XP Pagefile on a RAM disk?

[Turbo Data Weasel]

yet you're on a Windows enthusiast forum.

I should read some of the other threads here, because from this one it's not obvious to me that this is an enthusiast forum. I would think the enthusiast's response to the ideas kicked around here would have been more like:

"A pagefile on a ramdisk... Interesting... I wonder what you could do with that... How about ..."

"Yeah... I tried that... Unfortunately that doesn't help, because <details>... But maybe you could do this?"

"Hey! You can put the ramdisk into inaccessible memory using PAE! Then you get it for free without loosing memory!"

"Holy crap! For real? You mean I can get access to all of my 8GB of memory in 32-bit Windows??? That's awesome! How does that work?"

Here's the one I'm missing most: "COOL! I want to play with this! How do I do this again?"

...

Instead we have:

"Yeeaahhhh... That's possible... but it's dumb..."

"Windows really sucks about memory management!"

"No it doesn't!"

"Get Vista!"

"I'm ignoring you all!"

"Well, that's mature!"

<facepalm>

...

This is sort of the response that "Cold Fusion" gets from other Physics "enthusiasts"... So maybe this whole thread really is about a crackpot idea...

The thing that surprises me is: If the entire internet was littered with people having tried pagefiles in the inaccessible memory with no benefit, I would understand the reaction, but this does not seem like a well-explored concept. 95% of "experts" still claim that there is "absolutely no way to address more than 3.5GB of memory in 32-bit Windows", which I know is FALSE because I'm DOING IT RIGHT NOW!

For example, I would really like to get to the bottom of Brandon's claim that what I'm doing is dangerous. I really don't understand why. I'm not saying he's wrong. But - as an enthusiast myself (I WROTE a DOS virtual disk driver once!) - I would like to LEARN more about this issue (independent of whether it turns out to be useless). Given the density of pseudo-experts on the web, statements like "This is dangerous/impossible/stupid! (vague/no explanation)" are useless, because too many of them turn out to be just plain WRONG (see above). So, while I am not saying I wouldn't like to be able to just trust Brandon on this, I would like to have a bit of an explanation, because as of right now, all real-world data that I have points at him being wrong (but this might change one day when my system goes down in a giant fireball). Let's be scientists about this and work based off DATA or at least real API documentation rather than opinions and hear-say. Especially since it's soooo easy to get data (takes 5 mins to set all this up) and the docs are freely available!

Cheers,

~ a very sad TDW ;)

[soldier1st]

As I'm sure you know, people who POST that they are putting people on their "ignore list" are simply looking for attention. Why announce it? ---- just do it!

I Agree with you on that.

[The_Decryptor Veteran]

I should read some of the other threads here, because from this one it's not obvious to me that this is an enthusiast forum. I would think the enthusiast's response to the ideas kicked around here would have been more like:

...

I don't see why, just because an idea is presented doesn't mean everybody has to love it and want to do it, especially when it's just a way of preventing the inevitable.

People don't go nuts over NAT either.

[Panacik]

I should read some of the other threads here, because from this one it's not obvious to me that this is an enthusiast forum. I would think the enthusiast's response to the ideas kicked around here would have been more like:

"A pagefile on a ramdisk... Interesting... I wonder what you could do with that... How about ..."

"Yeah... I tried that... Unfortunately that doesn't help, because <details>... But maybe you could do this?"

"Hey! You can put the ramdisk into inaccessible memory using PAE! Then you get it for free without loosing memory!"

"Holy crap! For real? You mean I can get access to all of my 8GB of memory in 32-bit Windows??? That's awesome! How does that work?"

Here's the one I'm missing most: "COOL! I want to play with this! How do I do this again?"

...

Instead we have:

"Yeeaahhhh... That's possible... but it's dumb..."

"Windows really sucks about memory management!"

"No it doesn't!"

"Get Vista!"

"I'm ignoring you all!"

"Well, that's mature!"

