Pagefile & Ram Guide For Nt/xp

[Big Booger]

http://www.arstechnica.com/tweak/nt/pagefile-1.html

an excerpt:
Windows NT RAM & Pagefile Optimization
by Cæsar

Unlike the MacOS (for example), NT was designed to use virtual memory without regard for the amount of RAM in your box. Even if you have a 768MB of RAM, NT still wants a pagefile, and NT will still use it. It’s for this reason that knowing how to optimize your overall system memory configuration is so important. It doesn’t matter if you’ve got 128 or even 256MB of RAM – if your pagefile isn’t in tip-top shape, you’re taking a performance hit, and that extra RAM you paid for isn't truly paying off. But how much RAM do you need in the first place? Well, it depends - there's no real simple formula, other than to say that if you have less than 128MB of RAM, you're suffering.

Excellent article/guide on tweaking the pagefile and settings for memory. I found it very interesting and informative and I hope you do too. It says for NT, but I think that applies to XP since it is built on the NT technology.
BB

[tons of fun]

Very useful....thank you!!!

[Big Booger]

no problem.

One word of advice, write down the default values just in case something goes wrong, or just change it back to let windows manage it.
BB

[tons of fun]

Thanks....

[lynchknot]

Mr.BB, I don't understand this article too well, so I wanted to check with you. I set the max pagefile size to recommended as I found it set at 768 - 1500. I understood that the initial size and the max should match. Judging from my commit charge peak, is this still too excessive? I ran NAV, Adaware, photoshop, defrag, and DL a file all at the same time. In the SS it peaked at 382k but with a 1280k limit. Why does it not reflect my 768k max setting?

Screenshot here (http://www.boomspeed.com/lynchknot1/t1.jpg)

[lynchknot]

Booger?

[Big Booger]

I think pagefile is used after your ram is used, so if you have 256MB of ram, and 350 MB of pagefile, and the programs you are using don't require all of it, you more than likely will never hit the peak.
For example, you run 5 programs totalling 300MB, and you have 512MB of ram, and 256MB page file, it will most likely balance the two so that neither one is overtaxed...
taking a certain percentage from RAM and a certain percentage from pagefile..
not all ram or all page file..
Does that make sense? So you set the page file very high just in case it ever becomes taxed...
But remember page file uses HDD space to make the virtual ram, which by the way is a lot slower than hardware based RAM..

I doubt you will ever use the max pagefile, the most i have ever used was about 600Mb of ram and page file... gaming, while downloading, etc...

hope that helps,
BB

[lynchknot]

er.............so did I set it right? (screenshot above) :)

[Big Booger]

Yes it would seem that it is correct,
BB

[Conan]

With 512mb of RAM I set my pagefile to Min-768, Max-768.

[egghead]

i dont use a page file:

is that good?

396mb ram

[Big Booger]

I'd at least go with what MS says egghead:

What Microsoft says
Microsoft wants you to have RAM + 12MB of Virtual Memory available for the system, regardless of how much RAM you have (some documents say RAM + 11, but most of those relate back to NT 3.1). The reasoning for this is because Microsoft designed NT to dump the system's memory to the pagefile in the event of a crash (a setting you can turn of). So, you need at least as much pagefile as you have RAM (at least so far as this reasoning goes). The additional 12MB is for overhead in said crash eventuality. So, if you have 128MB of RAM, MS would tell ya to set yourself up with 140MB of virtual memory.

But if you are strapped for HDD space, and you aren't running programs that are going to use more than your available physical ram, then I don't see why it'd be a problem...
I have never ran XP without a pagefile so I don't really know.

I think it would be safer if you had more ram say 512mb-1gb of ram, but if it works, then it works.
BB

[egghead]

thanks big booger:

i dont seem to get crashes but what you posted makes sense

if i crashed and i have no page file

i could loose alot of stuff

as long as my disk stays error free
i'll be glad

but if m$ starts writing $%^&*(&^^%$ everywhere
and writing over directories

i'll give it 2gigs


i'll experiment

[perris]

well, my first post on this forum, and I know this is an old thread, but there's a lot of misinformation that needs to be set straight

this free pagefile deffrag program (http://www.sysinternals.com/ntw2k/freeware/pagedefrag.shtml)...


most of you have the wrong idea about the swap in xp.

xp is a paging os...the only way to get around that is to run dos.

