Eliminating the paging file in Windows 2000

Is it possible to eliminate the pagefile in Win2K? I set it's size to zero but when I reboot the computer I get an error message that the paging file is too small... meanwhile Windows is using 160MB of 700+MB I have installed.

I'd love to force Windows to utilize all my RAM rather than any disk space.

 

It's also not recommended at all. Even if you have a "zillion" megs of RAM. Windows will still use a swapfile. You can alter the minimum and maximum size. There are many suggestions available, some of them even correct. I prefer to let windows allocate the size. It's happy and so am I.

 

This is not something I could ever think of a good reason to do. 90% of paging happens behind the scenes, predictivly and without impact to performance. This could actually make the machine run worse, since you are robbing from the disk cache, and have all kind of things in memory occupying space that you may not be using.

If you could produce a perfmon showing working set trimming cutting into performance, i'd like to see it and know the scenario you used to create it.

 

If you want to improve performance, I heard that you can put the paging file on a different HDD. Is this correct? Second hand drives of around 6-10 Gb can be had for pennies

 

Grunty,

I have mine on my other drive to keep fragmentation of it and the boot drive to a minimum. It accomplishes those goal well.

As for bettering the performance, I doubt being able to tell any difference in most cases.

 

Grunty your question is deceptively simple. The answer requires quite a bit of information about how you use your machine, its configuration and your hardware specs.

Short and grossly oversimplified answer, if you MIN/MAX the page file on a defragged drive to the same number it will not be fragmented (okay maybe a little). The other potential speed boost is to move it to a different physical drive than your \windows or \program files directory. The theory there being that probable contention on the disk comes from loading programs and data, and swapping into pagefile. seperate the two, and your IDE drives on consumer class machines are going to be faster.

The amount of increase a "typical" home user could realize is perhaps 1-15% increase in paging operations, which are usually only 0-5% of system time. So, at best, 15% of 5%... Just because it can be measured, doesnt make it worthwhile.

The above numbers were pulled mostly out of the air. you would need to have large amounts of historical data to make specific calculations for your environment. I'm sure that someone has a scenario where this doesnt hold true, the above rough numbers apply for typical web browsing, light video gaming configs.

 

Putting the pagefile on a seperate HDD in theory as Ben explains is faster. But the second drive needs to be of the same spec or faster for any benefit. You wouldn't be advised to place the pagefile on a 5400 RPM drive if the system drive were 7200 RPM, by way of example. The pagefile should be on the fastes drive for any chance of noticeable performance improvement.

My main system drive is a SATA drive, and I have an ATA 100 secondary drive. Consequently I leave the pagefile on the system drive.

 

I did have a feeling that the stuff I read once on moving the pagefile would be out of date.

It was in some MCSE stuff and most of the literature on that was written for Pentium 2's or 3's. Thats why it still advises on how to create extra volumes from scattered bits of spare HDD space when anyone can get an extra 100Gb for pocket-change these days.

The gaming market has produced machines that will far outstrip the requirements of large page files, but it has been a while since I experimented with anything other than default setup so I just wondered if it still held true.