Any thoughts on using a high speed 1 or 2 GB flash drive for paging file?
 
If it could be used, HD usage could be greatly reduced and possibly extend the life of a harddrive.  It also make the computer a little quieter.

That's an interesting Idea.  I don't see why it wouldnt work, but I think it would be slower than if you used your HD.  I'm sure there is someone on these forums that has actually tried this and will be posting soon

I'm quite interested in the answer to this.  It depends on the read/write speed of drive ... also, why not just buy more RAM?

I have three gig ram and my system (D805) still uses the page file, even when the ram usage is below 50%

huh, I always thought that the windows page file was where windows put data that it couldn't fit into the RAM ... I know some programs make thier own page files, like photoshop.  The other question would be: Does USB 2.0 or Firewire have a bandwidth greater than SATA 3.0?  And if so, do any flashdrives actually use that full bandwidth?

The thing about NAND flash used in flash drives is that it has a maximum number of reads\writes that can be done, after that it just stops working. Doing what tou suggested might make things faster, but with the enomours number of times windows acceces the page file, its going to stop working very soon.
 
The bandwith of USB 2.0 is 480MBps = 60MB/s while SATA II is 3Gbps = 300MB/s, but that does not mean its faster... usualy a HD can sustain a transfer speed of 25MB/s.

@ smelly_feet:
 I guess this could help make a computer quieter, although I can't hear any of my 3 drives anyway.  I guess it could also benefit drive longevity, but I'm not sure if it would be significant.
 

You're on the right track saying that the theoretical speed of the connecting bus isn't a primary factor, but modern hard drives (7200RPM) can usually average at least 40MB/s over the entire surface of the drive.  Also, though I'm sure the decreased access time of using flash would help for pagefile usage, I'm not sure if it would be worth the tradeoff in throughput.  Fast flashdrives these days have transfer speeds in the 20-25MB/s range, but run-of-the-mill drives tend to be more in the 10-15MB/s range.  Unfortunately, faster flash won't help much since 25MB/s is already approaching the EFFECTIVE throughput of USB 2.0.  If I hadn't lost my flash drive I might have given this a shot, but I doubt the performance benefit would be worthwhile since I already have 2GB of Ram.  Also, transfers over the USB bus require CPU overhead, and the faster the transfer, the more CPU time is used.  This said, it may actually be detrimental to use USB based flash, especially if what you are doing is processor intensive.
 
-mcg

That is a very interesting idea. Don't know if it works. Let me know if it does please.  
 
I have a 1GB SD card that could find second life for this. I also have a 2 GB USB key that could do the same. Never touht of this, but it's a good idea "on paper" at least.  
 
Question to XZezin: I know what you say is true to a certain point, but I personally think the read/write cycle of NAND flash memory shouldn't be a problem. Otherwise Intel, Microsoft and everybody else is dead wrong with their idea of ROBSON and other technique to use NAND Flash memory to improve system performances exactly in a way that works like the one said above, only better on it's implementation. Or maybe i missed something, no offence I hope.

isn't vista going to have an option of doing almos this exact thing to increase proformance on low end systems?

why don't you just turn pagefiling off if you have 3 gigs of ram? set the amount to be used to 0 and no worries.

 
Agreed.

 
Not sure about XP but with win98se, turning the page file off causes problems with applications that expect a page file. My printer for instance, would not print without a page file (crappy drivers). It would complain about not enough memory when there was 500MB of ram sitting there waiting to be used.
 
I solved this by creating a ramdisk and sticking a small 32mb page file on it.
 
So I know this sounds dumb, but maybe creating a ram disk and putting the page file on this, is a better solution than having no page file. It's easy in win98 cos you create the ramdisk before windows starts,but not sure how you would do it in xp.
 
With win98se, I can boot to dos then run win98 in ram. Responsive and quiet, virus resistant, but takes a couple of minutes to load the os into ram. It's my prefered internet pc. A clean load every time.

MS doesn't agree with your conclusions regarding disk vs flash speeds.
 
From the Vista features list:
 
Windows ReadyBoost
 
Adding system memory (RAM) is often the best way to improve your PC's performance. More memory means applications can run without needing to access the hard drive. However, upgrading memory is not always easy. You need to know what type of memory you need, purchase the memory, and open your computer to install the memory—which sometimes can invalidate your support agreement. Also, some machines have limited memory expansion capabilities, preventing you from adding RAM even if you are willing to do so.
 
Windows Vista introduces a new concept in adding memory to a system. Windows ReadyBoost lets users use a removable flash memory device, such as a USB thumb drive, to improve system performance without opening the box. Windows ReadyBoost can improve system performance because it can retrieve data kept on the flash memory more quickly than it can retrieve data kept on the hard disk, decreasing the time you need to wait for your PC to respond. Combined with SuperFetch technology, this can help drive impressive improvements in system responsiveness.

 
...and you believe Microsoft ?
 
Maybe it's true, but a statement from Microsoft is about as believable as a statement from George Bush as far as I'm concerned. Better to ignore them and find more reliable sources. They just confuse the issue.
 
My favourite Microsoft statement, cos it's just so obviously not true, is the claim from Microsoft that Windows XP starts faster than any previous version of windows.
 
I haven't tried any versions before 3.1, but I still have a copy of windows 3.1 on an antique hard drive (less than 1 gb so very slow by todays standards). The last time I booted it up was on a celeron 300 running at 450: it booted in a couple of seconds. I suspect on a modern hard drive it would boot in under 2 seconds... I must try it out again.
 
Every version of windows they bring out is "faster and more secure" apparently. Can't wait for Vista, bound to be blisteringly fast and really secure.

 
...and you believe Microsoft ?
 
Maybe it's true, but a statement from Microsoft is about as believable as a statement from George Bush as far as I'm concerned. Better to ignore them and find more reliable sources. They just confuse the issue.
 
 
There are always people who won’t believe anything if it is inconsistent with their views. I tend to believe that MS actually knows more about Windows and technology than someone on this, or any other forum.

Readyboost does not rely on pure bandwidth between a USB device and the rest of the system to boost performance.  what it DOES though is allow lots of small bits of information to be accessed at the same time.  while not an extreme boost in performance, it would certainly be faster than accessing the HD for the same information as it can only access, aside from what is in its buffer, one request at a time (physical limitation, R/W heads being an actual physical arm).  flash allows hundreds.  
Major boost in performance?  not likely as overall bandwidth is limited.  will it hurt?  dont know why it would.

 
You must also remember that 480MBps on USB2.0 is it's burst speed and it cannot sustain that either.

They do make IDE and SATA connectors for Compact flash cards.  Why do we have to worry about the limits of usb/firewire?

Not that it would be financiallly responsible....I wonder how 8 large CF cards would perform in some raid setup.