Need a word of advice from someone who has had experience with editting the page file settings.

Some people say leave it at the default settings, some say cut it off completely if you have more than 2GB RAM and some say keep a tiny bit. So, for those who have actually had some real experience and tested with it, what is your recommendation? I set it from 6GB to 1GB for a month now, and have not noticed any difference. I'm not even sure what difference I should be looking for. What do you think? Worth it to spend time thinking about it or not?

 

If you're running XP and since you have 4gb RAM, you could actually disable the page file (even though windows only sees about 3gb). I always just leave it at the default, it doesn't hurt anything.

BTW, how did you get that cool font in your sig? I like the E in the Dell.

 

Franklin Gothic Medium

Looked at the page source.

 

I swear that's not an ' E'.

 

Hey thanks.

 

I’m with talin, for XP you can turn off page file with 4GB.

For about a week now, I'm running Vista without a page file; sofar it didn’t crash or give any memory errors. Apart from the normal word processing, internet and multimedia stuff, I do a bit of programing on Visual Studio 2008 Express edition, and some Photoshop work.

 

Hmm,

I've heard it's not safe to disable the pagefile since there are some programs that absolutely require it. What do you guys think about that?

And talin, how long have you actually disabled it for and what were your experiences with it?

 

It was when I had my inspiron. I had 2gb RAM though, and didn't do any gaming, just basic stuff, and I had no issues (XP). It was just for a few days though. With 4gb you'll be fine. But really, leaving it be wont hurt anything. I say leave it enabled.

 

I’ve run XP without a page file (having only 2GB RAM) for more than three years before I finally switched to Vista. I’ve seen few memory errors, but from what I can remember I never lost any important work or experienced critical system crashes. However I had to turn it back on to get some programs and games running (some of them simply wouldn’t load without a PF), but I ran more than 90% of the time without a page file.

Many are afraid of turning it off, and I suppose you could run into few problems; therefore I can’t really recommend it. But I don't care .. First two years after switching to Vista I only had 2GB RAM therefore couldn’t really turn it off, then I forgot about it, until last week.

P.S : No matter what others say, I personally experienced a difference when running without a page file in XP (programs would minimize/ maximize in an instant). Too early to say anything for Vista...

 

Oh great, here we go again. All these operating system experts who know more about operating systems than the folks who write them for a living. It boggles the mind.

I especially love the story about doing program development without a pagefile. Yeah, that just makes so much sense. Running the OS in a fashion that has the potential for an application crash if the OS runs out of memory, while creating code. Running the risk of trashing the code you are working on, is not my idea of fun. Sort of like the trapeze artist who likes to practice without a net. I have written code for a living for the past 40 years and managed many development projects. If I ever had someone on my staff who was willing to risk a days worth of coding on such a potentially dangerous, unproven tweak, I'd show them the door. (Before it starts, spare me the "well it has never happened to me" rhetoric. The potential is there for appication crashes without a pagefile, we all know that.)

All real tests I have seen that actually measured the performance differences with and without a pagefile, show no discernible improvement in performance. The only reports that indicate there is a difference are pure anecdotal. "I seemed faster." "I know it was faster." "It felt much quicker." No proof, no numbers, no test results with and without.

Yep, this just sounds like an absolutely unbelievably great idea... again.

Gary

 

I have 3GB of RAM, and people recommend setting the page file minimum and maximum size 1.5 times the size of your current amount of RAM. Lets do the math: 3GB (3000MB) x 1.5 = 4.5GB (4500MB) / 4500MB = Pagefile. Although, I usually let Windows manage the page file size automatically. But that might use up too much space, or raise disk write activity. But remember, don't let your eyes trick you.

 

@Scud: I don't feel a difference, it must be a placebo affect to people then. And I don't understand how you can actually just tell Windows to not use a page file and it will listen that easily. Probably more work involved, Task Manager shows 398MB of PF usage. On it goes, thank you Scud.