<facepalm>

...

This is sort of the response that "Cold Fusion" gets from other Physics "enthusiasts"... So maybe this whole thread really is about a crackpot idea...

The thing that surprises me is: If the entire internet was littered with people having tried pagefiles in the inaccessible memory with no benefit, I would understand the reaction, but this does not seem like a well-explored concept. 95% of "experts" still claim that there is "absolutely no way to address more than 3.5GB of memory in 32-bit Windows", which I know is FALSE because I'm DOING IT RIGHT NOW!

For example, I would really like to get to the bottom of Brandon's claim that what I'm doing is dangerous. I really don't understand why. I'm not saying he's wrong. But - as an enthusiast myself (I WROTE a DOS virtual disk driver once!) - I would like to LEARN more about this issue (independent of whether it turns out to be useless). Given the density of pseudo-experts on the web, statements like "This is dangerous/impossible/stupid! (vague/no explanation)" are useless, because too many of them turn out to be just plain WRONG (see above). So, while I am not saying I wouldn't like to be able to just trust Brandon on this, I would like to have a bit of an explanation, because as of right now, all real-world data that I have points at him being wrong (but this might change one day when my system goes down in a giant fireball). Let's be scientists about this and work based off DATA or at least real API documentation rather than opinions and hear-say. Especially since it's soooo easy to get data (takes 5 mins to set all this up) and the docs are freely available!

Cheers,

~ a very sad TDW ;)

We are enthusiastic only about things that are new.

Imagin for a moment, that this has been tried, tested, requested and put down a zillion times on the forums.

Why would people reiterate how pointless it is as it has been tested etc before?

As far as i can see, the original comment was, "Why do you want to do this? There is no real need", the normal response should be, "because i want to achieve a, b, or c". Instead we see the comment, [Vicky pollard] "OMG, I cant believe you just said that! Yeah, well, right, you cant tell me that, because Jason Nevis and Shirlly Bassy where down the road having a fling with a fish and got their knickers in a twist, and you said that and made it all worse!" [/Vicky Pollard].

Now take said response and reason as to why we should then care to take the time to help someone who has obviously decided they know better and will do it anyway?

Do yo understand my point?

[Fubar]

loved the vicky pollard section so much i had to put it in my sig , just brilliant lol

[Turbo Data Weasel]

We are enthusiastic only about things that are new.

Imagin for a moment, that this has been tried, tested, requested and put down a zillion times on the forums.

I guess I had gotten the wrong impression about the topic during my "journey" with it. There definitely are a bunch of forums that discuss the idea of putting the pagefile on the ramdisk. And on most of them, the initial post dates back a year or more. So, I agree, the idea isn't too new. What prompted me to assume that people might still care, though, are two things:

First: On pretty much every forum (including here) this topic blew up into a religious discussion about memory management with plenty of conflicting "expert" opinions, for example about whether you can turn off the pagefile altogether or not. This issue really does not seem understood (at least there is no consensus) and I have not found a forum where there is actual "closure" like cites to MS publications that settle the issue or instructions on how to reproduce an effect. Ideally, this would then be supported by whoever was wrong saying something like: "You know... you were right. I TRIED it and it does this..."

Second: People don't seem to know about the whole PAE business of being able to push the ramdisk into inaccessible memory. THIS is what I believe takes an "UTTERLY pointless" idea ("trade memory for RAM pagefile") into something useful ("add RAM pagefile for free and use all your memory"). I thought that this was the greatest thing since sliced bread! Because now I have an 8GB 32bit machine that actually allows me to USE all 8GB! This is barely mentioned anywhere. And wherever it's mentioned, it spawns another religious discussion about PAE: "It's enabled by default!", "No, it's not!", "It changes the way windows operates and needs drivers that support it!", "Not on my machine!". Not to mention the "experts" who claim that there cannot possibly be any way that any 32bit operating system can access beyond 4GB (WRONG: Cite: MS AWE API; Instructions: See my very first post).

