minifig

Putting caches on disc seems really strange from design point of view afaic. But then again, I don't work at µsoft. Thanks

The information in the link from technet you send, is something I already know. I am going to look around there some more as it's about performance in general. The problem I have is not usage of a paging file though. My problem is that it seems that a part of the System Cache can also be put in a paging file. If there is 2.8 GB of RAM free and 2.9 GB is in System Cache, then there has to be some of this cache in the paging file which is strange because for system cache to work efficiently, it has to be in real memory. There is no advantage putting caches on disk. The real question is: Are the 3 numbers in the Physical memory tab (Total, Available and System Cache) really in physical memory? Roger

Originally Posted By: DosFreak 2.8(Available)+2.9(System Cache)=5.7-1.7(pagefile)=4gb (Total Physical Memory) Can you tell me why Windows would put 1.7 GB in a pagefile when there is still 2.8 GB available? This means that System Cache also ends up in a paging file? This is even stranger when you consider the fact that there is only 1.8 GB of committed memory. Roger

Originally Posted By: Minifig Originally Posted By: DosFreak How big is your pagefile? About 6 GB. But I don't see how this can be relevant as free memory and system cache are ONLY in physical memory. Roger

Originally Posted By: DosFreak How big is your pagefile? About 6 GB. Roger

Hi, If you look at this image ( taskmanager ) , you'll notice something strange. At least, I think it's strange. What I see is 4 GB of physical RAM, 2.8 GB of free RAM and 2.9 GB of this physical is in a system cache. Then there is 1.8 GB of committed memory. This does not add up correctly. 2.8+2.9 is already 5.7 GB, way over the 4 GB of memory in the machine. And that's without incorporating extra memory of the committed memory in the processes. When I run the performance monitor, I see about 130 MB in Memory/Cache bytes. Can anybody tell me what's wrong? Maybe a bug (searched for it in the knowledge base of µSoft though)? Thanks, Roger