: Under certain circumstances, perhaps rare, there can be a
serious crash, if a maximum is set for the Swap File, and there is no
telling how big it may wish to grow. HE HAS SEEN IT, with his own eyes.
(Now, he is blind.) So, forget about the 2.5x usable RAM rule. (He now
sits at his keyboard, day & night, waiting to pounce, should anyone
suggest using it.) I didn't do it long! I couldn't bear the sight of
nearly a GB worth of Win386.swp, (although it seemed to function well).

Unless turned off, Windows is always managing VM. It manages it based on
what is shown shadowed in min & max at "R-Clk My Computer, Properties,
Performance tab, Virtual Memory button". That is OK at defaults of zero
& "No maximum". The only MVP approved alteration is to set the min
larger, based on careful observation over days. You would set it at 20%
larger than the greatest value you saw for Swap File "SIZE" (not "use"),
using System Monitor. If your observation discovers a constant &
frequent Swap File "USE" figure, THAT is when to consider additional
RAM.

WARNING: Once, I turned off VM for a zero swap file. All seemed to run
well, UNTIL I opened a Windows DOS (in a box) session. It produced an
orderly crash of the DOS box. Later, online, I did it again, just to
report the experience accurately. The crash this time was horrendous- IT
HAD A HEARTBEAT.

There is one other setting I've tried & rejected:
"ConservativeSwapfileUsage=1", in System.ini, [386.Enh] Section. This,
gave me a Win386.swp file generally 10-20 MB in size, sometimes zero.
That's with Windows in control of the size, and min=0, max=No max. It
was beautiful to watch. When the Swap file was in "use" (say, 20 MB),
the "size" went to around 100 MB. However, there was resizing going on,
even when "swap file in use" was zero (Windows tries to anticipate how
big to make it, depending on what your doing), and, again, my
sluggishness persisted.The Microsoft literature says it's less
efficient, as well. Although I generally loved seeing the small
Win386.swp sizes, I discontinued "ConservativeSwapfileUsage=1". (That
"sluggishness" is gone; it was something entirely unrelated to VM. In
fact, I was never able to pin anything on VM at all, except, I believe,
Win386.swp gets in the way of a proper Scandisk/Defrag.) This setting
might be appropriate for one who has so much RAM, as to never have a
"use" figure.

Do you have "System Monitor" in START... System Tools? If not, get it
from "START, Settings, Control Panel, Add/Remove Programs, Windows Setup
tab, D-Clk System Tools, check System Monitor, OK, Apply, OK". May as
well take "Resource Meter", too. Now, go through the menus and at least
have it display

(a) Swap file in use.
(b) Swap file size.
(c) Swappable memory.
(d) Unused physical memory.
(e) Allocated memory.
(f) Disk cache size.
(g) Locked memory
(h) Other memory
(i) Kernel Processor Usage
(j) Kernel Threads

Hopefully, by the time you have it set up and commence to asking
questions, someone will be around who can answer them.

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"JD" > wrote in message
...
| I think this will be easier for you:
|
| 1-Go to Control Panel / System / . Then set the virtual memory by
yourself. Use any max value (256 MB could be OK)
| 2-Reboot. You will find that a new and smaller swp file is in C
directory instead of Windows
| 3-Since no one like to set the virtual memory, you can go to Control
Panel / System and reset virtual memory by letting Windows manage your
virtual memory
| 4-Reboot, you will find a new swp file in the WINDOWS directory
|
|
| "Don Phillipson" wrote:
|
| > " Bill in Verona, WI" > wrote in message
| > ...
| >
| > > I'm getting KERNEL32.DLL errors and it was suggested that I
delete
| > > win386.swp but I get an "access denied" message.
| >
| > 1. WIN386.SWP is the essential Windows swap file i.e.
| > no normal software ever proposes deleting it. This is why
| > Windows will not let you do so (although DOS is not
| > thus limited, see BB's post.)
| > 2. Have you checked details? There is a notorious
| > virus that calls itself KERNAL32 (not the legitimate
| > Windows KERNEL32.)
| > 3. If you truly have some problem with WIN386.SWP
| > you should do two things first:
| > -- search to see if there are two or more on your
| > hard drive(s). There should be only one.
| > -- / Start / Settings / Control Panel / System / Performance
| > and click on button Virtual Memory.
| > If in doubt, select first option, Let Windows manage it.
| > You may assign the swap file to another drive if you want.
| >
| > --
| > Don Phillipson
| > Carlsbad Springs (Ottawa, Canada)
| >
| >
| >

PCR

July 28th 04, 09:52 PM

Careful! Harper is following this thread by the sound of key clicks!
Indeed, it is presumable the same horrible crash can occur that way, not
enough room for growth on the partition. However, that is no argument in
favor of setting a Max. Rather, keep the partition "fit as a fiddle &
feeling so good". Then, it may "jump over the moon, if it could". BUT...
you don't want it to crash it's way there!

Seriously, here is something I have done that I swear is beneficial...

I used to turn VM off, and I swear it makes for a better Defrag.
However...

(1) It is possible one may not be able to reboot with VM off. That would
be the case, if one has only 32 MB RAM. There was one poster who said he
had 64 MB & could not reboot when he turned off VM. However, "Scanreg
/Restore", choosing the most recent backup, should reset VM to being on.

(2) DO NOT "START, Programs, MS-DOS Prompt", with VM off. I had an
orderly crash doing that. Repeating the experience online to better
report it, my second crash doing it was HORRENDOUS-- it had a heartbeat!
Also, some apps will open a Windows DOS box on their own; so, beware!