Now take said response and reason as to why we should then care to take the time to help someone who has obviously decided they know better and will do it anyway?

I missed that response... Maybe it's in a different thread on this forum. So far I've only read this thread.

I don't know whether you are referring to me with "someone who decided that they know better". I was specifically trying to avoid that. I have a LOT of experience with programming and even messing with Windows. But that does not make me an expert on this topic since I don't know too much about how Windows manages the pagefile, and I have no clue about how PAE works. I do have some objective data, though: Things are WORKING RIGHT NOW on my machine that some people (mostly on other boards) claim are impossible. I am sure you would agree that half of all "expert" opinions on the internet are unfounded. I was specifically trying to NOT position myself as an expert, but instead to limit my statements to observations and documentation. Beyond that, I was just asking questions.

Quite frankly, posts like Brandon's annoy the **** out of me! Because now I have doubts in my head about the "safety" of what I'm doing, even though everything seems fine. I don't understand WHY it should be dangerous. Therefore, there's nothing I can do to fix a potential problem to make it safer. Given the information, I cannot agree with or trust Brandon. Not because I know better, but simply because he hasn't given me any information to support his point. Maybe it's because I'm working on my PhD in Physics, where we get trained to question statements until we UNDERSTAND WHY they were made. I find Brandon's post as it stands now entirely counter-productive. The annoying part is: If he was to add three lines about WHY he made the claim and where he gets his information, his post would be the most USEFUL of all of the one's I've gotten here so far. NOTE: I am NOT saying that I KNOW he's wrong! All I'm saying is "Tell me WHY you said this!".

Do yo understand my point?

I do. I didn't see any of the other threads in this forum talking about the same thing (I didn't look).

The reason I posted here was because this forum is the FIRST Google hit for "pagefile ramdisk" and as such it would be soooo useful if this thread was objectively discussing the issue and presenting a final solution. (My posts taken together at least describe now HOW to set this all up.)

Also, for people like me who came here via this search, THIS thread is the first impression they get of neowin.net. If the site owners are reading this, here's some free "consumer research" data. I'm going to be honest. For me this thread gives the following impression:

- This page tries to provide CONSERVATIVE TECH SUPPORT. It discourages experimentation in favor of stability and serviceability. Don't get me wrong! This can be VERY useful for MANY people. I just wouldn't call it an "enthusiast" site.

- Some guys here have tried certain things and thus seem to know about the issue. This could be very useful if it's possible to get them to share their insights! Unfortunately, those guys are for some reason disgruntled about the topic and just blow you off with "Don't do it!". Maybe if I add another piece to the puzzle (PAE) I can make the topic interesting again. failed...

- It's a very entertaining site (I LOVE the facepalm picture!). It'll probably make for a good discussion and maybe something will come of it. (So far, it didn't).

Overall, my current conclusion is:

This is the WRONG board for enthusiasts trying to make Windows do something that it wasn't necessarily intended to do, but if I ever have a mainstream problem like "my background picture is all small in the middle, how do I make it, like, fill the screen", this is the right place.

There's probably no point in venting here. It's just so frustrating that people with TRUE experience (e.g. allan who actually ran benchmarks) on forums everywhere get lost in the crowd of pseudo-experts who just have very strong opinions about something. This is Computer SCIENCE we're talking about here! There should be absolutely NO NEED for religious discussions along the lines of evolution vs. intelligent design! The information is out there! This is NOT a mystery! All data is available somewhere! And most of all: It's SOOOO easy to reproduce experiments! Argh!

I need to stop ranting and raving... Thanks to me, the first impression now probably is: "Man, these people need help... What a joke..."

~ TDW

PS: Question for the philosophers: Does having your page be the FIRST hit in Google for a search on a certain topic imply an obligation about addressing that topic in a useful way? Please vote and share your opinion and you will be entered for a drawing to win a RAMDISK (the driver, not the GigaByte hardware thing)!

