Putting Paging file on it's own partition

[trog100]

nope dosnt seem to be there..

i kinda like a story i once read.. the explanation for how radio theory worked.. the "wave" theory.. just like waves moving thru water it went..

some clever bugger after a while said waves move thru a medium.. like water or air.. how does it work in space..????

hmm said the scientists.. after much scratching of heads they invented a none existing medium called the "aether".. this is the medium said the scientists.. it must exists cos our theory says it needs to.. we cant have the none existence of a medium buggering up our nice "wave" theory..

he he he

trog

 

[Alec§taar]

Well, we are "lost in the ether" then...

(We are unable to make a determination of where it is you are paging to, period it seems @ this point)



* This IS going to bother me for some time... I don't like being "dumbfounded/stumped" @ all...

APK

P.S.=> Sorry I was unable to narrow this down for you... it happens! IF you do somehow find this out? Please, let me know, or point me to forums where you DO discover this... thanks! apk

 

[trog100]

i will check around for further info alec.. i dont hold much hope.. i think u know my conclusions which i arrived at thru my own observations.. i have/had been useing the little minimim 200 size pagefile just to go along with the common theory..

giving u more respect then i do most when u told me i was wrong.. its not uncommon when u go against commonly held beliefs.. i have run all the tests we can think of and posted the results.. so far nothing we have done has disproved my original conclusions.. that given enough physical memory windows dosnt seem to need or use the disk based variety..

i still accept the possibility of it needing disk based memory but all the tests i have run tend to suggest otherwise.. i also favour the idea that it simply pages to its physical memory..

it leaves me with a bit of a problem also when i poke my nose into pagfile threads and tell folks what they have read is all outdated rubbish based on machines of yesterday that rarely if ever had enough of the real stuff.. and dosnt really apply to a machine belonging to the cheap ram era..

but threads like this are where the real information comes from.. i hope this one at least makes people wonder about windows and how it uses its memory..

trog

 

[Alec§taar]

i will check around for further info alec.. i dont hold much hope.. i think u know my conclusions which i arrived at thru my own observations.. i have/had been useing the little minimim 200 size pagefile just to go along with the common theory..

I have @ this point, & others THINK as I did, that Windows creates tiny random.tmp files on the disk, someplace (in keeping, sort of, with my idea of the %TEMP%/%TMP% being the area this is done in, possibly TOO fast for us to see them used, & then deleted).

They may have been in the "root" (\.) of C: in fact, where the normal pagefile.sys forms, we never looked there, but they may occur TOO FAST for us to see (used/deleted cycle) as well, we don't know either way.

BUT, there was no proof there either for your idea or mine so far when others thought as I did (small tmp files on disk for paging)... it was on a forums like this one, & they didn't even ATTEMPT to analyze it as you & I had.

giving u more respect then i do most when u told me i was wrong.. its not uncommon when u go against commonly held beliefs..

BUT, you haven't proven yourself RIGHT, either... & neither have I, in your case @ least!

I have shown you this much though: You ARE paging...

& when OS of the VM utilizing design page? They use disks.

Now, afaik? They do NOT page "ram-to-ram", because that defeats the purposes of Virtual Memory period.

(& about going against commonly held beliefs? I do that quite a bit, were you to search my name online... I've gotten into some HUGE "battles" in that regard, lol, over time online).

Still, when it comes to "questioning the horses mouth" (Microsoft, the designer of this OS), that's when I get VERY LEERY of doing it - they did, after all, create this OS we use.

i have run all the tests we can think of and posted the results..

No, I have another... I will put it in my P.S. & hint @ it as I go in my reply to you here!

That test SHOULD tell us, what is what, I would think.

Fact is: We ALMOST hit it!

However, you didn't do the one test we should have.

You completely BURNED your pagefile.sys, but you never made a 1mb sized one (or as small as Windows LETS you make one)...

Try it, there is a good reason for testing that way, I will get into it in my P.S.!

so far nothing we have done has disproved my original conclusions.. that given enough physical memory windows dosnt seem to need or use the disk based variety..

BUT, you are paging, that much I proved to you... question is? Where to??

You can't prove it's "Memory-to-Memory" & I can't prove it's "Memory-to-RANDOMTEMPFILES-ondisk" either... @ least, in testing as we did, so far.

("Rock & A HARD PLACE!")

BUT, we're going to CHANGE that... read on!

i still accept the possibility of it needing disk based memory but all the tests i have run tend to suggest otherwise.. i also favour the idea that it simply pages to its physical memory..

Which defeats the typical operation of a virtual memory based OS... but, it could be possible, you never know (dumb, but possible).