Here is what I have now done, so that VM never need be turned off & the
Swap File will be out of the way...

(a) Set swap file minimum size to 300 MB, & reboot back to windows.
Do so at "Control Panel, System, Performance tab, Virtual Memory
button". This will create a 300 MB space for a swap file. It will not be
entirely contiguous. Set the minimum only.
(b) Boot to DOS
(c) SMARTDRV
(to speed up the following considerably)
(d) COPY Win386.swp Win386.jnk
(This fills in gaps between files, so that later Win386.swp will
be contiguous.)
(e) REN Win386.swp Win386.old
(Win386.swp will recreate at boot.)
(f) Boot to Windows
(Now, you have Win386.jnk, Win386.old, plus a new Win386.swp.)
(g) In Explorer, delete Win386.jnk & Win386.old. Hold Shift as clicking
to Delete to prevent it goes into the Recycle Bin!
(h) Do whatever you normally do, to keep the following from constantly
restarting.
(i) Scandisk (Thorough, usually w/o write checking)
(j) Defrag

Now I had my buffer between the swap file & the rest of my files. This
is meant to keep the swap file out of the way of the rest of the system
during a Defrag. There may have been some little bit of the swap file
still among the other files, but the vast bulk of it was one contiguous
block 300 MB away! I saw it, by pausing Defrag before it ended &
scrolling past a sea of white to the sea of red! (By the way, "REN
Win386.swp Win386.old" created a zero byte file, or I am a madman!)

Notes:
1. This need be done but once this way, except should you subsequently
delete Win386.swp. Anyhow, mine has stayed put!
2. The 300 MB swap file has moved to the root folder C:\, because a
minimum size was specified.
3. You still have a "C:\Windows\Win386.swp", but it is zero bytes in
size.
4. It has been promised "C:\Win386.swp" will never move physically on
the hard drive. Those 300 MBs will stay put. However, should the swap
file get larger than that, the excess will intermix with other files.
When the swap file shrinks, it will then return to the 300 MB area.
5. The purpose of Win386.jnk was to give the system room for growth,
before it must use disk space on the other side of the swap file. (It
does appear to work that way.)

Is it all worth it? I really think it did make a difference.

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"... et al." > wrote in message
...
| PCR wrote:
|
| > A very respectable name-- it was Harper, MVP (& not Martell, MVP)--
| > has said: Under certain circumstances, perhaps rare, there can be a
| > serious crash, if a maximum is set for the Swap File, and there is
no
| > telling how big it may wish to grow. HE HAS SEEN IT, with his own
eyes.
| > (Now, he is blind.) So, forget about the 2.5x usable RAM rule. (He
now
| > sits at his keyboard, day & night, waiting to pounce, should anyone
| > suggest using it.) I didn't do it long! I couldn't bear the sight of
| > nearly a GB worth of Win386.swp, (although it seemed to function
well).
|
| Hmm..
|
| With no set maximum but letting the OS dynamically change it. If the
| Swap for whatever reason starts to grow and grow and grow, wouldn't
the
| unavoidable limit, the size of the harddrivepartition it is located
| on, lead to the same blinding serious crash?
|
| With todays harddirvesizes, you could almost set a fixed maximum
larger
| then the harddrivesizes most people had in their computers back in
'99.
|
| --
| Please followup in newsgroup.
| E-mail address is invalid due to spam-control.

... et al.

July 29th 04, 03:28 AM

PCR wrote:

> Careful! Harper is following this thread by the sound of key clicks!
> Indeed, it is presumable the same horrible crash can occur that way, not
> enough room for growth on the partition. However, that is no argument in
> favor of setting a Max. Rather, keep the partition "fit as a fiddle &
> feeling so good". Then, it may "jump over the moon, if it could". BUT...
> you don't want it to crash it's way there!
>

The arguments are usually:
1) Don't let Windows manage your virtual memory (it does it poorly
(¿possibly by not doing #3 & #4?)).
2) Set the maximum size large enough so the limit isn't reached (to
avoid sudden severe crashes).
3) Set the minimum size equal to the maximum size (to avoid fragmentation).
4) If a second harddrive is available, place the swapfile on a partition
there (so if the swapfile is used, the heads on the first HDD doesn't
have to jump back and forth between the swapfile and where it is
otherwise working).

You claimed Mr Harper contradicted #2, and thereby also invalidated #1 &
#3. I didn't see the difference in running Win98 with a 1 GB HDD, 32
MB RAM and letting Windows manage the swapfile versus running Win98 with
a 40 GB HDD, 512 MB RAM and setting a fixed minimum equal with a fixed
maximum swapfile size of 1 GB.

FWIW, on this machine i have 768 MB RAM and the Swapfile size
min=max=256 MB, and i don't know that i've had any crashes or problems
related to it's use or limits.

--
Please followup in newsgroup.
E-mail address is invalid due to spam-control.

Richard G. Harper

July 29th 04, 04:10 AM

Someone calleth my name? :-)

The danger in setting a maximum to the swapfile size is that if you ever
exceed (memory+swapfile) with loaded programs and data, your system will
crash.

Guaranteed, 100% certain.

So what?

Well, if you're lucky, all it will do is crash. If you're not lucky, it
will crash and you'll lose all unsaved work. If you're REALLY unlucky, when
it crashes it will corrupt the hard drive and you can lose much more than
just unsaved data. I have seen this happen with my own two eyes, resulting
in a trashed hard drive and a complete wipe and reinstall.