[+BudMan MVC]

Yup something we can agree on - thanks to you, our first impression of you is YES you need HELP!! Lots of it, more then likely pharmaceutically based to start with ;)

As to this thread being the first hit on google -- having some obligation with it? What drugs are you on? Its a message board where people express opinions on topics.. Yes in this area they are suppose to be about supporting Windows NT/2k/XP clients.. That is all, there is no obligation of anything more than that.

You might be enthusiastic about placing your page file on a ram drive.. Most everyone else on the planet with a few brain cells to rub together and some common sense have already figured out such a thing is pointless and a waste of time.

But sure you have fun with it.. But as you can see from the rest of the internet with a simple search, most everyone else on the planet has already figured that out. Why you feel the need to rehash nonsense?

But sure if you want to tinker go right ahead.. Just don't feel bad when the rest of community does not feel your enthusiasm.

[Turbo Data Weasel]

As to this thread being the first hit on google -- having some obligation with it? What drugs are you on? Its a message board where people express opinions on topics.. Yes in this area they are suppose to be about supporting Windows NT/2k/XP clients.. That is all, there is no obligation of anything more than that.

I was simply posing the question and saying that I thought it "would be useful". The whole "obligation" thing was a somewhat unrelated, very general thought that came to me while writing. That's all. I wonder if it's possible (or already in the works) to get a PhD in philosophy examining that question. Probably so.

I'm surprised you didn't agree with more points from my post besides me needing help. ;)

After all, you are almost the only one on this thread who was consistently trying to reference docs and data from the beginning. I really appreciate that. So I'm a little bummed that you think I'm on drugs.

As an aside:

Without implying anything about your or my viewpoint on the following question or the implications of such viewpoints, I would be interested to know what you think about this related question of much (in my view legitimate) philosophical debate:

"People in Africa are dying of hunger and AIDS by the millions. The Western world, specifically the US, has (somewhat easy) access to food, money, medication, and knowledge which could help alleviate the problem. Does this create an obligation to help?"

Sooo many things to ponder.... Isn't this a marvelous world?

~ TDW

[Panacik]

I was simply posing the question and saying that I thought it "would be useful". The whole "obligation" thing was a somewhat unrelated, very general thought that came to me while writing. That's all. I wonder if it's possible (or already in the works) to get a PhD in philosophy examining that question. Probably so.

I'm surprised you didn't agree with more points from my post besides me needing help. ;)

After all, you are almost the only one on this thread who was consistently trying to reference docs and data from the beginning. I really appreciate that. So I'm a little bummed that you think I'm on drugs.

As an aside:

Without implying anything about your or my viewpoint on the following question or the implications of such viewpoints, I would be interested to know what you think about this related question of much (in my view legitimate) philosophical debate:

"People in Africa are dying of hunger and AIDS by the millions. The Western world, specifically the US, has (somewhat easy) access to food, money, medication, and knowledge which could help alleviate the problem. Does this create an obligation to help?"

Sooo many things to ponder.... Isn't this a marvelous world?

~ TDW

On a side note, your writing style is the same as another, now ex member, of Neowin, who has a desire to create multiple accounts, and take quotes from peoples posts and wrote long replies to each section, which actually make people switch off after reading a line or two...

[Panacik]

I missed that response... Maybe it's in a different thread on this forum. So far I've only read this thread.

Let me show you:

Below is the first response given. He even says there is no real advantage. He does not imply he does not want to give a reason why not. In fact, knowing allan, im sure he would happily give an explination, IF it was asked of him.

Just change the setting to 768 Mb min and max and you'll be fine. But to answer your question yes, you can put the PF on a ram disk, but unless you are very short of hd space there's really no advantage.

Now we see the reply from the OP. This indicates he will do what he wanted anyway. Nothing wrong with that, but he asked a question, was given an answer and then decided to do it anyway.

