Do you play games with pagefile on or off?

[alpha_omega]

Turning off the pagefile is a bad idea for sure, but wouldn't manually setting both the min and max value to the same amount prevent the pagefile from getting fragmented?

That is the perfect indicator for low memory (or an overly fragmented partition). Time to upgrade (resp. defrag)!

By the way, the OS pushes unneeded "stuff" to the pagefile first, so it being fragmented is of little relevance...

[Saul Goodman]

Turning off the pagefile is a bad idea for sure, but wouldn't manually setting both the min and max value to the same amount prevent the pagefile from getting fragmented?

Oh god talk about taking me back, we used to do that supporting Windows NT4.

[spaceelf]

There is no such thing as a secret page file. This is a myth that I think originated from the task manager in older versions of Windows, which incorrectly labeled certain graphs "page file usage" when that is not what they were. If you disable all page files, it really disables them.

The rest is right, which is why I suggest that he actually measures it.

It isn't secret, but it's never fully disabled either. Windows will use a pagefile whether you disable it or not, as I understand it.

[WastedJoker]

I split my paging file across two drives.

[The_Observer]

just one and on another drive other than windows. Works fine... never turned it off

[hdood]

If you open a program that uses 1GB of RAM and you have 512MB of RAM and your page file is disabled, what happens?

When it tries to allocate that much, Windows will return an error to the program saying that there is not enough memory. The program can then either gracefully deal with this or crash, depending on how well it is written.

Turning off the pagefile is a bad idea for sure, but wouldn't manually setting both the min and max value to the same amount prevent the pagefile from getting fragmented?

You can defragment the page file if you really want to though.

It isn't secret, but it's never fully disabled either. Windows will use a pagefile whether you disable it or not, as I understand it.

I showed you that it is not true. You aren't arguing with me now, you're arguing with Microsoft. I say secret because the people who make the claim can never document it. It's the kind of misconception that has given birth to the "put your page file on a RAM disk!" nonsense.

I think the belief stems from several things. The first is that Windows is a virtual memory-based OS (as is Linux and OS X), but this in itself has nothing to do with page files.

Another is the fact that Task Manager in XP and earlier had graphs that said "page file usage" that would still show the same even if you disabled all page files. The reason it did this is because it was incorrectly labeled. The graphs actually showed the commit charge, which simplified is the total amount of both physical RAM and page files in use.

The "page faults" column in Task Manager might also have some of the blame, because people think it actually shows paging to disk. It doesn't necessarily, because it contains both soft and hard page faults. A soft fault is when the data is already in memory, but not in the process's address space. Those are routine on any virtual memory OS, and have little performance impact. Hard page faults on the other hand, are when it has to read data from disk (but not necessarily a paging file.)

Of course all of this is a bit of a derail. The point is that Windows can run just fine without any page files, which can be useful for certain specialized setups, but not of much value on your normal general purpose computer. If people insist on doing it, they should at least do some testing and benchmarking to establish that it actually helps them. If it does, knock yourself out.

[spaceelf]

I split my paging file across two drives.

Supposedly the best advice. I'll be doing that again once my Samsung drive comes back from RMA.

[Joshie]

I have a lot of old page file 'tips' in my head that I've accumulated over the years (oh man, some of the old 9x optimization tips...) that I wonder today about the relevance or effectiveness.

The whole manually setting min/max to the same value one is still in my head and is almost a reflex after a clean install for me, if just out of habit. I know it doesn't hurt anything, but I wonder if it really makes much of a difference (if Windows keeps the drive bare up to the max value point anyway, it probably makes no difference).

I wonder, though, if it can still help to set your page file to a separate harddrive. Time was, if you had a spare, relatively low-capacity, fast drive lying around (the upgrade cycles always leave us with these), it seemed a good idea to throw it in and use it for virtual memory to slightly reduce access to your system drive. Is this still a fundamentally solid idea?

[spaceelf]

I showed you that it is not true. You aren't arguing with me now, you're arguing with Microsoft. I say secret because the people who make the claim can never document it. It's the kind of misconception that has given birth to the "put your page file on a RAM disk!" nonsense.