You didn't try the 1 test we still need to - a SUPER SMALL physically diskbound located pagefile.sys (smallest Windows lets you make)... there's a reason for it, we want to see if when it is as TINY as you can make it, IF IT GROWS.

That will prove to us that it uses disk for the most part, when your explorer.exe pages (which we KNOW it does, as do your apps like winamp.exe when minimized).

it leaves me with a bit of a problem also when i poke my nose into pagfile threads and tell folks what they have read is all outdated rubbish based on machines of yesterday that rarely if ever had enough of the real stuff.. and dosnt really apply to a machine belonging to the cheap ram era..

It belongs to the design foundation of these OS' we use though, they are Virtual Memory Operating systems.

but threads like this are where the real information comes from.. i hope this one at least makes people wonder about windows and how it uses its memory..

trog

True... now, time to outline what to do to test this all out! The one test we missed... IF THIS TINY PAGEFILE.SYS grows? We KNOW you use disk, when you page, and WE KNOW YOU PAGE YOUR APPS already, that much I proved.

NOW, LET'S GET THE REAL ANSWERS (hopefully) YOU & I ARE LOOKING FOR!

(You've provided such an ODD scenario, I want the answers! lol...)

APK

P.S.=>LAST EXPERIMENT TO PERFORM (we never did this one):

Make the SMALLEST pagefile.sys Windows lets you make, under 12 mb as far as you can go, & then, start running things on your system...

I want to see, if say for instance, a 1mb pagefile.sys (IF POSSIBLE) will grow on you over there...

This alone will PROVE well enough, that even IF you have 2-4gb of RAM, you use a pagefile.sys because we DO KNOW you page, via pagefaults generating on explorer.exe constantly, & apps like winamp.exe when you minimize it/restore its window...

While doing it?

REALLY "push your system around" & bring TONS of programs into memory, with the biggest data you can find (most likely your biggest game & tons of instances of other apps you use)...

This is, the TRUE test, imo!

HOWEVER: You have to have enough programs + data in RAM to exceed your 2gb in RAM chips though...

(Because, We already KNOW you page, but where is the question & thinking it pages "Ram-to-Ram" is sort of outrageous, because the OS itself just is NOT designed that way. It is designed to use diskspace as "fake RAM" as you called it.)

Try that... get back to us! apk

 

[trog100]

loading more data than i have room for in physical ram made my small-ish 200 meg pagefile grow.. this has never been in dispute.. this to me just backs up my "only uses its pagefile when it has to theory"..

i know my system during normal usage and that includes some very memory hungry games.. the most memory hungry games in fact.. never made my 200 meg minimum size one grow..

there is the possibility that windows was busy paging away to my 200 meg file on a smallish scale and my assumption that because it didnt grow in size it wasnt being used was wrong..

i will see how small a one i can set.. but even if a 1 meg one dosnt grow.. windows could still be useing it on a very small scale..

how small would it have to be before we can except the fact it isnt being used cos it dosnt grow in size.. ??

before i do this by the way i am pretty sure it wont grow however small.. but if it dosnt grow.. the fact it has one set to a minimum size should put paid to any suggestion of it creating any other ones..

one thing is clear to me.. if it dosnt grow in size and windows is useing it.. it aint exactly shoving much data into it..

trog

ps.. something interesting just happend.. i just tried a min of 2 meg and max of 2000 meg.. windows does seem to say amin of 2 meg.. when i went back to the desk top i got the error message "windows is increasing the size of your swop file cos it aint big enough"..

more odd behaviour in my book.. windows seems happy not to have one but dosnt like a min 2 meg one..

it has given me a 44.032 k one.. it also did this without a reboot.. basically it is ignoring my 2 meg and giving me 43 meg one.. i will try and get a smaller one..

 

[trog100]

just tried a 2 and 200 meg one.. this time it tells me i have to reboot.. it has now given me a 2 gig one.. he he he

i will make sure it gets removed and try again..

trog

ps.. okay gonna give up on this.. windows behaves in an irrational way when u try and get a very small pagfile.sys created.. i have had anything produced from 2 gig down to 10 meg.. 10 meg isnt small enough to prove much..

let me come at this from a different angle.. in control panel.. windows lets u customize its pagefile.. i am gonna have to assume it means the disk based part of its system memory.. a different meaning than in task manager which seems to refer to system memory of any kind when it says pagefile in use..

okay why does windows quite clearly give u the no pagfile option.. by pagefile i assume it means the disk based part of its virtual memory.. it is that which is being altered after all..

i would suggest the answer to this is very simple.. it means exactly what it says.. windows will function or attempt to function without useing the disk based part of its virtual memory..

i think we are chasing ghosts alec.. we cant find a pagefile temp or otherwise because contrary to popular belief there isnt one.. when windows says no pagfile it means exactly what it says..

trog

 

[Alec§taar]

just tried a 2 and 200 meg one.. this time it tells me i have to reboot.. it has now given me a 2 gig one.. he he he

Yes... & I'd wager that NOW, you may have a temppf.sys (temp pagefile.sys) under %WinDir%\System32 (%systemroot) as the page from MS has been stating as well.