Just around 2-300MB. It's a laptop with a 5400rpm disk which is fairly slow, so instead of copy data from the disk to the pagefile (on the disk), wouldn't it be faster to copy it to the RAM?

I'm only using windows for playing CS:s, BF2, and some other games. Otherwise i use GNU/Linux. But i'll try to disable the pagefile instead... If i only have 300MB for it, and if then and the RAM is full, i think i get the same result as disabling it: Probably a crach. ;)

Oh, and PLEASE stop "quoting" "everything" tis "most" annoying.

[Turbo Data Weasel]

Let me show you:

I misunderstood you. I was looking for the Vicky Pollard thing and couldn't find it. Sorry.

But, honestly, that reply by the OP reads to me as if he took the suggestions to disable the pagefile instead of putting it on a ramdisk to heart. His comment about the crash reads to me like: putting the pagefile on the ramdisk will be no better than turning it off, because once the ram is full (for either case) the computer will probably crash (as in kill apps with an insufficient-memory-error) just the same. So going with disabling the file makes more sense.

I'll stop the quoting, and also the caps. I think one of the reasons I was catching so much flak here was because my posts read like I'm angry and accusing, which is actually not the case. This is the first time that I was trying to get involved in any message board ever (great start, huh?) and it looks like I still have to find a style that works. Are italics and bold ok? Is there an associated meaning? (Like the SHOUTING, which I hadn't kept in mind for my posts)

I hope I didn't frustrate too many people here. Although, I'd still like to get more data from Brandon! ;)

Keepin' it shorter this time,

~ TDW

[tm1]

TDW, do you want to place pagefile.sys to RAM drive with compression or not ?

There is only advantage when using compression (to reduce size of swapped out memory pages), I think.

Yesterday I does some experiments with help of QSoft RAMDisk (128 meg drive formatted using NTFS with compression enabled). When I place 120 meg pagefile.sys at this RAM drive it's "compressed" attribute is cleared by Windows itself. By the way it works as is, without compression.

After reboot I am prepared disk image formatted NTFS (compression enabled) with size about 128 megs, and place manually pagefile at it with size 240 meg, entirely filled by same byte (nonzero) and "compressed" attribute set. It really takes about 15 megs space of the drive. Then with help of regedit I'm adjusted size of pagefile from 120 to 240 (remember, RAM disk size is only 128 meg), configured RAMDisk to load this disk image at boot of driver and rebooted system. Unfortunately, Windows (XP Pro SP3) is not booted in "normal" mode, only in "safe" mode with RAM drive disabled, of course. So I conclude that this trick with compressed pagefile was wrong way to achieve results, because Windows rejects it, not because it's not possible.

[Turbo Data Weasel]

After reboot I am prepared disk image formatted NTFS (compression enabled) with size about 128 megs, and place manually pagefile at it with size 240 meg, entirely filled by same byte (nonzero) and "compressed" attribute set. It really takes about 15 megs space of the drive. Then with help of regedit I'm adjusted size of pagefile from 120 to 240 (remember, RAM disk size is only 128 meg), configured RAMDisk to load this disk image at boot of driver and rebooted system. Unfortunately, Windows (XP Pro SP3) is not booted in "normal" mode, only in "safe" mode with RAM drive disabled, of course. So I conclude that this trick with compressed pagefile was wrong way to achieve results, because Windows rejects it, not because it's not possible.

Hey tm1,

thanks for trying this out! I somewhat expected that Windows wouldn't be willing to compress the file. I don't know too much about NTFS compression, but it seems like it's highly optimized for speed. For example, when I a put the disk image for a VirtualBox machine on an NTFS compressed disk and install an OS, the file grows steadily as expected, but the free space on the drive seems to jump around quite a bit. This makes me think that Windows saves the data as quickly as it can (uncompressed) and then compresses things in the background later. Since this probably causes quite a performance hit, I would expect compression to be disabled for a pagefile where every ms counts.