THIS IS VERY IMPORTANT; the page file will always be active, though you will hardly ever use the paging file.

what's happening, is the page file is always ready, and getting ready, and that's the pf activity that you see;

RAM IS ALWAYS USED BEFORE THE HARDRIVE INFORMATION.

NOW, those of you that have dissable expansion...foolhardy.

the pf will only expand when the commit charge reaches the commit limit...never before...doesn't want to doesn't try to.

therefore, if the pf is expanding, it obviously needs to expand you are suffering perfirmance by not allowing expansion.

is expansion desireable?...no...but the way to avoid expansion is to raise your initil minimum, you don't prevent expansion by disabling expansion.

that's the same thing as removing the safety net under the tite rope, because falling into the tite rope is a bad thing...

my recomendation, initial minimum=2x ram, with expansion enabled to 3x that figure.

DO NOT MISSREAD THE ADVICE THAT 1.5 IS THE RECOMENDED VALUE...IT'S NOT!!!

in xp, 1.5 is the minimum allowed AS THE INITIAL MINIMUM... EXPANSION... IS STILL REQUIRED WITH THIS ABSOLUTE MINIMUM SETTING .

[lynchknot]

So you're saying, for me, the standard windows setting is right? Because that is where I found my settings. Very close to 2x-3x. (recommended windows setting)

[lynchknot]

Well that was weird. After viewing this thread, I received a notification from windows stating that my virtual memory was low (!) and that windows was automatically adjusting it for me.
I checked and the settings have not changed.
I don't know if i'm correct, but I have now set it to 900x900. It was set at 768x768. I ran up my PF to peak at 419, with a lot of apps, this time. I never use more than 3-4 apps at a time anyway. I think 900 is a good ballpark - though i'm not fully understanding this.

[perris]

if you have the nt kernal, yes, those are correct settings for most people, however, those of you that use ram intensive programs can even go with a higher initial minimum, there is never a penalty for having a pagefile tooo big, but there is a huge penalty for having a pagefile too small.

also, some nyth has been long told about a dynamic pagefile causing fragmentation.

this is a myth, it does not happen and this is provable by any user.

dissregard any one who tells you a static pagefile will give ypu some benefit.

this is a fact...there is no uiser that can benefit from a static pagefile, and quite a few users will be punished in performance for having one

[perris]

sorry, lynchnot, my post was for the posted ahead of this one, where you said you pf was at the default.

I don't know the amount of ram you're running, but in any event, you'r setting is way too small if you are going to use a static pf..do not set the max and minimum to the same value, this is a huge mistake.

nobody should ever use a static pf.

another urban myth, most people think the more ram you have, the smaller the pagefile.

this is not true, the more ram you have, the bigger the pagefile needs to be, as the pagefile likes to have enough room to assign all the ram you have to a place on the harddrive.

now, the os assigns a place, or an address, but it doesn't ever write to disc unless necceassary, so never fear that harddrive is used before ram, that is never the case.

this address signment is the pagefile activity that many people have mistaken for pagefile use...this activity is not pagefile use, it's just priming the engine, so it's ready to speed your use if you do run out of ram

whenever you get a message that the os is resizing your pf, just raise the initial minimum, but never set both values equal

also, when you do increase the size of your pagefile, you need to then defrag it with a boot defrag, shich I will provide free.

what happens when you personally make the pf bigger, is it will be fragmented, and your regular defrag is not capable of defragging the pf.

when the os makes the pf larger, the extra pf is thrown out on reboot, so obviously, if it was contigous before expansion, it is mpossible for it to be fragmented after expansion...on the next boot
also, lynchnot, it might be a good idea to dissable memory dump if you don;t plan on using it...tha twill usually resolve the issue you've just experienced

[Tinker]

Thanks for the information perris. I have 768 meg of RAM and have set my pf to 1550. I did however set both min and max to this amount. Could you please explain why we do not want to set both the same.

Also I have a P4 so I think I want to use your x86 defrag program, is this correct?

I was doing some large graphics work the other day and ran out of memory using only the RAM and no pf. I enabled the pf and it was set to 1150 by the system. I returned to the graphics program and ran the RAM down to 3 meg and used the pf to render a graphic image that I wanted.

