ViksitA
asked on
Page-File Size Range setting
Is it essential to set a range (different values of min. & max.) for page-file size in Win2000 ? If I set both min. & max. page-file size to a same value, how is it going to affect performance of the machine ? Will there be any problems while executing some high memory hungry processes ?
-ViksitA.
-ViksitA.
ideally your page file should be the size of your physical memory * 2. although it dont hurt to have it bigger, much bigger in fact is okay.
you may find better performance by fixing the size of the page file but you will have set it at the max you ever expect to need. this of course could be wastefull if it only reaches this once in a blue moon.....
really to answer this question properly you need to look at the applications that you run and their behavior pattens. you can use performance monitor to monitor page file usage along with commit charge and a whole host of other statistics
Paul
you may find better performance by fixing the size of the page file but you will have set it at the max you ever expect to need. this of course could be wastefull if it only reaches this once in a blue moon.....
really to answer this question properly you need to look at the applications that you run and their behavior pattens. you can use performance monitor to monitor page file usage along with commit charge and a whole host of other statistics
Paul
The page file sizing should be RAM * 1.5 (not *2).
Fixing Pagefile size will avoid the system to handle size dynamically while working. Better performances. But you can't look at your pagefile while working in order to see rapidly if his size is enough for your system, because his size is fixed !!! if you set a size between 100 and 500, and you see that the pagefile size is 200, you can see that you've enough paging space.
If you really want to have the best performance of paging, you will have to look closely at your applications, and how you work with them, by using perfmon or other tool.
Fixing Pagefile size will avoid the system to handle size dynamically while working. Better performances. But you can't look at your pagefile while working in order to see rapidly if his size is enough for your system, because his size is fixed !!! if you set a size between 100 and 500, and you see that the pagefile size is 200, you can see that you've enough paging space.
If you really want to have the best performance of paging, you will have to look closely at your applications, and how you work with them, by using perfmon or other tool.
People, let's not argue pointlessly, Ram +12 mb is another valid formula for initial size, with 40% more for maximum size. Yet another formula is initial size = ram +10%, and max=remaining disk space. everyone has their own way.
the problem is that Microsoft wants it left alone, and doesn't publish official information. every pc installation is different enough that this setting is best custom configured to each PC. if you hear the HDD making noise all the time, for several days, increase the numbers by 5 or 10 % every few days until it quits. that's it.
the problem is that Microsoft wants it left alone, and doesn't publish official information. every pc installation is different enough that this setting is best custom configured to each PC. if you hear the HDD making noise all the time, for several days, increase the numbers by 5 or 10 % every few days until it quits. that's it.
hi bro(makerp)
when you install windows 2k it sets the intial size to an average of 1.5 times your RAM (recmm from microsoft). If you have your int and max sizes the same this I believe stops windows increasing the page file itself.. therefore the difference is the amount windows is allowed to increase it to (upto the max)
with hard disk sizes being so large now days, you could have a paging file of really what ever size you wanted, for i have found best perfomance with having a small one on your c:\ partiton, then create a partition that contains a page file only.
regards.
makerd
when you install windows 2k it sets the intial size to an average of 1.5 times your RAM (recmm from microsoft). If you have your int and max sizes the same this I believe stops windows increasing the page file itself.. therefore the difference is the amount windows is allowed to increase it to (upto the max)
with hard disk sizes being so large now days, you could have a paging file of really what ever size you wanted, for i have found best perfomance with having a small one on your c:\ partiton, then create a partition that contains a page file only.
regards.
makerd
Here is my setup 394MB's of RAM
Pagefile
Min 200MB's MAX 200MB's
I have ran into no problems with this setup what-so-ever.
It took some time to find the settings that worked best for me. I purposely started out with settings much lower then they are now and kept increasing the value until I stopped getting messages of the type "Windows needs to increase pagefile size". I do get this message now but rarely so I haven't increased the pagefile size in a long time.
I personally recommend setting the MIN and MAX to the same size because it cuts down on defragmentation and Windows doen't have to keep resizing the pagefile which helps the performance issue.
