Notably the standbylist is ram that can be used with its original data or can be immediately zeroed and reasigned as it resides both in memory and has been paged out into the pagefile, this means any memory assigned to the standby PFNDB list will be immediately sent to storage, this is good, however this means there is a second pfndbl, the modified list storing, aka what used to be in the standbylist but was modified without having been backed again, if windows incurres a #pf and decides to use a modifiedpage it means it has to send that memory to disk, which takes time and times out the current thread for an irp, the good thing is that this all doesnt happen all to often, it happens when you have more memory in commit than physical available, so always have some free ram at your disposal, and then there is the other aspect of the argument just cause windows runs on 4gb's or less doesnt mean it wants to, the reason here is that this means it will have to more often clean out the modified list, you can manually flush the modified list via undocumented api calls to the kernel iirc ZwSetInformationSystem, doing this on a highend system can actually stall the system for a few seconds, as cleaning that list is not a very light process
just some windows memory managment trivia)