Thanks for the information...:msnstar: :msnstar:

[perris]

ok, yours is a perfect example of why you should never set the pf static.

first, let me make one thing perfectly clear..nt is a paging os, it is impossible to run without a pf...though you may think you dissable the pagefile, all you do is tell the os to page somewhere else..you can prove this by monitoring the activity with perfmon...you will notive MORE paging without a pagefile, not less...true.

now, the minimum and maximum that you suggest would be a setting that MOST people will not suffer a perfirmance hit, but, they also would get no benefit from not allowing a bigger maximum.

you on the other hand, while for most instances, are not suffering performance with the setting you suggest, WILL suffer performance on those occasions where you do graphics, as you are only hardly at a useable setting, and when you load a couple more graphics then you've been, you will be at a loss for vm.

in your case. double the ram is hardly sufficient...just set the maximum to whatever the os allows from there...you see, nothing lost, as the os will only expand when it needs to...so...if you are right, and double ram is sufficient, then the pagefile will remain static anyway...follow?

IF YOUR PAGEFILE HAS THE CORRECT INITIAL MINIMUM, IT IS STATIC, EVEN THOUGH YOU HAVE ALLOWED EXPANSION

you see, I'll say this again;

THE PAGE FILE WILL NEVER EXPAND UNLESS IT NEEDS TO EXPAND..doesn't want to, doesn't try to.

in techno speak, the pagefile only expands when the commit charge approaches the commit limit

therefore, you accomplish abslutely nothing if you never need a bigger pf due to your use.

and therefore again, you might as well leave room for the os to grow the pf if the ocomputing occasion arises

[perris]

here's a perfect analogy of how the pagefile works...very few people understand wht the pf does, and why it's always active, even when you have tons of ram...first, let me tell you, the pf is neccessary for lkernal stability, and it is the very reason the nt kernal is the most stable of the windows kernals...but that's yet another function of the pf that I won;t get into...for now, this is how the pagefile works for your computing needs;

This does a good job explaining the pagefile, for those that enjoy the discussion of it...

this is pretty close;

Imagine that the pagefile is a big storage facility, with no walls inside the facility...Ram are the workers, and packages are program information

The boss, (the os) likes to give all his workers, (ram) a place to put the packages they're carrying, so they can carry other packages if it becomes necceassary...that's why he hired so many workers, he wouldn't hire say 512 workers if he didn't intend on using them, would he, so he wants them to be able to carry lots of packages(4) at once.

so, while a worker is carrying a package, the boss says, well, if I need him to carry something else, I'll have him put the package he's carrying right over here, ... he thus assigns a spot for all the packages his workers are carrying, but the spot remains empty.

when the boss sees he is out of spots for the packages, he rents more space, say up the block. this is not as efficient, he doesn't want to turn a client down, does he, and he surely doesn't want to keep a client waiting...at least he can keep bussiness moving in this fassion, and absolutely does not have to turn a client down..(this is expansion, and thank goodness for it

now, when he runs out of workers for a task, he says, "well, bill is carrying that package for the longest time, and the package has just not been opened, jeeze, I don't think that package will ever get opened, I'm going to have bill wait in the storage spot I gave him, but I won't have him put the package down,untill he absolutely needs to carry another package...this way, if the customer wants to open bills package, (HARDLY LIKELY),he'll be ready, and I won't have bill waiting in the storage area anymore, (boy, I'm glad I had the forsight to do that!)I'll have to put pete in waiting instead of bill.

you see? it's important for the storage facility to be the right size, so the boss doesn't have to rent space up the block!!!...plus, here's the best part, the boss realizes this inneficient space is not good, so he gets rid of it the very next time he opens the doors for bussiness!!! You see, that;s what some people didn't understand in the past, the ineficient space, was actually extremely efficient fot the moment, and gone on the next day...boom...beutifull!!!

Obviously, it's important for there to be places up the block to rent, just incase the storage facility is too small.(otherwise, the boss will be forced to put those packages in other peoples yards...unbelievably inneficient)

but rest assured, if the day comes that the boss needs to rent space up the block, the next day, the owner, (you), should build a bigger storage facility, so you never have to rent ineficient space...you of course will never get mad at the boss for renting more space when it was necceassary, instead, you should give him a raise, and say, "well done, good job"

there

[lynchknot]

Ok, I have 512MB what should I set both at? It recommends 766 initial. The is the option "system managed size"

[perris]

go with double ram as the initial minimum, allow the maximum to whatever the os allows.

if the os ever tells you it is expanding the pf, then after you are done computing, you must make the initial minimum begger the 2x ram

also, since you are permanently changing the size of the pf, you must defrag the pf with the free program I posted for you on my first post

once you make the pf contigous, it will remain contigous for eternity, unless you again, permanently change the size, and then again, you would have to give it a defrag

[lynchknot]

how do you know "whatever it allows" - do you keep typing #'s in until it won't take? I set the initial at 1024 then...........