Ideally as has been mentioned moving the pagefile to a partition that only houses the pagfile is a good way to optimize but most folks don't have this as an option.
The more RAM you have the less the pagefile is used and the size of the pagefile can be reduced. It takes some time but experiment to find the sweet spot for you.
The Crazy One
Pagefile
Min 200MB's MAX 200MB's
I have ran into no problems with this setup what-so-ever.
It took some time to find the settings that worked best for me. I purposely started out with settings much lower then they are now and kept increasing the value until I stopped getting messages of the type "Windows needs to increase pagefile size". I do get this message now but rarely so I haven't increased the pagefile size in a long time.
I personally recommend setting the MIN and MAX to the same size because it cuts down on defragmentation and Windows doen't have to keep resizing the pagefile which helps the performance issue.
Ideally as has been mentioned moving the pagefile to a partition that only houses the pagfile is a good way to optimize but most folks don't have this as an option.
The more RAM you have the less the pagefile is used and the size of the pagefile can be reduced. It takes some time but experiment to find the sweet spot for you.
The Crazy One
>>Ideally as has been mentioned moving the pagefile to a partition that only houses the pagfile is a good way to optimize but most folks don't have this as an option
this makes little difference... really you want to have the page file on a _seperate_physical_disk_ as this would break up the I/O across seperate IDE chanels
I think we can summarise by saying.... The size of the page file should reflect the type of applications that the machine in question runs. although Microsoft recommends *1.5 in reality it all depends on the above and besides all OSs recommend different sizes but yet a page file on all modern protected mode OSs is fundamentally the same thing.
also it is worth noting that page file size should in theory decrease as the amount of physical ram increases as there is less chance of the OS actually needing it, therefore in CrazyOnes case he/she only needs 200MB because he/she has loads of real memory. on the other hand I need a page file of 500mb because I only have 16mb of ram (not really but it emphasises my point)... therefore it looks like we could derive a puka formula whereby the size of the page file decreases linear as ram increases.....
if you are interested in your page file usage then simply use performance monitor to view some memory counters...
on a final note windows uses (or atleast did use) a page replacement algorithm that seems to use the page file regardless of how much ram you have. Therefore you will find that the more page file you give the more it _may_ use, this is paradoxical because there more of the page file it uses instead of memory would result in a decrease of performance, yet you add more ram and page file to increase performance....
this makes little difference... really you want to have the page file on a _seperate_physical_disk_ as this would break up the I/O across seperate IDE chanels
I think we can summarise by saying.... The size of the page file should reflect the type of applications that the machine in question runs. although Microsoft recommends *1.5 in reality it all depends on the above and besides all OSs recommend different sizes but yet a page file on all modern protected mode OSs is fundamentally the same thing.
also it is worth noting that page file size should in theory decrease as the amount of physical ram increases as there is less chance of the OS actually needing it, therefore in CrazyOnes case he/she only needs 200MB because he/she has loads of real memory. on the other hand I need a page file of 500mb because I only have 16mb of ram (not really but it emphasises my point)... therefore it looks like we could derive a puka formula whereby the size of the page file decreases linear as ram increases.....
if you are interested in your page file usage then simply use performance monitor to view some memory counters...
on a final note windows uses (or atleast did use) a page replacement algorithm that seems to use the page file regardless of how much ram you have. Therefore you will find that the more page file you give the more it _may_ use, this is paradoxical because there more of the page file it uses instead of memory would result in a decrease of performance, yet you add more ram and page file to increase performance....
and also setting the min and max to be the same is more efficent because windows can not spend time adjusting the file itself.
i do beleive this question needs closing.
makerp, beers this week some time? howz the decorating?
makerd
makerp, beers this week some time? howz the decorating?
makerd
ViksitA, do you have any other questions or queries about the information provided.
>>>this makes little difference
Actually it can make difference depending on how the user has setup the pagerfile. If the max an min are different vaules then the pagefile has a tendency to fragment which in time decreases performamce. By putting the pagefile on partition that is only used for the pagefile virtually eliminates the fragmentation of the file.
