One of the things I've noticed with my new Windows 7 64-bit machine is that apps seem to never give up their "unused" memory. Specifically, Photoshop CS5. Under XP, the Task Manager would show Photoshop CS3 memory going up and down as images were loaded, processed, saved and then closed. When you minimized the Photoshop window, Task Manager would show Photoshop's memory dropping to a very low amount. Under Windows 7, the Task manager shows Photoshop memory only going up, never going down, until you finally close the program. My Photoshop grows to about 3 gig and stays there, even when I close all images within Photoshop and minimize the app. Ditto for the few other apps I track. So far that's no problem because I've got memory to burn (8 gigs), and rarely use over half of it. But I'm curious whether this is a Windows 7 memory management "feature" or whether Photoshop CS5 is different. If I ever reach a point where all loaded apps need more memory than avaialble, will Windows 7 start taking memory away from apps that don't need it anymore?

Have you asked Adobe about this? I would expect the memory management to be within the program while it's open and running.

When the main window of an application is minimized, XP empties the application's working set. This is not specific to Photoshop - you should see the same behavior with most other apps (though some apps like Firefox explicitly override this policy). On Windows 7, minimizing the main window doesn't do anything to the application's working set, and this is by design. If half of the total 8 GB is available there is no reason for the memory manager to reduce the Photoshop's working set, but if you start another memory-intensive app that consumes most of the remaining 4 GB you should see Photoshop's memory usage decrease.

Thanks. I suspected Windows 7 memory management was changed over XP. A related question if anybody knows: My system has not yet used more 60% of its 8GB, yet it still shows thousands of page faults in Task Manager. My understanding of page faults dates back to the 1980's on mainframes and DEC minis. In that era a page fault was a bad thing. It meant a progam needed a mapped memory page that was no longer resident and had to be reloaded from the virtual memory swap file on disk. What exactly is a Windows page fault, any why would they occur on a system that has unused memory.

These are most likely soft page faults, which can be resolved without accessing the disk: http://msdn.microsoft.com/en-us/library/cc441804(VS.85).aspx You can monitor hard faults (which do cause disk accesses) from the Memory tab of the Resource Monitor. Note that even hard faults don't always indicate memory pressure, and are often expected. For example, if an application is using memory-mapped files to read large amounts of data, it will generate lots of hard faults, unless the file contents have already been cached in memory.