Personally I'm not arguing with anyone. That was the impression I was under, and it doesn't bother me either way.

http://blogs.technet.com/b/markrussinovich/archive/2008/11/17/3155406.aspx

Some feel having no paging file results in better performance, but in general, having a paging file means Windows can write pages on the modified list (which represent pages that aren?t being accessed actively but have not been saved to disk) out to the paging file, thus making that memory available for more useful purposes (processes or file cache). So while there may be some workloads that perform better with no paging file, in general having one will mean more usable memory being available to the system (never mind that Windows won?t be able to write kernel crash dumps without a paging file sized large enough to hold them).

It probably doesn't matter if you disable it given enough memory, but I'll pass.

[hdood]

I wonder, though, if it can still help to set your page file to a separate harddrive. Time was, if you had a spare, relatively low-capacity, fast drive lying around (the upgrade cycles always leave us with these), it seemed a good idea to throw it in and use it for virtual memory to slightly reduce access to your system drive. Is this still a fundamentally solid idea?

I suppose it still applies, but at the same time you have to factor in that times have changed and that it's not uncommon for modern systems to never run low on RAM. That means that even though the logic is sound, it doesn't make much difference in practice.

Personally I'm not arguing with anyone. That was the impression I was under, and it doesn't bother me either way.

Maybe, but you have to keep in mind that others read these threads. A lot of people read them. That's what the problem is.

I see you're quoting Russinovich (one of the few Technical Fellows in the core OS division of MS.) It might be helpful to know that he is also the principal author of the Windows Internals books that I quoted earlier.

[asusmc]

Windows will complain that it is low on memory and ask you to close one of more applications.

When it tries to allocate that much, Windows will return an error to the program saying that there is not enough memory. The program can then either gracefully deal with this or crash, depending on how well it is written.

Yes but while that message shows up you can keep opening and opening and opening and opening and............ programs.

There was to be a point where a page file, even if it is disabled, HAS to exist because of this.

[hdood]

Yes but while that message shows up you can keep opening and opening and opening and opening and............ programs.

Yes you can, and they will all just fail.

There was to be a point where a page file, even if it is disabled, HAS to exist because of this.

I'm not sure what you're trying to argue. All that happens when you run out of memory is that memory allocation functions (like malloc) fail and return an error code to the program that called it. The program simply doesn't get what it asks for, and can deal with that in whatever manner it chooses. This is the same whether you run one or ten or a hundred programs.

[c3ntury]

Its on by default for a reason :)

^ This. Considering Windows / Linux or whatever have whole teams dedicated to improving preformance, surely if the pagefile was so bad, it wouldn't be enabled?

[asusmc]

I'm not sure what you're trying to argue.

What Im trying to argue is that there HAS to be a page file, even if it is disabled, for a critical situation.

What if the program is badly programmed and even though it returns a error code, it doesnt exit the process thus not freeing the memory? Windows draws up the GUI stating the error code (taking up memory itself as well) and it has to draw memory from somewhere.

[+Anarkii Subscriber²]

I just let Windows 7 manage my RAM. It is the best solution I found.

[kjx]

^ This. Considering Windows / Linux or whatever have whole teams dedicated to improving preformance, surely if the pagefile was so bad, it wouldn't be enabled?

A lot of things in Windows are geared to what they view as the "average user", and many settings are set to "safe" ones which might result in decreased performance but will lower the number of calls coming through to support services.

Sure, some settings will be optimized around to account for different types of hardware but beyond a point performance is affected by how each individual user uses their system. Operating systems can do a better job by asking the user what their usage scenarios are, like if they are never going to use wireless, there's no real point in running related services (like WLAN Autoconfig). But that would introduce a degree of "complexity" of novice users who might one day decide to use the wireless card and find it not working (If you want to be pedantic, yes, Diagnostic Policy service could detect it; but this is just an example).

When considering the pagefile, the consensus is that it is needed, particularly to allow Windows to allocate memory to applications that request it without physically allocating it in RAM. But what about someone with 8GB of RAM but never needs to use more than 2GB. Going by MS recommendations, that would mean a 12GB pagefile. Yes, that's not harmful but what if the user had a 64GB SSD?

My point? The people coding OSes might be (are!) really smart and do a lot of research into improving performance, but they are delivering their binaries to 100s of millions of users, and some corners get cut so that the "average user" is happy. Things will get better over time but we are have not reached the point where the OS knows better than the most knowledgable user (but I'll admit that it knows better than the "average user").