[perris]

put in 10 gigs, and the os will tell you it can't allow more then your maximum.

on my box, it's 4096

on your box, it might be more, mine's a laptop...in any event, 4096 is a fine number

[lynchknot]

yes, it's 4096 here also. I DL your program (non alpha version) a couple days ago. Does it run automatically?

[perris]

reboot, now that you've changed the size of the pf, it won't take affect till you reboot.

then, launch the program, and it will tell you how fragmented your pf is, along with some other files it monitors.

follow the gui on the program to defrag 1 tome on the next boot...don't bother setting it to defrag everyboot.

reboot...launch the program again...if you have enough room on your drive fir the defrag, you will get 1 fragment per file,\.

one fragment is no fragnmentation

[lynchknot]

Thanks for your help perris. I started out with 3 frags. I was wondering how long it was going to take to defrag. It was not very long. Thanks again.

[Tinker]

O.K. , I have set my min to 1550 and max to 4096 (using the above mentioned method to find max setting). Lots of disk activity now that I did not have before. We will see if it is better or worse.

Thanks to BB and perris for the help on this issue, it is a hard one to figure out.....

:msnstar: :msnstar: :msnstar:

[Big Booger]

A few more notes and things to consider about the pagefile of XP:


When your computer is running low on RAM and more is needed immediately, Windows uses hard drive space to simulate system RAM. This is known as virtual memory, and is often called the paging file. This is similar to the UNIX swapfile. The default size of the virtual memory paging file (named pagefile.sys) created during installation is 1.5 times the amount of RAM on your computer.

You can optimize virtual memory use by dividing the space between multiple drives and removing it from slow or heavily accessed drives. To best optimize your virtual memory space, divide it among as many physical hard drives as possible. When you select drives, keep the following guidelines in mind:
Try to avoid having a paging file on the same drive as the system files.
Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance, and some fault-tolerant systems suffer from slow data writes because they write data to multiple locations.
Do not place multiple paging files on different partitions on the same physical disk drive.


To have Windows choose the best paging file size, click System managed size. The recommended minimum size is equivalent to 1.5 times the amount of RAM on your system, and 3 times that figure for the maximum size. Example, if you have 256 MB of RAM, the minimum size would be 384, the maximum size would be 1152.
For best performance, do not set the initial size to less than the minimum recommended size under Total paging file size for all drives. The recommended size is equivalent to 1.5 times the amount of RAM on your system. Usually, you should leave the paging file at its recommended size, although you might increase its size if you routinely use programs that require a lot of memory.
To delete a paging file, set both initial size and maximum size to zero, or click No paging file. Microsoft strongly recommends that you do not disable or delete the paging file.

I agree that having too much shouldnt be an issue other than fragmentation.. Thanks Perris for the great information about the PF and for offering that program that helps defragment the PF. Welcome to Techzonez by the way.

Additionally,
If for some reason your PF is all over your several HDDs and you want to move it, MS has a guide available.
http://support.microsoft.com/default.aspx?scid=kb;en-us;Q307886

That article outlines moving of the pagefile within your windows XP operating system
Hope that helps.

One last thing. If you are like me and have Diskeeper 4.0+ defragging software then you can set it to:
Paging File Frag Guard
When Paging File Frag Guard is enabled, Diskeeper will monitor the paging file and attempt to keep it defragmented while the computer is online.

I have this set and have never noticed that my PF is fragmented. If you dont have Diskeeper, I highly recommend it.

