>> Carl, do you monitor your system's memory/file paging needs every
>> few microseconds?
>> Windows does.
>There is imo no need for me to monitor it that often. Peak memory usage
>is easily read from the system ;-)
>> >Maybe I'm dead wrong here?
>> Hey, it can work under some limited circumstances, but it is
>> POOR advice that you're spreading around.
>> >I'm curious about the file-ops you talk about? That is something I'm not
>> >aware of. Could you point me to some current reference-litterature? I've
>> >been searching the MSDN but I cannot find the file-operations you
>mention.
>> When a virtual-memory enabled operating system, such as Windows
>> gets a file/record request, it automatically -assumes- that the file
>> fragment is available in RAM. If it isn't, it issues a 'page-fault' and
>> 'pages out' to the swapfile to find what it needs. If it is there it
>> retrieves it. If not, it issues a disk read to disk controller(s), which
>in
>> turn request the file open/fragment through the FAT system, and
>> eventually to the physical file.
>Doesn't that imply that it has to be swapped out in the first place?
>If there is always 100-300mb ram avaliable, why wouldn't the fragment be in
>memory?
>If you are talking about a fileserver, I'm completely with you, but this is
>single user gameing-machines we talk about, with normally a maximum uptime
>of 10 hours.
>> It's cheaper, performance-wise, to use RAM, or the swapfile to
>> get a data fragment ('page'), than to deal with the hardware
>> and normal FAT file system of the computer.
>Of course the swapfile is positioned in the fastest sections of the disk,
>but Windows (Xp) does the same with all frequently used programs.
>> Your problem is that you have it in your head that the -entire-
>> swapfile is brought into RAM. You stated that in your prior post.
>> That's ***and I pointed that out to you in my earlier reply,
>> but you are apparently oblivious to new knowledge.
>No, I do not think that. It was a misunderstanding, probably due to my bad
>english.
>I said:
>"Why would you want a cache-file that big?
>I mean swapping that amount data from the memory, would make the PC so
>slow... "
>With that I mean that if you fill up your memory, maybe with a huge bitmap
>or a wavefile, that is so big that it needs to be swapped out to disk, the
>PC will be a pain to work with. I do not mean that the swapfile is placed
>into ram, all at once. Obviously the disc-cache is used to put chunks of
>data to free ram. When the cached data is needed, the current ram needs to
>be cached to disk (if the ram is full), before the cached data can be moved
>to ram.
>> From your comments, you have no clue as to how a VM paging
>> system (swapfile) works, and have decided to not use one. That's cool.
>> Just don't post your misguided theories about VM here and not
>> expect to be challenged on their voracity.
>I think I was unclear in my statement.
>My bad.
>I've hope I've explained it better above?
>> >Again I'm sorry if I'm giving someone bad advices. That is something I do
>> >not want to do.
>> Then learn more before recommending that people not
>> use a swapfile.
>> FYI, the largest/fastest systems in the world with incredible
>> amounts of 'RAM' use Virtual Memory.
>> All the time.
>I do not doubt it, but it's not the same hardware as we are using. Neither
>the same apps.
anything about this subject.
I'm through with you.