Actually it can make difference depending on how the user has setup the pagerfile. If the max an min are different vaules then the pagefile has a tendency to fragment which in time decreases performamce. By putting the pagefile on partition that is only used for the pagefile virtually eliminates the fragmentation of the file.
okay, fair enough... but is fragmentation really a problem. after all there may be simultainious disk reads going on causing the disk head to move all over the shop
True I was just trying to clarify it. Putting it on a seperate disk would be the best way to do it so I agree with that idea whole heartedly.
there is a reg hack for clearing the pagefile on shutdown, therefore i would believe that fragmentation would not be a problem>?
makerd
makerd
makerd ?????....
how would clearing the page file reduce fragmentation
fragmentation is to do with the layout of the file blocks of the page file on disk... to de-fragment them would be to re-arrange them so they are on consequtive blocks in the hope that linnear reads are then speed up because the disk head aint gota fly all over the place (in reality this dont always matter cos the disk will be searving other processes which may be reading from different parts of the disk causing disk head movment anyway.... although some drivers do try to order reads to minimise this... thats another story)...
clearing the page file surley means just to write an array of known values through its contents. the benifit of this i can not see... it just sounds like a way to stop hackers gaining access to previous memory contents.... maybe this is an option for the paraniod user...
how would clearing the page file reduce fragmentation
fragmentation is to do with the layout of the file blocks of the page file on disk... to de-fragment them would be to re-arrange them so they are on consequtive blocks in the hope that linnear reads are then speed up because the disk head aint gota fly all over the place (in reality this dont always matter cos the disk will be searving other processes which may be reading from different parts of the disk causing disk head movment anyway.... although some drivers do try to order reads to minimise this... thats another story)...
clearing the page file surley means just to write an array of known values through its contents. the benifit of this i can not see... it just sounds like a way to stop hackers gaining access to previous memory contents.... maybe this is an option for the paraniod user...
i would actually think that windows automatically disallows the page file to fragment if it can (disk space permitting). this is atleast tru for the min size, as it grows it may need to allocate non-consequtive blocks
yep, sorry bro, thought it would remove it then re-write on startup. KB:
http://support.microsoft.com/default.aspx?scid=kb%3ben-us%3b182086
maKERD
http://support.microsoft.com/default.aspx?scid=kb%3ben-us%3b182086
maKERD
this document justifies my statement
>>it just sounds like a way to stop hackers gaining access to previous memory contents.... maybe this is an option for the paraniod user...
password etc
any road, this question is dead i think.... worst still i have a strange feelin that ViksitA aint ever gonna use EE again... not even a whisper from him...... ho hum
>>it just sounds like a way to stop hackers gaining access to previous memory contents.... maybe this is an option for the paraniod user...
password etc
any road, this question is dead i think.... worst still i have a strange feelin that ViksitA aint ever gonna use EE again... not even a whisper from him...... ho hum
ASKER
Many thanks to all for your responses. Pretty good amount of information.
I thought setting of min and max page file size is the only factor that lead to some inconsistencies in the performance testing results. Perhaps having page file in a seperate partition would be another important consideration.
Are there any other such settings to be performed to reduce the inconsistencies in the performance results between executions in the same test environment and the version of software under test?
I thought setting of min and max page file size is the only factor that lead to some inconsistencies in the performance testing results. Perhaps having page file in a seperate partition would be another important consideration.
Are there any other such settings to be performed to reduce the inconsistencies in the performance results between executions in the same test environment and the version of software under test?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks to all for their valuable comments.
Special thanks to CrazyOne for giving an idea about having an exclusive partition for page-file and for providing information regarding performance tweaks.
Since, I got information from all comments. I can't accept a single comment as answer. But for the sake of closing the question, I would like to accept CrazyOne's comments as answer.
Special thanks to CrazyOne for giving an idea about having an exclusive partition for page-file and for providing information regarding performance tweaks.
Since, I got information from all comments. I can't accept a single comment as answer. But for the sake of closing the question, I would like to accept CrazyOne's comments as answer.
increase page file size to slightly beyond your estimated max requiremnets