BB

[perris]

tinker, once you defragment your pagefile, there really is no way you can have more disc activity with a larger pagefile then a smaller one...that cannot possibly happen,...you will have more dosc activity if your pagefile is too small...I'll be back with that ms documentation.

now, there is a chance you are running a program that requires a pagefile for all the features to be realized, and those features could possibly be turned off if you don't have enough vm...photoshop will not even run without a pf, so in the case of photoshop, hehe..you wll see more hd activity with a big pf, because now the program is running...and without the pf, the program is not running...that's the only thing that can cause more harddrive activity...it's the more efficient use of your programs, which would be the function of the program, not the function of the pagefile.

big booger...this white paper you provide is actually obsolete.

when ms released xp, due to stbility issues with 2000, they raised the reccomended threshold.

in xp, and also all nt kernals, butr specifically for xp, the threshhold is;

NO LESS THEN 1.5 as the initial minimum...no less then, and 1.5 is the initial minimum, meaning expansion is enabled at theis absolute minimum.

I'll be back with both of these white papers

[perris]

allright, here're the updated recomendations for the nt kernal...notice, ms admonishes never to set less then the recomended 1.5 as the initial minimum, but bigger won't hurt.

most people missinterperate this document to read the recomended is 1.5...that is absolutely not what ms recomends here...here, they recomend no less then 1.5 with expansion enabled



"For best performance, do not set the initial size to less than the minimum recommended size under Total paging file size for all drives . The recommended size is equivalent to 1.5 times the amount of RAM on your system. Usually, you should leave the paging file at its recommended size, although you might increase its size if you routinely use programs that require a lot of memory. "

now, if you read the apragragh, the recomended size stated is the initial size, meaning at this low setting, expansion is neccessary.

This is from Knowledge base article Q308417. HOW TO: Set Performance Options in Windows XP

[perris]

allright tinker, here's the qoute from the ms document for you;

it was originally written for nt...later, microsoft recognized that pagefiles too small were causing great performance problems, and therefore, set the default pagefile bigger in xp...this is very important for everyone, especially those who think a small pagefile is a good idea...xp is by design made to perform with a big pagefile...here is the exact qoute from microsofr, as far as making a pagefile too big, versus too small in the nt kernal;

"...A pagefile that's set too small can lead to overactive disk swapping, or "disk thrashing." The only real drawback with a relatively large swapfile is that you might not have as much disk space available for other uses as you would if you'd followed the pagefile setup recommendations."...

for your referance, document number; Q102020:

[Big Booger]

So basically,
Don't set the page file less than the recommended minimum of 1.5 times your current amount of RAM.
And if you have the space to use for pagefiling, the larger the better, though with a larger pagefile, you will more than likely have to deal with fragmentation.
One last thing perris, I agree that the recommended minimum is not the optimum. So one thing that gets me, in general, when you let the OS manage the pagefile, it has at least for me, always set it to 1.5 times my amount of RAM. Why is that?

Thanks for your information and insight. It is most thought provoking.
BB

[perris]

allright booger, yes, if you ahve plenty of drive space, xp will set the default minimum to 1.5.

if you are short on drivespace, xp will try to get by with a smaller pf.

that's the first answer, now the second.

actually, once you defrag the pagefile, there is no chance for it to get fragmented at all...it doesn't matter how big you make it, and in fact, if it's so big the os will never enlarge it, that's even better.

let me be a little more specific about the pf fragmentation issue.

first, sometimes it's best to leave unimprotant information out, as there are people that twist the information into something that's damaging.

such is true about pagefile information.

here's what actually happens, and why it was missinturperated;

suppose your pf is too small, and the os therfore expands the file.

FOR THIS INSTANCE, AND ONLY FOR THIS INSTANCE the pagefile will be in fragments, because the os had to put an extra part onto it, and the os grabbed whatever area on the hd it could.

so, for this boot, the pf is fragmented, and this is where the myth began.

but, this was shortsighted, for on reboot, the excess pagefile, the part that was added by the os is disscarded.

the orogonal pf remains in the exact physical location, the exact physical size that it was before expansion, so if it was contigous before expansion, it is impossible for it to be fragmented after expansion.

you see, the pagefile is completely differant then most of the other files in the os...the pagefile actually behaves a little like a partition, in that absolutely no other information can enter the area that is set aside for the pagefile...the physical area set aside for the pagefile is exclusive to the pagefile...as if it were a partition...it's not a partition but in this behavior, it's similar


so no, a bigger pagefile actually causes fewer circumstances of fragmented pf use, because fewer times the os will have to expand the pf.

it is important to try to make the pf so large, the os never needs to expand it.