Maybe you can check on that now as well.

ps.. okay gonna give up on this.. windows behaves in an irrational way when u try and get a very small pagfile.sys created..

It behaves exactly as is noted on that MS page you felt was "out of date rubbish/bullshit" apparently.

okay why does windows quite clearly give u the no pagfile option.. by pagefile i assume it means the disk based part of its virtual memory.. it is that which is being altered after all..

Well, again as I have stated - to Windows? Disk based pagefile.sys &/or temppf.sys + your RAM IN CHIPS, is ALL VIRTUAL MEMORY.

i would suggest the answer to this is very simple.. it means exactly what it says.. windows will function or attempt to function without useing the disk based part of its virtual memory..

Oh, of course... if you don't exhaust the RAM in chips first. Except that, we STILL haven't figured out WHERE winamp & Explorer.exe were paging to, when you DIDN'T have a pagefile.sys!

i think we are chasing ghosts alec.. we cant find a pagefile temp or otherwise because contrary to popular belief there isnt one.. when windows says no pagfile it means exactly what it says..

trog

Well, the only thing I question is, when you had NO pagefile.sys period? Where then, was explorer.exe paging to?

After all:

We both verified, it INDEED, is paging via taskmgr.exe PAGEFAULTS column in the PROCESSES tab, as well as winamp as well when you min/restored its window.

APK

P.S.=> Ever think you & I may have found a "bug/hole" in the OS design? It's possible... apk

 

[trog100]

when i was checking the size of the pagefile windows gave me.. i wasnt going by the amount stated in the control panel by the way.. i was looking at the actual size of the pagefile.sys file windows created in my C drive root..

"Well, the only thing I question is, when you had NO pagefile.sys period? Where then, was explorer.exe paging to?"

if my way of seeing it is correct when its been told not to use the disk based part of its virtual memory (no pagefile is set) it simply pages out to whatever spare physical memory is there.. in my case there is always spare physical memory there for it to use.. in a system where there wasnt spare physical memory for it to use it wouldnt function..

once u accept the fact that windows by its very nature dosnt need disk based memory it can make do with just physical if told to.. it all fits and there is no puzzle..

i am back to the "no pagefile" setting now by the way..

changeing the subject.. i like your ramdisk idea but i think it would all work better if u increased your physical ram from 512 meg.. i can see your logic in thinking u dont need it.. but..???

trog

ps.. going away for a week boating.. if i dont sink i should be back after.. he he

 

[Alec§taar]

when i was checking the size of the pagefile windows gave me.. i wasnt going by the amount stated in the control panel by the way.. i was looking at the actual size of the pagefile.sys file windows created in my C drive root..

"Well, the only thing I question is, when you had NO pagefile.sys period? Where then, was explorer.exe paging to?"

if my way of seeing it is correct when its been told not to use the disk based part of its virtual memory (no pagefile is set) it simply pages out to whatever spare physical memory is there.. in my case there is always spare physical memory there for it to use.. in a system where there wasnt spare physical memory for it to use it wouldnt function.

Assuming you are correct, you run the risk of system (OS) &/or application crash, which you state is a possibility when using NO pagefile.sys... a risk I personally cannot take for how I use my machines.

Still, I think the theory of Windows using small tmp files for paging to disk when no pagefile.sys is present (such as in the case of explorer.exe paging which we did on your system & verified as true, &/or your winamp.exe doing the same when min/restore of its window) is the more likely - because this is the design of this OS family, & many others that use Virtual Memory utilizing designs.

BUT, you & I have no proof, either way (unfortunately).

I can see your point, it is possible, but imo? Sort of impractical, because you run the risk of crash/abend/errs.

once u accept the fact that windows by its very nature dosnt need disk based memory it can make do with just physical if told to.. it all fits and there is no puzzle..

It fits, but runs the risk of system (OS) &/or applications abends/errs/crashes. BUT, then again, the theory that using small temp files for paging of explorer.exe (constantly paging) may be a possible too...