So the question is ... "If you can set a minimum and raise the maximum as
high as it will go and avoid a crash, why would you ever set a maximum size
and risk a crash?"

There is no benefit - zero, zip, zilch, none, nada, nein - to setting a
maximum swapfile size that is smaller than the largest allowable value.
None whatsoever. No performance benefit, no memory benefit, no resource
benefit. So why take an action that you know is risky?

--
Richard G. Harper [MVP Win9x]
* PLEASE post all messages and replies in the newsgroups
* for the benefit of all. Private mail is usually not replied to.
* HELP us help YOU ... http://www.dts-l.org/goodpost.htm

"... et al." > wrote in message
...
> PCR wrote:
>
>> Careful! Harper is following this thread by the sound of key clicks!
>> Indeed, it is presumable the same horrible crash can occur that way, not
>> enough room for growth on the partition. However, that is no argument in
>> favor of setting a Max. Rather, keep the partition "fit as a fiddle &
>> feeling so good". Then, it may "jump over the moon, if it could". BUT...
>> you don't want it to crash it's way there!
>>
>
> The arguments are usually:
> 1) Don't let Windows manage your virtual memory (it does it poorly
> (¿possibly by not doing #3 & #4?)).
> 2) Set the maximum size large enough so the limit isn't reached (to avoid
> sudden severe crashes).
> 3) Set the minimum size equal to the maximum size (to avoid
> fragmentation).
> 4) If a second harddrive is available, place the swapfile on a partition
> there (so if the swapfile is used, the heads on the first HDD doesn't have
> to jump back and forth between the swapfile and where it is otherwise
> working).
>
> You claimed Mr Harper contradicted #2, and thereby also invalidated #1 &
> #3. I didn't see the difference in running Win98 with a 1 GB HDD, 32 MB
> RAM and letting Windows manage the swapfile versus running Win98 with a 40
> GB HDD, 512 MB RAM and setting a fixed minimum equal with a fixed maximum
> swapfile size of 1 GB.
>
> FWIW, on this machine i have 768 MB RAM and the Swapfile size min=max=256
> MB, and i don't know that i've had any crashes or problems related to it's
> use or limits.
>
> --
> Please followup in newsgroup.
> E-mail address is invalid due to spam-control.

... et al.

July 29th 04, 05:14 AM

Richard G. Harper wrote:

> Someone calleth my name? :-)
>
> The danger in setting a maximum to the swapfile size is that if you ever
> exceed (memory+swapfile) with loaded programs and data, your system will
> crash.
>
> Guaranteed, 100% certain.
>
> So what?
>
> Well, if you're lucky, all it will do is crash. If you're not lucky, it
> will crash and you'll lose all unsaved work. If you're REALLY unlucky, when
> it crashes it will corrupt the hard drive and you can lose much more than
> just unsaved data. I have seen this happen with my own two eyes, resulting
> in a trashed hard drive and a complete wipe and reinstall.
>
> So the question is ... "If you can set a minimum and raise the maximum as
> high as it will go and avoid a crash, why would you ever set a maximum size
> and risk a crash?"

[Arguing the side of trying to avoiding swapfile fragmentation]
If setting the minimum size so that it /almost/ never is used up. Then
assuming that Win386.swp is the size of set minimum, and if/when a need
arises to use the swap the OS first uses this *without* /feeling/ like
suddenly growing the file a bit just in case it might need more in the
future. - Suddenly. - I see the point. :-)

>
> There is no benefit - zero, zip, zilch, none, nada, nein - to setting a
> maximum swapfile size that is smaller than the largest allowable value.
> None whatsoever. No performance benefit, no memory benefit, no resource
> benefit. So why take an action that you know is risky?
>

The benefit would then be that even as you experience the hard crash,
you can rest assure that the swapfile haven't exceeded the limit *you*
set for it.

As i stated i don't think i have had any problems related to this but,
since now i can see, i feel an urge to change my settings. :-)

--
Please followup in newsgroup.
E-mail address is invalid due to spam-control.

PCR

July 29th 04, 06:06 AM

I think you are on the verge of a correct decision. I can testify
that... After establishing a contiguous Minimum for the Swap File, that
space will stay put. (Naturally, should you delete it in DOS or readjust
it's size, then it will go to different tracks/calendars/whatever.)

I don't believe it has ever exceeded the 300 MB I set for it, as
explained in an earlier post. I have a notarized affidavit from Terhune
& others that, should it ever exceed the Minimum, it will shrink back
into it afterwards.

So, the 300 MB stay put, but the file may grow out & shrink back in.

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"... et al." > wrote in message
...
| Richard G. Harper wrote:
|
| > Someone calleth my name? :-)
| >
| > The danger in setting a maximum to the swapfile size is that if you
ever
| > exceed (memory+swapfile) with loaded programs and data, your system
will
| > crash.
| >
| > Guaranteed, 100% certain.
| >
| > So what?
| >
| > Well, if you're lucky, all it will do is crash. If you're not
lucky, it
| > will crash and you'll lose all unsaved work. If you're REALLY
unlucky, when
| > it crashes it will corrupt the hard drive and you can lose much more
than
| > just unsaved data. I have seen this happen with my own two eyes,
resulting
| > in a trashed hard drive and a complete wipe and reinstall.
| >
| > So the question is ... "If you can set a minimum and raise the
maximum as
| > high as it will go and avoid a crash, why would you ever set a
maximum size
| > and risk a crash?"
|
| [Arguing the side of trying to avoiding swapfile fragmentation]
| If setting the minimum size so that it /almost/ never is used up. Then
| assuming that Win386.swp is the size of set minimum, and if/when a
need
| arises to use the swap the OS first uses this *without* /feeling/ like
| suddenly growing the file a bit just in case it might need more in the
| future. - Suddenly. - I see the point. :-)
|
| >
| > There is no benefit - zero, zip, zilch, none, nada, nein - to
setting a
| > maximum swapfile size that is smaller than the largest allowable
value.
| > None whatsoever. No performance benefit, no memory benefit, no
resource
| > benefit. So why take an action that you know is risky?
| >
|
| The benefit would then be that even as you experience the hard crash,
| you can rest assure that the swapfile haven't exceeded the limit *you*
| set for it.
|
| As i stated i don't think i have had any problems related to this but,
| since now i can see, i feel an urge to change my settings. :-)
|
| --
| Please followup in newsgroup.
| E-mail address is invalid due to spam-control.