it is also important to defrag the pg once you manually change the size of it.

ok, that's a little wordy, I hope it's not too hard to follow

[Big Booger]

Perris,
That made perfect sense and I am glad you clerified the fragmentation issue. Because nearly every article or post that I have read about the PF always included a bit or two about fragmentation and how the pF gets fragmented.
I myself use tweak Xp to have the Pagefile cleared everytime I reboot or shutdown.
Thanks for your explanations. It seems you know quite a bit about the page file of the NT kernel.
BB

[perris]

big booger..hehe...here's another myth I'm going to cast dispersion on.

the os ignores the old information in the pagefile on every reboot.

in other words, it is already deleted, without you doing anything to it.

the tweak of clearing the pagefile is a security issue.

what this does, on shutdown, it ovewrwrites the pf with zeros, so that the pf can supposedly not be read by a hack.

this is actually a waste of time, since anyone that can get into your pf, can surely overcome a simple overwrite of zeros.

the only way to secure a file is to over write 9 x's, with various randome sequenses. one time 1's, one time 0's, one time random, and that sequence 3x's is the only time your info is safe.

in other words, don't waste your time with this tweak

[Big Booger]

well again I am baffled. I always thought the rewrite was like an optimization to help improve the speed or reliabilty. I never knew it was a security issue and now that I have this new bit of information I am going to put an end to that "tweak". You have been most informative.
BB

[perris]

ya...there is one instance when the information in the pf is not ignored by the os.

if you have a crash, and therefore a memory dump, this dump is marked as such, and it is not ignored on the first sequential reboot.

if you don't ever intend to use the memory dump for debugging, you amy as well turn memory dump off, though you won't get a benefit from doing it, it won't hurt less you plan on using the feature

[shadow_warez]

what ify ou have a fast hard drive, and a giog of fast memory? can you disable the page file? or should i keep the page file intact,
i was wondering i can disabl e my page fille and still work in Photo shop 7,

[Tinker]

You can do that. I have tried it both ways and with a fast HDD you really do not see improvement not using PF. One of the major issues I see on my system is over time the system becomes unstable. This time period would be 6 to 8 weeks. It gets to the point that I have to do a clean install of XP to get a stable system again. Now I can not say that this is the total cause because there are many things that could cause this to happen (system being unstable I mean).

I am using a Maxtor HDD 80 gig with 8 meg cache and have 768 meg RAMBUS. Not using the PF really does not give me a faster system for my use......

As well one of the things that has been pointed out is that a NT based system uses PF as part of its SOP unlike 98 or ME that use it in laue of sufficient RAM......

:D

[shadow_warez]

so if i get a srial ata drive, and i run my paging at say 2000-4096 then it should in theroy have faster access times,?

[Tinker]

Well the thing to remember here is we are talking about access time, not sustained time. If you have a lot of data such as say a large graphic file that you are trying to work with and you were able to keep it all in RAM then the sustained time would be better from the RAM rather than a HDD.. Trying to work faster from a HDD rather than RAM is not going to happen, or at least not as far as I know.

Quote from shadow_warez:

so if i get a serial ata drive, and i run my paging at say 2000-4096 then it should in theory have faster access times,?

End quote........

It would be faster than from a IDE drive yes, faster than from memory? That would depend on your memory speed and of course the speed of the HDD, etc.

I was out looking at the 128 MB USB 2 memory cards today and they should be something close to 430 MB transfer all the time. This would be a good way to go I think, if you have USB 2. The cost of the plug n play card was $130 US. I have used my USB 1.1 memory card reader and it works but is slower than a HDD. The USB 1.1 standard is only 12 MB where as USB 2 is 430 MB... What do you all think about this idea?

:D

[Big Booger]

I wonder if you could create a permanent RAM drive and use it as your PF...

I know RAM drives are faster than normal hard drive space.

More about Ramdrives (http://cyberwizardpit.net/article3.htm)

I don't know if this is possible but it might be... I have no idea how to set it up, or anything.

And seeing as the RAM drive contents are deleted on shutdown, surely this would help deter the need to defrag the PF.