I am currently running 8GB of memory in my 32-bit WinXP system, which, of course, is way more than I ever use. I used a PAE enabled ramdisk driver to create a 5.25GB ramdisk in the memory that my 32-bit Windows can't see anyways. On that I have my pagefile which never really gets much use (but frees up space on my small SSD harddisk). So, I don't think I'd gain too much if I could compress it. It was more just a crazy is-this-possible-in-principle thought. I think at some point (back in the Win3.1 times) there were tools that claimed to be able to compress your ram. But I never played with them.

What's fun, though, is to run a VirtualBox machine off the ramdisk. It boots Win2K in 20 seconds (10 of which are network prep) and is actually much more zippy than my host computer. HD Tune inside the VBox reports average reads of 350MB/s!!! It's NUTS! :)

Oh, and, Brandon: My system still seems very stable! ;) (Maybe he'll get annoyed enough to answer me at some point unless he already blocked me...)

Cheers,

~ TDW

[Udedenkz]

Alright skimming through this thread I found a few opinions,

1 - People like me that disable the pagefile and see no problems

2 - People that say that disabling the pagefile only disables the ability to manage it

3 - Crazy People Who Run the PF on a RAMDISK

Personally I think that the second group does not know what it is talking about. I disabled PF on both computers, rebooted deleted that big file, and nothing bad happened. Ofcource if you run out of RAM, windows will kill that app - so no running leaky software.

Also I think that putting the PF unto a RAMDISK will make it slower that just putting it into RAM directly. To evaluate, my superspeed RAMDISK only reads at 1380MB/s avarage.

[throwexception]

I am currently running 8GB of memory in my 32-bit WinXP system, which, of course, is way more than I ever use. I used a PAE enabled ramdisk driver to create a 5.25GB ramdisk in the memory that my 32-bit Windows can't see anyways. On that I have my pagefile which never really gets much use (but frees up space on my small SSD harddisk). So, I don't think I'd gain too much if I could compress it. It was more just a crazy is-this-possible-in-principle thought. I think at some point (back in the Win3.1 times) there were tools that claimed to be able to compress your ram. But I never played with them.

What's fun, though, is to run a VirtualBox machine off the ramdisk. It boots Win2K in 20 seconds (10 of which are network prep) and is actually much more zippy than my host computer. HD Tune inside the VBox reports average reads of 350MB/s!!! It's NUTS! :)

Cheers,

~ TDW

I wanted to congratulate you on your setup, and I would very much love to get more info on how you did it.

I've read several places on the net about similar configs, some people used SuperSpeed ramdisk; some used Gavotte Ramdisk. The idea is the same -- use the Ramdisk to map memory that couldn't be addessed by 32-bit Windows XP; and then put Pagefile onto that ramdisk. Superspeed calls it "unmanaged memory", I see the term PAE in this forum.

The net result is that without changing OS, messing with drivers, application compatibilities, motherboards, etc, they can now utilize more memory than they could without doing this.

I understand that compared to a full 64-bit OS; this still limits each process to 4GB of memory total -- but that's a LOT OF MEMORY for ONE process even with today's standards!

This is a non-trivial problem for me. I very much would love to do this. My PC has too many legacy hardware and software that the manufacturers may not even be in business anymore; and I would like to stretch out it's life at least until Windows 7 comes along or reasonable hardware replacement arrives.

For various reasons, my business is also choosing to skip Vista installation and wait, so it's on XP as well.

In case you wonder why I need so many processes with so much memory? I do server development work, and run huge sized processes. The processes are already "trimmed down" production services. I would like to develop/debug in the same OS as my company's to minimize compatibility issues; so that's another reason to stick with XP instead of something else.

Having ability to run all these 32-bit processes in 1 box will speed up troubleshooting and proof of concepts for me tremendously, without having to acquire multiple server boxes, install them, etc etc.

The ramdisk is the most elegant solution.