Richard G. Harper

July 29th 04, 12:02 PM

You can set the minimum size to be as big as you want, and if set
sufficiently high the swapfile should never feel the need to resize. Thus
you avoid fragmentation without any of the down side issues.

I think the only reason any contrarian discussion continues on this point is
because some web sites and/or "experts" that like to discuss swapfile
optimization have failed to keep up with the state of the art in computers.
When Windows 98 came out computers came with very small hard drives and
comparatively little memory, and upgrading either was horrendously
expensive. Today both memory and hard drives are commodity-priced and
there's really no reason to stint on either.

--
Richard G. Harper [MVP Win9x]
* PLEASE post all messages and replies in the newsgroups
* for the benefit of all. Private mail is usually not replied to.
* HELP us help YOU ... http://www.dts-l.org/goodpost.htm

"... et al." > wrote in message
...
> Richard G. Harper wrote:
>
>> Someone calleth my name? :-)
>>
>> The danger in setting a maximum to the swapfile size is that if you ever
>> exceed (memory+swapfile) with loaded programs and data, your system will
>> crash.
>>
>> Guaranteed, 100% certain.
>>
>> So what?
>>
>> Well, if you're lucky, all it will do is crash. If you're not lucky, it
>> will crash and you'll lose all unsaved work. If you're REALLY unlucky,
>> when it crashes it will corrupt the hard drive and you can lose much more
>> than just unsaved data. I have seen this happen with my own two eyes,
>> resulting in a trashed hard drive and a complete wipe and reinstall.
>>
>> So the question is ... "If you can set a minimum and raise the maximum as
>> high as it will go and avoid a crash, why would you ever set a maximum
>> size and risk a crash?"
>
> [Arguing the side of trying to avoiding swapfile fragmentation]
> If setting the minimum size so that it /almost/ never is used up. Then
> assuming that Win386.swp is the size of set minimum, and if/when a need
> arises to use the swap the OS first uses this *without* /feeling/ like
> suddenly growing the file a bit just in case it might need more in the
> future. - Suddenly. - I see the point. :-)
>
>>
>> There is no benefit - zero, zip, zilch, none, nada, nein - to setting a
>> maximum swapfile size that is smaller than the largest allowable value.
>> None whatsoever. No performance benefit, no memory benefit, no resource
>> benefit. So why take an action that you know is risky?
>>
>
> The benefit would then be that even as you experience the hard crash, you
> can rest assure that the swapfile haven't exceeded the limit *you* set for
> it.
>
> As i stated i don't think i have had any problems related to this but,
> since now i can see, i feel an urge to change my settings. :-)
>
> --
> Please followup in newsgroup.
> E-mail address is invalid due to spam-control.

... et al.

July 29th 04, 06:39 PM

Richard G. Harper wrote:

> You can set the minimum size to be as big as you want, and if set
> sufficiently high the swapfile should never feel the need to resize. Thus
> you avoid fragmentation without any of the down side issues.
>
> I think the only reason any contrarian discussion continues on this point is
> because some web sites and/or "experts" that like to discuss swapfile
> optimization have failed to keep up with the state of the art in computers.
> When Windows 98 came out computers came with very small hard drives and
> comparatively little memory, and upgrading either was horrendously
> expensive. Today both memory and hard drives are commodity-priced and
> there's really no reason to stint on either.
>

Repeating the advice i previously followed regarding Virtual memory:
1) Don't let Windows manage your virtual memory (it does it poorly
(¿possibly by not doing #3 & #4?)).
2) Set the maximum size large enough so the limit isn't reached (to
avoid sudden severe crashes).
3) Set the minimum size equal to the maximum size (to avoid fragmentation).
4) If a second harddrive is available, place the swapfile on a partition
there (so if the swapfile is used, the heads on the first HDD doesn't
have to jump back and forth between the swapfile and where it is
otherwise working).

I saw the point you made but after testing for a while i'm back at using
settings based on the 4 points made above, because two things bugged me.

I experienced just what PCR was mentioning, that the swap-file started
out bigger then the set minimum and then changed (shrinking) over time,
i.e. it didn't stay a fixed size like i wanted it to.

If setting a fixed minimum, you are also forced to set a maximum, there
is no way to tell it to use maximum available on the partition. *You
have to give it a value in MegaBytes.* So after having put other files
on that partition and opening the Virtual Memory settings page again,
one can notice that Windows have noticed that there is less space
available and adjusted the set maximum down to what is currently
available. Taking files off that partition and looking again one can see
Windows setting the max-setting higher again, but it doesn't set it
higher then what was origianlly set.

