Hard drive speed Windows 7 + Page file

[oblivion666]

Keep in mind we are talking about 64 windows 7, so I do not know if the page file or other things are changed compared to xp and vista. I run my windows on a 32MB cache hard drive--- the WD Black 1TB to be exact. I have 6GB of tripple channel RAM at 1800. I was wondering what would be faster for my overall performance with games mainly, should I keep my page file on or off? Should I use the policy in windows 7 Enable write caching on this device and turn off windows write-cache buffer flushing? Heres a photo to show the exact details of the options.

http://i63.photobucket.com/albums/h158/DeathNeedle/a1-1.jpg

So, what would be best for my game performance?

 

[GhislainG]

Keeping the pagefile doesn't hurt performance. If you have a UPS, enable write caching and don't turn off windows write-cache buffer flushing. If you don't have a UPS, then check both options (or none if data loss is not an option). Regardless of the settings, gaming performance shouldn't be affected.

 

[oblivion666]

But, would this slow down my data rates for transfers? Does anyone know if it is better performance to have something like FRAPS on another hard drive instead of the same one?

 

[GhislainG]

But, would this slow down my data rates for transfers? Does anyone know if it is better performance to have something like FRAPS on another hard drive instead of the same one?

Not unless the system requires more memory than is available and swapping occurs. If you use your system for benchmarking a game that creates a lot of disk I/O (which one?), then you could use a dedicated drive for FRAPS. I have yet to read a review where the reviewer stated that a dedicated hard disk was required for FRAPS, but your requirements might be different.

 

[sminlal]

Keeping the pagefile doesn't hurt performance.

It can under certain circumstances. The issue is that Windows will use RAM for caching disk I/O to the point where it ends up paging programs out of memory. This leads to the well-known "post lunch slowdown" problem where a virus scanner, having detected that the system is idle, fires up. After running for half an hour all of the programs end up being turfed out of memory, and when the hapless user comes back from lunch to use the machine it takes forever to get anything started.

I've had this issue on XP when programs like Firefox have been minimized while I run my backups. When I try to restore the Firefox window, it can take upwards of 30 seconds to get to the point where you can do useful work with the program. Disabling the page file solves the problem because it prevents the system from paging the programs out of memory.

So far my experience with Windows 7 has been much better in this regard. Some people have said this is because Superfetch is working behind the scenes to reload the application memory pages after an I/O-intensive program completes. If so, it's a stupid solution, IMHO. Rather than paging the application out in the first place and then having to page it back in again, there should be a way to limit how much memory can be grabbed for file caching in the first place. In most cases a desktop system doesn't need very much file caching beyond getting the most active directories and file system mapping information into memory. There's certainly no reason to cache a copy of every virus-scanned or backed-up file.

 

[cjl]

Keeping the pagefile doesn't hurt performance. If you have a UPS, enable write caching and don't turn off windows write-cache buffer flushing. If you don't have a UPS, then check both options (or none if data loss is not an option). Regardless of the settings, gaming performance shouldn't be affected.

Actually, if you have a UPS, turn off write cache buffer flushing. It will improve performance at the cost of a slightly higher risk of data loss in the event of an unexpected powerdown.

 

[GhislainG]

Actually, if you have a UPS, turn off write cache buffer flushing. It will improve performance at the cost of a slightly higher risk of data loss in the event of an unexpected powerdown.

You are correct.