[Singh400]

I know some games/programs don't run well with pagefile off.

Really? Like what? I've set mine to a static size and left it on my SSD.

[DDStriker]

What Im trying to argue is that there HAS to be a page file, even if it is disabled, for a critical situation.

What if the program is badly programmed and even though it returns a error code, it doesnt exit the process thus not freeing the memory? Windows draws up the GUI stating the error code (taking up memory itself as well) and it has to draw memory from somewhere.

if it receives an error code from malloc it means the memory could not be allocated, in which it never receives the memory in the first place, so it doesn't need to free it later

[Southern Patriot]

Why not create a ram drive and point your pagefile to it?

Seriously?? :rolleyes: :no:

[Dashel]

Those days are well behind us.

[Salty Wagyu]

I leave it on, but since I use an SSD and 8GB of RAM I just shorten the size of mine so it doesn't take up 8gb of space on the SSD too. Say, around 2gb.

[argonite]

My point? The people coding OSes might be (are!) really smart and do a lot of research into improving performance, but they are delivering their binaries to 100s of millions of users, and some corners get cut so that the "average user" is happy. Things will get better over time but we are have not reached the point where the OS knows better than the most knowledgable user (but I'll admit that it knows better than the "average user").

Judging from the lunacy of some of the tweaks that "advanced users" make to Windows 7, I'd say the people at Microsoft accounted pretty well for all user types. Even advanced users have to admit, at a certain point, they have no idea what really goes on behind the scenes better than the people who made Windows.

And who's to say there's nothing in the Windows source code to account for all the niche situations that people keep tweaking about? Just because you don't see an option in the Control Panel or registry, doesn't mean Windows won't do some desperate measures when its stability is compromised.

[hdood]

What Im trying to argue is that there HAS to be a page file, even if it is disabled, for a critical situation.

It seems like you believe Windows is designed so that nothing must ever fail. It isn't. It's also worth pointing out that having a page file only increases the amount of memory available, it doesn't make it infinite. In other words, everything that applies to running out of memory without a page file, also applies to running out of memory with one. Both are easy to do if you try.

What if the program is badly programmed and even though it returns a error code, it doesnt exit the process thus not freeing the memory?

Like DDStriker says, the memory is never allocated and so does not have to be freed.

Windows draws up the GUI stating the error code (taking up memory itself as well) and it has to draw memory from somewhere.

Windows doesn't tell the user, it tells the program that made the function call, either by returning an error code, or by raising an exception. It tells the program--in code--that "sorry, the memory you asked for was not allocated."

Now, Windows will occasionally display warning messages to the user to notify him that memory is running low, but this is a separate thing and there's no 1:1 relationship with the number of failed allocations. It won't flood the screen with dialog boxes. There are also provisions to make sure that this dialog box can still be displayed in most cases, and to help you kill programs that are using excessive amounts of memory to help you restore the system to a usable level.

Disabling the page file is like running an OS that has no virtual memory. I don't know if you are old enough to remember, but those used to exist. The Amiga, for instance, did not have it. Once you ran out of memory there, you had to exit stuff to make more available. It's the same with Windows.

My point? The people coding OSes might be (are!) really smart and do a lot of research into improving performance, but they are delivering their binaries to 100s of millions of users, and some corners get cut so that the "average user" is happy. Things will get better over time but we are have not reached the point where the OS knows better than the most knowledgable user (but I'll admit that it knows better than the "average user").

Many people consider themselves expert when they really aren't though, or when they are experts on something completely unrelated. Obviously many advanced options have to exist, but it's probably safe to say that many people who go near them don't really know what they're doing.

[Udedenkz]

No page file.

Also,

There is no secret page file.

[asusmc]

No time to really post to hood so Ill keep it short

if it receives an error code from malloc it means the memory could not be allocated, in which it never receives the memory in the first place, so it doesn't need to free it later

When programming in C (simple programs), when malloc failed to reserve memory for say a dynamic struct, it would say the program has a error cant continue bla bla bla.......but the program is still running in the background even though that dynamic struct could never be reserved because (in this case) a illegal memory reservation..