(So the OS kind of changes both the minimum and maximum settings on its
own accord.)

The solution to both these annoyances, and maybe you agree the optimal
setting, and the case where your suggestion and the 4-part advice given
above coincide, would be to put the swapfile on a partition (say 1 GB)
on its own, set the minimum size high, as high as the maximum, which in
turn then is set as big as the partitionsize. Hmm, sounds pretty
analougous to a Linux Swap-partition. The next step would be to make it
not mount, so you wouldn't see it in 'My Comuter' and thus wouldn't muck
around with it by misstake.

I have no such dedicated partition currently, maybe i will after i
reformat my drives next time, but i've returned to using Swapfile on a
not often used partition on the second drive with minimum = maximum. But
i have upp'ed the size from 256 to 384 .. just in case. ;-)

--
Please followup in newsgroup.
E-mail address is invalid due to spam-control.

Richard G. Harper

July 30th 04, 12:52 AM

#1 is an ignorant statement with no basis in fact. #2 is okay, #3 is
reasonably okay if you don't mind tying up as much as 2gb of hard drive
space, #4 is bang-on correct as long as you remember the words "on a second
physical hard drive". If all you do is create two partitions on a single
drive and put the swapfile on the second partition you're already fighting
against yourself.

--
Richard G. Harper [MVP Win9x]
* PLEASE post all messages and replies in the newsgroups
* for the benefit of all. Private mail is usually not replied to.
* HELP us help YOU ... http://www.dts-l.org/goodpost.htm

"... et al." > wrote in message
...
> Richard G. Harper wrote:
>
>> You can set the minimum size to be as big as you want, and if set
>> sufficiently high the swapfile should never feel the need to resize.
>> Thus you avoid fragmentation without any of the down side issues.
>>
>> I think the only reason any contrarian discussion continues on this point
>> is because some web sites and/or "experts" that like to discuss swapfile
>> optimization have failed to keep up with the state of the art in
>> computers. When Windows 98 came out computers came with very small hard
>> drives and comparatively little memory, and upgrading either was
>> horrendously expensive. Today both memory and hard drives are
>> commodity-priced and there's really no reason to stint on either.
>>
>
> Repeating the advice i previously followed regarding Virtual memory:
> 1) Don't let Windows manage your virtual memory (it does it poorly
> (¿possibly by not doing #3 & #4?)).
> 2) Set the maximum size large enough so the limit isn't reached (to avoid
> sudden severe crashes).
> 3) Set the minimum size equal to the maximum size (to avoid
> fragmentation).
> 4) If a second harddrive is available, place the swapfile on a partition
> there (so if the swapfile is used, the heads on the first HDD doesn't have
> to jump back and forth between the swapfile and where it is otherwise
> working).
>
> I saw the point you made but after testing for a while i'm back at using
> settings based on the 4 points made above, because two things bugged me.
>
> I experienced just what PCR was mentioning, that the swap-file started out
> bigger then the set minimum and then changed (shrinking) over time, i.e.
> it didn't stay a fixed size like i wanted it to.
>
> If setting a fixed minimum, you are also forced to set a maximum, there is
> no way to tell it to use maximum available on the partition. *You have to
> give it a value in MegaBytes.* So after having put other files on that
> partition and opening the Virtual Memory settings page again, one can
> notice that Windows have noticed that there is less space available and
> adjusted the set maximum down to what is currently available. Taking files
> off that partition and looking again one can see Windows setting the
> max-setting higher again, but it doesn't set it higher then what was
> origianlly set.
>
> (So the OS kind of changes both the minimum and maximum settings on its
> own accord.)
>
> The solution to both these annoyances, and maybe you agree the optimal
> setting, and the case where your suggestion and the 4-part advice given
> above coincide, would be to put the swapfile on a partition (say 1 GB) on
> its own, set the minimum size high, as high as the maximum, which in turn
> then is set as big as the partitionsize. Hmm, sounds pretty analougous to
> a Linux Swap-partition. The next step would be to make it not mount, so
> you wouldn't see it in 'My Comuter' and thus wouldn't muck around with it
> by misstake.
>
> I have no such dedicated partition currently, maybe i will after i
> reformat my drives next time, but i've returned to using Swapfile on a not
> often used partition on the second drive with minimum = maximum. But i
> have upp'ed the size from 256 to 384 .. just in case. ;-)
>
> --
> Please followup in newsgroup.
> E-mail address is invalid due to spam-control.

PCR

July 30th 04, 01:35 AM

#4 is probably best, but some say it ought go to the first primary
partition of a second drive. Also, if one begins to use other partitions
on that second drive, then it's heads naturally suffer, just as they
would were all partitions on the first drive.

HOWEVER, with enough RAM (I have 376 MB usable to Windows), the Swap
File rarely gets a "Use" figure in System Monitor. In that case, it
hardly matters where the Swap File is located so far as head travel,
that I can think of. And setting a high enough Min figure prevents
resizing.

In that circumstance, I think isolating the Swap File on the C:partition
so that it is out of the way of all other system files makes for a
better Defrag. Other files will not have to pussyfoot around Swap File
bits/pieces as they are optimized.