@Thong: What is the point of the 1.5 factor?

 

I'm not sure. It's just a normal recommendation that most people have told others to use sometimes.

 

I've always just stuck with a fixed pagefile equal to the installed physical RAM.

That being said, system managed pagefiles don't suffer from the same kind of fragmentation/performance issues they used to back in the day, there's probably no difference between running a manual fixed and a system managed one nowadays.

 

Let me clarify few things, With Vista I am yet to see any difference; Maybe turning off page file has no effect or the effect is minimal. However, with XP you definitely should feel a difference. Note that your programs will NOT load faster, your OS will NOT boot any quicker, nor any CPU/IO bound applications will run any quicker either! BUT if you have several different applications running (MS word, couple of browser instances, a media player etc..) then switching between them is almost instantaneous without a page file.

Before you say I'm talking rubbish, Please do this on a XP computer with 2GB or more ram and see for yourself. Load several programs and wait a while then try switching between them. This is the only way to really test it.

 

An out of memory exception will crash the program, but it will not trash the latest code modifications, at least not with Visual studio. VS saves all the modifications when you build a program, therefore even if the program crashes the code modifications are NOT lost. And it will not crash the OS or the IDE, OS will handle the exception and control will be returned to the IDE after terminating the program.

In the most unlikely "worst" case scenario where it does crash the whole OS (yet to experience this), all you have to do is reboot.

 

Im with Scuderia on this one. Its not worth disabling it

 

I am well aware of how Visual Studio works. You would loose all program modifications made since the last build (if that option is turned on). If you are working on debugging that might mean only the loss of a couple of minutes work. If you are writing a whole new function or section of code, the amount lost would be significantly more, potentially measurable in hours. Significant enough that, as I said, were it done in my shop the offender would find themselves looking for employment elsewhere. Any PERCEIVED improvement is not worth the productivity loss, period.

As for the testing method suggested, it is laughable at best. Testing involves real measurement of results in a controlled environment. As I said before, every single such REAL test I have seen, dispels this myth. The only claims of performance improvement are anecdotal and even then minimal.

Gary

 

Apparently, Mark Russinovich, author of many of the Sysinternals utilities and co-author of Windows Internals, is also with ScuderiaConchiglia on this one.

Mark recently posted a discussion of virtual memory management in the various WinOSes on his Microsoft blog page, in which he briefly describes some of the performance issues that the size (or existence) of the paging file affects. In particular, the size of the paging file determines, in part, the amount of virtual memory that all of the processes running on your system can "commit" - i.e., virtual memory that the process intends to use, and which therefore requires that the OS be able to guarantee that it can "maintain all the data the process stores in that committed memory either in physical memory or on disk." (to quote from Mr. Russinovich).

Now, on 32-bit WinOSes, each process can have its own virtual address space of up to 2GB, which means that, theoretically, it could potentially reserve, and commit, almost 2GB of virtual memory. Most processes, obviously, will not do that; however, if you've got a process that's got a memory leak, or you're running an app that is very data-intensive, you could quickly reach the point where that process has reserved and committed the maximum of 2GB of virtual memory.

So, for example, put two misbehaving processes with memory leaks together on a 32-bit WinOS system with 4GB of physical memory and no paging file, and sooner rather than later, you'll start getting out of memory exceptions because those two processes, by themselves, will have committed a total of 4GB of memory (actually, somewhat less since then there'd be no physical memory for anything else, including the OS itself ).

Adding a paging file relieves that stress and, in point of fact, could probably serve as a useful early warning indicator of a misbehaving process with a memory leak before you hit an out of memory exception - just set a monitor to monitor the page file usage, and if it starts to nudge up consistently with no explainable reason, such as that you've slowly been running more and more apps at the same time, then you've got an indicator that there might be a memory leak or something else fishy going on.

At any rate, Mark's article makes for very enlightening reading, along with the first part of the series that precedes the article I mentioned above, and which can be found here.