We have no answers here, & neither does anyone else studying this online that I have seen, thusfar.

changeing the subject.. i like your ramdisk idea but i think it would all work better if u increased your physical ram from 512 meg.. i can see your logic in thinking u dont need it.. but..???

trog

It works, as is, & has given me no trouble thusfar... & SHOULD be even faster, once that DDRDrive x1 PCI-e model of a SSD for the masses releases, given that it uses faster memory & a larger bandwidth bus type than my CENATEK RocketDrive (PC-133 SDRAM memory & PCI 2.2 bus) does.

ps.. going away for a week boating.. if i dont sink i should be back after.. he he

Have fun, & it was an INTERESTING experiment/study here, albeit a pity we did not get SOLID answers on either account (other than you WILL crash/abend/errout if you run out of RAM using your "no pagefile.sys" setup IF you exhaust your RAM in chips).

APK

P.S.=> Sorry for delay in reply, Wed. & Thurs.? Tough to fit in forums-stuff for me, as those are days my friends & I take off & "enjoy the real world" outside of "the Matrix" here, lol...

Anyhow/anyways:

You & I, we have SORT OF diverged also into a very "shades-of-gray" area in the sizings of pagefiles as well (one where imo @ least? There are some very valid ways of doing that, based on RAM onboard your mobo in chips & also HOW YOU USE YOUR MACHINE... & it always boils down to that!)... & there is little question about performance gains, IF you use a pagefile.sys that is (safety net in ANY configuration against crashes), about placements however.

Sizing pagefiles: It's one of the most "hotly debated topics" I have ever seen online in fact, as I mention earlier on in this thread... apk

 

[Alec§taar]

Been "reading up" on this... here is what I think you are seeing Trog100

Trog100:

(ALSO, per our last discussion/analysis above: I think I know why you're showing paging on explorer.exe too now, by the by, even though you don't use a pagefile.sys period - CacheMgr subsystem can cause paging by the memmgr kernel subsystem )

I.E.-> Those 2 subsystems & the filesystem driver: They DO work massively close together, & especially depending on if a read is read in as noncached, or cached, but in YOUR case (no pagefile.sys use, & 2gb of RAM online)?

It's not paging in/out data from a pagefile.sys, but from data files it reads AND from data itself for itself in the "VACB" (more on that later) & when changed? You see paging from explorer.exe in taskmgr.exe PROCESS tab with pagefaults column visible! Sounds "weird" I know, but read on:

Hence the cached vs. uncached read-in data which can be marked thus by apps!

E.G. -> The Cache/Memmgr/FileSystem Driver are "punching back" files/data read in & used by Explorer.exe & also its OWN "VACB" data, memory to memory in YOUR case, since no pagefile.sys is present (if memmgr driven, & it will be since the cachemgr can force this or the filesystem driver, to the VACB (or from 512mb max size afaik cache array/buffer), but, will "hit disk" 'paging' on cached lazy writes)!

& if they change?

They are marked as "dirty" in the cache, & the VACB has to note it also (the allocated area for an app in VIRTUAL MEMORY & THE OS SEES IT ALL AS THAT, RAM IN CHIPS + PAGEFILE.SYS if present), & pushed back to file they used, BY THE MEMORY MGT SUBSYSTEM, in the data cached in from said files & THEN, changed!

Since the cachemgr, &/or FileSystem driver may be 'forcing' writes by the memmgr, it is being shown as PAGING on your end, via VACB changes of data while cached, & it is paging memory-to-memory (VACB is a buffer, a VM buffer, & since you have NO PAGEFILE.SYS? IT IS PAGING MEMORY TO MEMORY, until the filesystem + cache do a "lazy write" back to disk & the files data used was changed)!

(It's a WEE BIT DIFFERENT THOUGH, in the case of a file being shared by more than 1 app! That is when "copy on write" goes into effect when 1 instance of an app, changes the data file being shared too, making sure BOTH APPS HAVE THE DATA UPDATED while a file is shared).

THIS IS WHY (ALSO) WHEN YOU MINIMIZED WINAMP.EXE THERE PER MY TEST FOR YOU? IT TOO WAS SHOWING PAGING imo, albeit paging memory to memory to the VACB buffer, until (if needed) lazy written by memmgr forced to do so my Cachemgr requests on lazy writes - thus, the memmgr takes over, & when it does a write to disk, even to a data file it used? It will/is showing as "PAGING"!

* So far, it seems to make sense in YOUR case, where you do NOT use a pagefile.sys, & have 2gb of RAM online!

(This is good for your use patterns @ least, UNTIL YOU HIT A DATASET OR TOO MANY APPS IN RAM (which we tested & you began seeing hassles, but NOT your usual use pattern there but instead really "PUSHING" your system per my instructions to test that out) THAT OVERFLOWS IT, you can crash the OS or apps, but a risk you take only & CAN avoid).



* Sound about "write" (lol, pun intended)...?

APK

P.S.=> Memmgt, & cachemgr + file system driver interaction? IMO, one of the HARDEST THINGS TO GRASP about Windows... apk