BB

[Tinker]

Seeing that any information that is placed in a RAMdrive is lost on reboot this may not be a good idea since XP uses the same space where the page file is located to store important information about the crash. As well any data that has been stored there will be lost so you would not want to put any important files on the RAM Drive. IMHO.........:)

** More....... ** (http://user.netomia.com/darkfaq/ramdisk.htm)

[Big Booger]

Perhaps you are right Tinker. It was just a thought. :D I am sure if it were possible, it would have already been done by now.

BB

[perris]

Originally posted by shadow_warez
what ify ou have a fast hard drive, and a giog of fast memory? can you disable the page file? or should i keep the page file intact,
i was wondering i can disabl e my page fille and still work in Photo shop 7,

it is impossible to turn of the pagefile in xp

you think you have dissabled it by setting it to zero, but just go to perfmon while it's set to 0, and you will see more pf useage, not less...xp is a vm os...there is nothing you can do about this.

it's the vm strategy that makes the nt kernal so fast and stable.

without vm, you would need more then 2 gigs of ram, as this is about what is allocated to memory (the os assigns an address in thepf for code that is not likely needed.)...for instance view souce in ie...this feature is not likely to be used by most users, but the feature is in code...so xp assigns an address for the few people that will use the feature...then if the feature is accessed, it remains in ram, and a differant feature is addressed to the pf.

efforts to dissable vm in xp will slow the os down, and create an unstable envirnment...and, you will not prevent paging one bit in the effort

as far as photoshop...this is one of the programs that write backup files directly to the pagefile, instead of waiting to let xp write to it.

if there were no pagefile, then photoshop would be writing the same files to backup.

this is all the pf is...a contigous envirnment for backing up information you are not likely to be using

I wonder if you could create a permanent RAM drive and use it as your PF...

it's absolutely possible...there are free ram discs from microsoft, as well as after market ram discs from second party comapanies...this idea has been experimented with by ms to no avail, and the notion was discarded as a perfomance liability.

here's why;

the ram disc will take the ram out of memory use whatever ram you allocate to the ram disc

the os will obviously allow smaller working sets for all your apps since it has less ram to work with. (if you didn't know this, the more ram you have, the bigger the working sets allocated to your useage...more features are available in any given program to people that have more ram...wherefore, if you have a gig of ram, you are using more ram then I am with 512)

also, you will obviously need to access the pagefile sooner, since you have less available ram for use then you would have without the ram disc

you are using ram to store ram information...think about that...pretty rediculous.

now, there are people that have found benefit to a ram disc, though as a whole the os will slow down.

you see, when you minimize a window, xp assumes you will not be coming back to that window for a while.

it will agressively release the working sets of minimized windows, and when you maximize, say a half hour later, the window will have to be paged in.

a ram disc will circumvent that strategy, as pagining back in will come from the ram disc.

so, some people like this;

however, the proper way to handle this strategy is to simply not minimize windows that you want the ram information to remain resisident.

as long as you don't minimize, the working set is not aggressively relesed.

this is a tool that most people do not realize is available to xp

you can actually release the ram of programs that you want to leave open, but don't think you'll access in a while, simply by minimizing the program.

[Big Booger]

Perris,
Thanks for clearing that up. I just have one thing I need to clear up. Are you saying that the more RAM you have the better performance gain you would get from using a RAM Disc as your PF?

Just wondering.
BB

[perris]

no

there is no performance gain with a ram disc, as you are just trading ram that you would be using, and forcing it to do nothing...letting it sit doing nothing, waiting for the pf to access it...you see the os never accesses the pf until you are out of ram...so what's the gain here?

what I am saying is that there are some people that think the trade is worth it.

the trade would be a slower computer over all for your down side.

the plus side would be windows that are minimized would open more swift...as ram in minimized windows is deliberate and aggressively released by xp for other use until you put focus back on the minimized window

but, if you want ram to stay in residence, jsut use the windows key+d, (desktop)..

or open a window on top of the window you're working, or any thing but minimize.

so in just about every case, there is no user that will benefit from a ram disc.

I'll tell you another scenario where you might like a ram disc.

if you are one of the people that clear your temproary intenet files every time you close ie, then a small, say 50mb ram disc will speed your intenet expeerience.

now, you need to be the kind of person that is on broadband, have a minisule temp file, and clear the file every ie close for you to notice the benefit.

but for those like me that like to keep temporary internet file relatively large, (500mbs to 1 gig), then the ram disc in this case will slow your internet experience as well, for every boot, the ram disc would need to be repopulated.

to me, here again this is counterproductive.

but again, if your temp file is about 50mbs, and you like to clear the temp file every close, then in this case, (only for those with more then 512 mbs), you might spare 50 mbs to a ram disc for the temp files

[Big Booger]

Perris,
Again thanks. An additional question. What other uses of a RAM disk do you think would benefit a windows XP user?
Just wondering about your thoughts on it.