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"Richard G. Harper" > wrote in message
...
| #1 is an ignorant statement with no basis in fact. #2 is okay, #3 is
| reasonably okay if you don't mind tying up as much as 2gb of hard
drive
| space, #4 is bang-on correct as long as you remember the words "on a
second
| physical hard drive". If all you do is create two partitions on a
single
| drive and put the swapfile on the second partition you're already
fighting
| against yourself.
|
| --
| Richard G. Harper [MVP Win9x]
| * PLEASE post all messages and replies in the newsgroups
| * for the benefit of all. Private mail is usually not replied to.
| * HELP us help YOU ... http://www.dts-l.org/goodpost.htm
|
|
| "... et al." > wrote in message
| ...
| > Richard G. Harper wrote:
| >
| >> You can set the minimum size to be as big as you want, and if set
| >> sufficiently high the swapfile should never feel the need to
resize.
| >> Thus you avoid fragmentation without any of the down side issues.
| >>
| >> I think the only reason any contrarian discussion continues on this
point
| >> is because some web sites and/or "experts" that like to discuss
swapfile
| >> optimization have failed to keep up with the state of the art in
| >> computers. When Windows 98 came out computers came with very small
hard
| >> drives and comparatively little memory, and upgrading either was
| >> horrendously expensive. Today both memory and hard drives are
| >> commodity-priced and there's really no reason to stint on either.
| >>
| >
| > Repeating the advice i previously followed regarding Virtual memory:
| > 1) Don't let Windows manage your virtual memory (it does it poorly
| > (¿possibly by not doing #3 & #4?)).
| > 2) Set the maximum size large enough so the limit isn't reached (to
avoid
| > sudden severe crashes).
| > 3) Set the minimum size equal to the maximum size (to avoid
| > fragmentation).
| > 4) If a second harddrive is available, place the swapfile on a
partition
| > there (so if the swapfile is used, the heads on the first HDD
doesn't have
| > to jump back and forth between the swapfile and where it is
otherwise
| > working).
| >
| > I saw the point you made but after testing for a while i'm back at
using
| > settings based on the 4 points made above, because two things bugged
me.
| >
| > I experienced just what PCR was mentioning, that the swap-file
started out
| > bigger then the set minimum and then changed (shrinking) over time,
i.e.
| > it didn't stay a fixed size like i wanted it to.
| >
| > If setting a fixed minimum, you are also forced to set a maximum,
there is
| > no way to tell it to use maximum available on the partition. *You
have to
| > give it a value in MegaBytes.* So after having put other files on
that
| > partition and opening the Virtual Memory settings page again, one
can
| > notice that Windows have noticed that there is less space available
and
| > adjusted the set maximum down to what is currently available. Taking
files
| > off that partition and looking again one can see Windows setting the
| > max-setting higher again, but it doesn't set it higher then what was
| > origianlly set.
| >
| > (So the OS kind of changes both the minimum and maximum settings on
its
| > own accord.)
| >
| > The solution to both these annoyances, and maybe you agree the
optimal
| > setting, and the case where your suggestion and the 4-part advice
given
| > above coincide, would be to put the swapfile on a partition (say 1
GB) on
| > its own, set the minimum size high, as high as the maximum, which in
turn
| > then is set as big as the partitionsize. Hmm, sounds pretty
analougous to
| > a Linux Swap-partition. The next step would be to make it not mount,
so
| > you wouldn't see it in 'My Comuter' and thus wouldn't muck around
with it
| > by misstake.
| >
| > I have no such dedicated partition currently, maybe i will after i
| > reformat my drives next time, but i've returned to using Swapfile on
a not
| > often used partition on the second drive with minimum = maximum. But
i
| > have upp'ed the size from 256 to 384 .. just in case. ;-)
| >
| > --
| > Please followup in newsgroup.
| > E-mail address is invalid due to spam-control.
|
|

Ron Martell

July 31st 04, 08:09 PM

"... et al." > wrote:

>Repeating the advice i previously followed regarding Virtual memory:
>1) Don't let Windows manage your virtual memory (it does it poorly
>(¿possibly by not doing #3 & #4?)).

That is totally wrong.

>2) Set the maximum size large enough so the limit isn't reached (to
>avoid sudden severe crashes).

Never repeat never specify a maximum size limit for Virtual Memory in
Windows 9x. There is no benefit that can ever be achieved by doing
so. The only possible outcomes, in order of decreasing probability
and increasing severity, are:
- Reduced performance as your system reduces disk cache in order to
meet the total memory needs of your loaded applications and data
files.
- Applications refusing to load due to "insufficient memory" errors.
- Applications crashing due to "out of memory" errors resulting in
loss of data and possible corruption of data files.
- Total system lockups or crashes due to "out of memory" errors
resulting in loss of data and possible file corruption.

All of the benefits purportedly achieved by having a fixed or
permanent swap file can in fact be obtained by specifying a minimum
size only.

>3) Set the minimum size equal to the maximum size (to avoid fragmentation).

Swap file fragmentation in Windows 95/98/Me is a total non-issue. Any
purported performance impacts are totally mythical.

The fundamental fact is that Windows 95/98/Me handles each and every
4K memory page written to or read from the swap file as a separate I/O
transaction. So even if two consecutive memory pages are being
written to or read back from the swap file this will still result in
two separate disk read operations, even if they physically adjacent to
one another on the hard drive.

>4) If a second harddrive is available, place the swapfile on a partition
>there (so if the swapfile is used, the heads on the first HDD doesn't
>have to jump back and forth between the swapfile and where it is
>otherwise working).

That does make sense, provided the hard drive is not in active use for
other purposes. Assume, for example, that programs are stored on the
second hard drive, and one of these programs is being loaded. In
order to load this new program it is necessary to for Windows to move
some currently inactive items from RAM to the swap file so as to
provide the RAM needed for the new program. Both the swap file and
the program being loaded are on the second hard drive. Therefore in
this specific circumstance there will be a lot of head movement back
and forth between the program location and the swap file location.