Another less optimal solution I'm considering as an alternative to install Linux 64bit and then run VMware instances of 32bit XP of 4GB RAM each. That's a far less elegant solution to achieve the same purpose -- to have more 32-bit processes running inside 1 PC without penalties associated with swap space trashing. It's better isolation wise; but it's also wasteful, as having a VM layer simulating hardware; and having multiple XP OS kernels running. For what I'm doing right now, the isolation benefit isn't helpful and starting / managing multiple VMs is definitely a chore. Plus, I'll still lose access to my legacy 32-bit era hardware which I can't easily replace.

Until hardware and software go to 64 bit reliably -- maybe when Windows 7 comes out -- I don't see any other non-disruptive solution.

[Turbo Data Weasel]

... I would very much love to get more info on how you did it.

There were really only two things that gave me trouble when setting this up:

a) Finding the Gavotte Ramdisk (if that is the one you would like to use; works great for me)

b) Realizing that a pagefile cannot be on drive B: (no matter what type that drive is)

The solution to the latter is easy: don't make your ramdisk drive B:.

The solution to the former is buried in this thread in all the ranting and raving. To spare you the searching, here it is:

The Gavotte Ramdisk is (supposedly) based on the M$ Ramdisk.sys sample driver available here:

FILE: Ramdisk.sys sample driver for Windows 2000

It turned out to be EXTREMELY difficult to actually find the Gavotte version. I finally dug it out of a Japanese hosting page (Thank you, Google Translate!) that you can find here:

Gavotte_RAMDisk_1.0.4096.5_200811130.7z

To download it, you need to "solve" the captcha in the top right corner of the page and click the long button right below the text field. You will then be directed to a page with an ad and a big button in the bottom middle that counts down 40 seconds or so. After, the button turns blue and you can click it to download the file.

I am currently running this version of the ramdisk on my system and have had no problems with it. Avast! does not seem to object to it either, BUT: I DO NOT KNOW WHO WROTE THIS DRIVER AND WHETHER IT IS MALICIOUS OR NOT! SO, USE IT AT YOUR OWN RISK!!!!

Setup could not be easier (other than the page file business). This page describes how:

Gavotte Ramdisk - Free virtual harddisk

There's one extra setup-step that's not explained in the page listed above: The zip file contains a file called "ram4g.reg" you need to execute it to add a parameter to your registry that tells the driver to use PAE to push the ramdisk into the inaccessible memory. The drive will then be called RamDisk-PAE (might require reboot first). BTW: I did NOT need to do anything to enable PAE in Windows. It seems to be enabled by default if your system supports it.

After the ramdisk is running, simply use Control Panel -> System -> Advanced -> Performance -> Settings -> Advanced -> Virtual memory -> Change (phew....) to move your pagefile onto the ramdisk.

Reboot... Done.

If you want to be sure that it all worked, temporarily turn off Hide protected operating system files under My Computer -> Tools -> Folder Options... -> View -> Advanced Settings and make sure that only your ramdisk contains a file called pagefile.sys in the root directory.

Good luck and let me know if you have any further questions.

Cheers,

~ TDW

PS:

The ramdisk gives me 4.5GB/s (yes, that's GB, not MB) for sequential reads. I love it! The one thing that blows my mind, though: My Intel X25-E SSD gives me BETTER performance for random reads with block sizes up to 1KB (both around 10MB/s)! So, maybe the Gavotte driver isn't quite optimized yet. Or the AWE API is slow at remapping memory. I would be interested in seeing whether the SuperSpeed ramdisk performs better (if it works for this at all). One day I might try it, but for now I'm happy with what I have.

[Emergent Comprehension]

Just want to say thanks to TDW.

He seems to provide sound reasoning and balance.

I shall be trying out a RamDisk in XP 32-bit with 12GB of ram.

Like a previous poster I need to run many huge apps, and also run as many parallel c++ linker processes as memory will allow.