My thoughts about a RAMdisk and a PF, would be that since RAM is faster, using a RAMdisk would, when needed make the PF faster. YOu have cleared that up. It wouldn't because the PF only comes into play, when the system runs out of RAM.

Wouldn't it be nice to make a HDD partition act like ram, or perform at speeds near to RAM.. would be nice :D
BB

[perris]

here's my prediction bb

I believe longhorn will not be the same big step forward that nt was.

I believe the next great improvement in an os will be a strategy that uses ram for everything but migration on boot.

for instance, the files will be compressed and put to mechanical memory when you shut down, and will be written entirely to soft memory on boot.

I believe it will begin about 50/50

50% mechanical memory, 50% soft memory.

then every couple of years, ms will increase the percentage, untill the os is entirely soft memory.

this will mean a differant type of soft memory that can maintain with reliability information that is written to it, even when it is shut down.

anyway, in essence, this is xp allready does.

xp tries to put as much of your ram into use as it can.

the addage is "free ram is wasted ram"

a person with a gig, who is running the same programs as I am is actually using more ram then I am with 512.

512 though is the point of deminishing returns with xp though, so a gig is usually not noticed over 512 untill you load that much more to access.

so, even with xp, if ms decided to let it use all additional ram with the same strategy it uses up to 512, then in essence, you would have a ram based os.

god...I go on, don't I?

did any of this make sence?

[Big Booger]

made sense to me. I look forward to the future of personal computing. Thanks for your insight.

Here is a little information I dug up from the CENATEK website:

11. Can I install the Windows swap file on a RamDisk 9x/Me drive?
The short answer is yes, but we really don't recommend this. It is unlikely to really improve performance. Taking RAM away from the system to create a RAM disk only increases the frequency of swapping, so the likely effect is a net decrease in performance as the system becomes loaded. For lightly loaded systems, however, it may make the system seem "snappier" to the user since Windows tends to go to the swap file much earlier that it probably should.

Undaunted? Here are some basic instructions on how to set this up:


1. Install RamDisk 9x/Me and configure it for the desired size. Set the RAM disk to save an image on shutdown and load the same image on exit. Do not select FAT32 format.
2. Set the Recycle Bin properties so that the Recycle Bin is turned off for the RAM disk.
3. Edit the following entries in the [386Enh] section of the System.ini file:
PagingFile=r:\win386.swp PagingDrive=r:

where r: is the drive letter of the RAM disk.

Alternatively, you can use the "Virtual Memory" settings button, which you can access from the performance tab of the System applet in the Control Panel.

4. Reboot and enjoy.

When using the RamDisk 9x/Me drive for the swap file you may see message in the Performace tab in the System Control Panel application that states "Compatibility mode paging reduces overall system performance." This is related to how RamDisk 9x/Me suports memory-mapped files. The performance loss suffered in compatability mode should be more than offset by the increased paging performance. This issue, along with FAT32 support for the paging file, will be addressed in a future release of RamDisk 9x/Me.

[perris]

me is a differant animal then xp

I believe me does not use the swap properly, and maybe a ram disc might help, but I don't know

I am an expert on the nt kernal, and not anything else from microsoft

in other words, I can't help one bit in milenium

[Big Booger]

I think anything you can try to improve the performance of ME is a god send.
hehehe
BB

[lynchknot]

I ran across this and thought of this thread
10. Use FAT32 for the Paging File

Finally, if you have a second physical disk in your machine, you can boost performance by moving your paging file (pagefile.sys) onto your second drive. To make this work best, do the following:

1. Create a volume on your second drive, making sure the volume is big enough to hold your paging file. (Three times your RAM amount will be more than enough.)
2. Format the new volume using FAT32 instead of NTFS, since FAT32 gives slightly better read performance on smaller volumes.
3. Don't create any additional volumes on your second drive--that is, leave this drive for exclusive use by the paging file.

So in other words, our final NTFS tweak is to not use NTFS for your paging volume.