>

The ultimate answer for all performance related issues regarding the
swap file is to have sufficient RAM so as to totally eliminate the
need for Windows to actually use the swap file at all.

Note that this does not eliminate the need for the swap file
completely, as Windows will still use space in the swap file to
satisfy the memory address requirements for the unused portions of
memory allocation requests.

Ron Martell Duncan B.C. Canada
--
Microsoft MVP
On-Line Help Computer Service
http://onlinehelp.bc.ca

"The reason computer chips are so small is computers don't eat much."

PCR

August 1st 04, 12:08 AM

You're making me start to worry all over again!

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"Ron Martell" > wrote in message
...
| "... et al." > wrote:
|
|
| >Repeating the advice i previously followed regarding Virtual memory:
| >1) Don't let Windows manage your virtual memory (it does it poorly
| >(¿possibly by not doing #3 & #4?)).
|
| That is totally wrong.
|
|
| >2) Set the maximum size large enough so the limit isn't reached (to
| >avoid sudden severe crashes).
|
| Never repeat never specify a maximum size limit for Virtual Memory in
| Windows 9x. There is no benefit that can ever be achieved by doing
| so. The only possible outcomes, in order of decreasing probability
| and increasing severity, are:
| - Reduced performance as your system reduces disk cache in order to
| meet the total memory needs of your loaded applications and data
| files.
| - Applications refusing to load due to "insufficient memory" errors.
| - Applications crashing due to "out of memory" errors resulting in
| loss of data and possible corruption of data files.
| - Total system lockups or crashes due to "out of memory" errors
| resulting in loss of data and possible file corruption.
|
| All of the benefits purportedly achieved by having a fixed or
| permanent swap file can in fact be obtained by specifying a minimum
| size only.
|
|
|
| >3) Set the minimum size equal to the maximum size (to avoid
fragmentation).
|
|
| Swap file fragmentation in Windows 95/98/Me is a total non-issue. Any
| purported performance impacts are totally mythical.
|
| The fundamental fact is that Windows 95/98/Me handles each and every
| 4K memory page written to or read from the swap file as a separate I/O
| transaction. So even if two consecutive memory pages are being
| written to or read back from the swap file this will still result in
| two separate disk read operations, even if they physically adjacent to
| one another on the hard drive.
|
|
| >4) If a second harddrive is available, place the swapfile on a
partition
| >there (so if the swapfile is used, the heads on the first HDD doesn't
| >have to jump back and forth between the swapfile and where it is
| >otherwise working).
|
| That does make sense, provided the hard drive is not in active use for
| other purposes. Assume, for example, that programs are stored on the
| second hard drive, and one of these programs is being loaded. In
| order to load this new program it is necessary to for Windows to move
| some currently inactive items from RAM to the swap file so as to
| provide the RAM needed for the new program. Both the swap file and
| the program being loaded are on the second hard drive. Therefore in
| this specific circumstance there will be a lot of head movement back
| and forth between the program location and the swap file location.
|
| >
|
| The ultimate answer for all performance related issues regarding the
| swap file is to have sufficient RAM so as to totally eliminate the
| need for Windows to actually use the swap file at all.
|
| Note that this does not eliminate the need for the swap file
| completely, as Windows will still use space in the swap file to
| satisfy the memory address requirements for the unused portions of
| memory allocation requests.
|
|
| Ron Martell Duncan B.C. Canada
| --
| Microsoft MVP
| On-Line Help Computer Service
| http://onlinehelp.bc.ca
|
| "The reason computer chips are so small is computers don't eat much."

... et al.

August 6th 04, 06:37 PM

Ron Martell wrote:

> "... et al." > wrote:
>
>>Repeating the advice i previously followed regarding Virtual memory:
>>1) Don't let Windows manage your virtual memory (it does it poorly
>>(¿possibly by not doing #3 & #4?)).
>
>
> That is totally wrong.
>

Well, it's not really an advice on *how to* set your virtual memory
settings, more a what not to do. I never seen any real qualification for
the argument, hence my guess given in parenthasis above. But it's really
more of a preamble to the then 3 pieces of advice on *how to* adjust the
settings.

>
>>2) Set the maximum size large enough so the limit isn't reached (to
>>avoid sudden severe crashes).
>
>
> Never repeat never specify a maximum size limit for Virtual Memory in
> Windows 9x. There is no benefit that can ever be achieved by doing
> so. The only possible outcomes, in order of decreasing probability
> and increasing severity, are:
> - Reduced performance as your system reduces disk cache in order to
> meet the total memory needs of your loaded applications and data
> files.
> - Applications refusing to load due to "insufficient memory" errors.
> - Applications crashing due to "out of memory" errors resulting in
> loss of data and possible corruption of data files.
> - Total system lockups or crashes due to "out of memory" errors
> resulting in loss of data and possible file corruption.
>
> All of the benefits purportedly achieved by having a fixed or
> permanent swap file can in fact be obtained by specifying a minimum
> size only.
>

I know you read my last post, but .. one of the two gripes i had after
understanding Richard's argument was that i didn't see any possibility
to avoid setting a number-value for a maximum *if* you had set one for
the minimum.
Repeating, if i set a minimum size limit, then i can't see how to avoid
setting a maximum limit as well.

