[GUIDE] Using ImDisk to set up RAM disk(s) in Windows with no limit on disk size

PAE is not enabled by default. PAE is only enabled if a processor supporting NX (AMD) or DX (Intel) is detected as both require PAE to be enabled to function.

 

Microsoft introduced DEP in XP/SP2. It is enabled by default. DEP requires PAE, thus PAE is enabled by default.

The point is, it is enabled by default, and I don't think you can find a CPU today that doesn't meet the requirements.

When Microsoft speak of stability problems it is mostly regarding memory above 4G, not PAE-mode.

 

Absolutely nothing. However, the Page File is actually used for keeping some state information which helps the operating system. It's not only used for "virtual memory". Keeping it on RAM also helps ensure that information is flushed on every reboot which is pretty good seeing how there's almost no way to retrieve the RAM info unless you implement a cold boot attack (which isn't guaranteed to work unless you actually know how to do it).

This goes the same with web browser caches, it's actually safer to use a RAM drive to store your web browser caches so that any personal data that may end up being cached will reset every time you reboot your system. Again, yes, it's prone to the cold boot attack but the attacker will need physical access to your desktop and be able to do the entire attack fast enough before the RAM loses all of its contents while in the OFF state.

As for PAE, I wrote extensively about it a few years ago in a small FAQ I created on my website: http://www.kwokinator.com/book/operating-systems-faq/windows

PAE in Windows desktop is rather broken and is usually used with servers. It's not recommended to use 32-bit Windows if you have more than 4 GB of RAM as PAE only extends the physical address but not the virtual addresses.

deltatux

 

Putting page file on ramdisk is pointless.

http://www.overclock.net/t/1193401/why-it-is-bad-to-store-the-page-file-on-a-ram-disk/0_20

Mark Russinovich, who has a Ph.D in computer engineering and works at Microsoft, said, "Yes, putting the pagefile on a RAM disk is ridiculous."

 

We'll I'll be damned. I never knew that the Ram disk actually takes away from the commit memory. I'll have to change my advise. Thank you for the information.

From Mark Russinovic:

Following link from article from OP information:

Pushing the Limits of Windows: Virtual Memory

You might need to log in with a Windows Live ID to be able to read the article : ex Hotmail, Messenger, Xbox LIVE — and other Microsoft services.

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

Here is an article about " The Windows 7 Page file And Running Without One " which I have posted / mentioned before:

http://www.tweakhound.com/2011/10/10/the-windows-7-pagefile-and-running-without-one/

Virtual Memory in Windows XP:

http://aumha.org/win5/a/xpvm.php

RAM, Virtual Memory, PageFile and all that stuff:

http://members.shaw.ca/bsanders/WindowsGeneralWeb/RAMVirtualMemoryPageFileEtc.htm

Lots to read. Good to know.

 

What did you think? Everytime you execute something, you take a bite from the commit pool

 

I never gave it a thought. Any time you start a program you take from the commit pool. I never really thought about the limit it might have.

Back in the good old days of DOS, Ram Disks were common. Never heard about it being an issue with other limits on memory. Then again, I'm talking about DOS, not a modern OS.

 

From the TweakHound article

lol, that wasnt necesary at all. He mad.

 

I was thinking about setting it to 1 - 1.5gb, i've read this guide
http://www.ocztechnologyforum.com/f...ks-Utilities-*&p=442160&viewfull=1#post442160
and it says this

Will it dump complete 1.5gb or just used part? btw it would dump it to normal hdd.. so i guess if its whole 1.5gb then ~ 25sec longer?

I would use it for Firefox cache and maybe temp files, but idk if 1.5gb is enough for temp files?

Thanks :nerd:

 

Try it, at worst, Windows will tell you that you're out of free space.

 

It would dump the exact size of the RAM disk.
e.g, if you set it to 1GB and only filled it to around 200MB, then it'll dump the entire 1GB, not just the 200MB you used.

It's enough in most cases.
But since you use it along with FF cache, then you might want to limit the space that FF cache will use by limiting the maximum cache size.

 

I use 500mb on my system for internet temp files, and system temp files and it works perfectly. I have never used all that space at once.

 

Ahh ye but he has a Ph.D so he's a damn fine philosopher but he knows jack **** about systems engineering.

Effectively, having no page file at all can give you a minor performance boost but the problem with that is some apps refuse to run without one (regardless of whether you have enough physical RAM).

So, rather than simply disabling the page file altogether, a nice workaround to this specific problem is to put the page file on a ramdisk and, optionally, to reduce the page file's size to the minimum size allowed by the OS (or set it to the size required by the app) so that it doesn't have to waste more RAM than what's necessary for the workaround to be able to work around.

However, obviously, if you don't have enough RAM to begin with then the point is still moot (same as when you don't have enough RAM to be able to run without a page file, anyway). A small side bonus IMO is it's quicker to re-enable the page file on a ramdisk than it is to re-enable it on a harddisk, should you encounter an app that requires it.

http://www.romexsoftware.com/en-us/vsuite-ramdisk/index.html

:deflate:

EDIT: And if you just read through that whole OCN thread you linked, you might find a few more good reasons why putting the page file on a ramdisk can, in some peculiar situations, be a great idea after all.

 

Ok thanks you two for all the info.

I'll try something later tonight

 

Which apps?
Provide some proof before comming which such a claim.

 

It will be less than 8GB dependent on the filesystem and allocation uinit size.
So there is some overhead caused by MFT and cluster size.
I think it's around 13% for NTFS? 12.5 default MFT and approximate 0.5% for data.

 

It's in that same OCN thread to which Riou kindly posted the link (the link, which, for your convenience, I had put in my quote). It was also pointed out by a fellow guru3d forum member a while ago but I can't remember the name.

Anyway, I am by no means claiming it all makes sense. Putting the page file on a ramdisk should not be necessary but, on a rare occasion at least, it very well might be and I hope this shows how trial and error is often still the one best way to finding solutions that work.

 

Maybe I'm blind, but I do not see a single claim about any app that requires a paging file to run on a system with plenty of memory.
Heck, I have not seen any single app that I had to test eating up my memory neither require a paging file, so I would really love to know which app's your on about.

It is kinda like your claim about Mark.
He knows a bit more about system engineering than you are aware of, so saying he knows nothing about it is not a fact, stick to facts please.

 

I presume you've never played Dawn of War II or ran any other apps that do check for the existence of a page file, regardless of the currently available memory.
I'm pretty sure in that "8GB RAM with no PF" thread some people have mentioned that game quite a few times.

Yes, the check can be disabled with -disablepagefilecheck, but my point is that there are still some apps that can't run properly without a page file when started normally.

 

However this is not the same as there are some apps that can't run without a pagefile. Period.

Bottom line is Dawn of War can run without a pagefile...