>
> The ultimate answer for all performance related issues regarding the
> swap file is to have sufficient RAM so as to totally eliminate the
> need for Windows to actually use the swap file at all.
>

No arguments there :-)
Though .. Win98 will apparently run into trouble for other reasons when
you exceed 768 MB, 1 GB, 1.5 GB ... something.
Adding MaxFileCache=524288 in the [VCache] section in System.ini helps
raising the limit a bit.

> Note that this does not eliminate the need for the swap file
> completely, as Windows will still use space in the swap file to
> satisfy the memory address requirements for the unused portions of
> memory allocation requests.
>

ahh.., ok.

--
Please followup in newsgroup.
E-mail address is invalid due to spam-control.

Ron Martell

August 6th 04, 09:15 PM

"... et al." > wrote:

>
>
>I know you read my last post, but .. one of the two gripes i had after
>understanding Richard's argument was that i didn't see any possibility
>to avoid setting a number-value for a maximum *if* you had set one for
>the minimum.
>Repeating, if i set a minimum size limit, then i can't see how to avoid
>setting a maximum limit as well.
>

If you leave the maximum setting alone when entering a minimum then it
will default to the same value as the amount of free space on the
drive.

If you then exit the virtual memory window, reboot the computer, and
go back in to the virtual memory window the status will now be set to
"let windows manage..." However if you look closely at the
grayed-out values you will see that your specified minimum is there
and the maximum reads "no maximum".

If you click on the "let me specify.." button then the values show
clearly and the maximum will be the same as the free space available
on the drive.

Ron Martell Duncan B.C. Canada
--
Microsoft MVP
On-Line Help Computer Service
http://onlinehelp.bc.ca

"The reason computer chips are so small is computers don't eat much."

... et al.

August 6th 04, 10:53 PM

Ron Martell wrote:

> "... et al." > wrote:
>
>>I know you read my last post, but .. one of the two gripes i had after
>>understanding Richard's argument was that i didn't see any possibility
>>to avoid setting a number-value for a maximum *if* you had set one for
>>the minimum.
>>Repeating, if i set a minimum size limit, then i can't see how to avoid
>>setting a maximum limit as well.
>>
>
> If you leave the maximum setting alone when entering a minimum then it
> will default to the same value as the amount of free space on the
> drive.
>
> If you then exit the virtual memory window, reboot the computer, and
> go back in to the virtual memory window the status will now be set to
> "let windows manage..." However if you look closely at the
> grayed-out values you will see that your specified minimum is there
> and the maximum reads "no maximum".
>
> If you click on the "let me specify.." button then the values show
> clearly and the maximum will be the same as the free space available
> on the drive.
>

Not really selfexplanatory, eh? Thanks for clarifying.

So if that is the case there is only one gripe left. And that, that the
size of the Win386.swp file start up larger then the specified minimum
directly after bootup and then shrinks back to what you specified, can
most probably be done away with by setting the minimum high "enough". I
wonder if there is any way to know what Windows demands as "enough" for
the minimum to never start out higher then that. Is it a fixed MB value
or does it depent on the amount of real RAM you have in the machine
and/or some other things? Do you have to find that out with a trial and
error approach?

--
Please followup in newsgroup.
E-mail address is invalid due to spam-control.

PCR

August 7th 04, 01:21 AM

Do you have "System Monitor" in START... System Tools? If not, get
it from "START, Settings, Control Panel, Add/Remove Programs, Windows
Setup tab, D-Clk System Tools, check System Monitor, OK, Apply, OK". May
as well take "Resource Meter", too. Now, go through the menus and at
least have it display

(a) Swap file in use.
(b) Swap file size.
(c) Swappable memory.
(d) Unused physical memory.
(e) Allocated memory.
(f) Disk cache size.
(g) Locked memory
(h) Other memory
(i) Kernel Processor Usage
(j) Kernel Threads

But don't ask me what thry all mean. JUST watch the swap file over a
period of days to see how large it normally wishes to grow. Then, set
the Min to 20% above it.

--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
should things get worse after this,
PCR

"... et al." > wrote in message
...
| Ron Martell wrote:
|
| > "... et al." > wrote:
| >
| >>I know you read my last post, but .. one of the two gripes i had
after
| >>understanding Richard's argument was that i didn't see any
possibility
| >>to avoid setting a number-value for a maximum *if* you had set one
for
| >>the minimum.
| >>Repeating, if i set a minimum size limit, then i can't see how to
avoid
| >>setting a maximum limit as well.
| >>
| >
| > If you leave the maximum setting alone when entering a minimum then
it
| > will default to the same value as the amount of free space on the
| > drive.
| >
| > If you then exit the virtual memory window, reboot the computer, and
| > go back in to the virtual memory window the status will now be set
to
| > "let windows manage..." However if you look closely at the
| > grayed-out values you will see that your specified minimum is there
| > and the maximum reads "no maximum".
| >
| > If you click on the "let me specify.." button then the values show
| > clearly and the maximum will be the same as the free space available
| > on the drive.
| >
|
| Not really selfexplanatory, eh? Thanks for clarifying.
|
| So if that is the case there is only one gripe left. And that, that
the
| size of the Win386.swp file start up larger then the specified minimum
| directly after bootup and then shrinks back to what you specified, can
| most probably be done away with by setting the minimum high "enough".
I
| wonder if there is any way to know what Windows demands as "enough"
for
| the minimum to never start out higher then that. Is it a fixed MB
value
| or does it depent on the amount of real RAM you have in the machine
| and/or some other things? Do you have to find that out with a trial
and
| error approach?
|
| --
| Please followup in newsgroup.
| E-mail address is invalid due to spam-control.