J
John Weiss
Rod Speed said:How odd that we aint seen even a single substantiation from you, ever.
The only "odd" thing is your "we". Though you may be blind to reality, that
is not the case with most posters here.
Rod Speed said:How odd that we aint seen even a single substantiation from you, ever.
Robert Heiling said:Rod Speed wrote
Are you certain that won't be paged, even though a copy remains in ram?
What puzzles me most about this whole discussion is the lack of mention
of the demand that the running tasks place on memory requirements.
Yes. The size of physical ram is less important than the total memory
required by the running tasks for program code and data space. That
figure plays the most important role in the size of the pagefile and
its role has not even been mentioned.
That is obviously going to vary depending upon what use is made of the
systems and which, and how many, applications are being run concurrently.
The only "odd" thing is your "we".
Though you may be blind to reality, that is not the case with most posters here.
Rod said:I in fact said that Win does in fact put stuff in the page file even when
there is plenty of free physical ram, AND leaves it in physical ram,
and it does that in the background, essentially because its never going
to be possible to predict what some app will want physical ram wise
or what the user will choose to run, so you might as well put the best
candidates for the page file in the page file ahead of time, in the
background, so you can just mark that stuff as no longer in physical
ram if there become a need for more physical ram than is installed.
That is done essentially instantly when its already in the page file.
What was being discussed was what Win does page file use wise WHEN
THERE IS ENOUGH PHYSICAL RAM TO NEVER NEED THE PAGE FILE.
Only the user can ever know that, the OS can never be sure what some
app will request, or what the user may choose to run for the first time etc.
So it makes sense to put some stuff in the page file in the background,
just in case more physical ram is needed than is actually installed.
Because what was being discussed was what Win does page file use wise
WHEN THERE IS ENOUGH PHYSICAL RAM TO NEVER NEED THE PAGE FILE.
Duh. Pity that was assumed when we were discussing the use of the page file
WHEN THERE IS ENOUGH PHYSICAL RAM TO NEVER NEED THE PAGE FILE.
Robert Heiling said:Rod Speed wrote
Right. What you state there is my understanding also.
Win pages everything that is pageable.
I thought I was in agreement with you all along.
If you say so. That's not readily apparent from the rhetoric.
Ok, It got mentioned there. Is that what you meant by:
"What was being discussed was what Win does page file use wise WHEN
THERE IS ENOUGH PHYSICAL RAM TO NEVER NEED THE PAGE FILE."?
That was a couple of messages ago that was said.
There was also discussion of which HD to put it on and which IDE etc.
I didn't realize the discussion had narrowed like that.
Whatever! The point is that the size of ram is irrelevant.
You need a pagefile that is large enough to hold the
maximum code & data that your applications requires.
Robert Heiling said:Rod Speed wrote
And summarizing, the answer given appears to be that the location
of the pagefile can make a difference if the pagefile gets used.
Duh! Why didn't I think of that?
Robert Heiling said:And summarizing, the answer given appears to be that the location
of the pagefile can make a difference if the pagefile gets used.
Duh! Why didn't I think of that?
Bob
Rod said:Thats always been obvious. What was being discussed was WHEN
THE LOCATION OF THE PAGE FILE CAN MAKE A DIFFERENCE.
See "Virtual Memory in Windows XP" http://aumha.org/win5/a/xpvm.htm
A snippet (see the link for full context) says:
"For any given workload, the total need for virtual addresses will
not depend on the size of RAM alone. It will be met by the sum of RAM
and the page file. Therefore in a machine with small RAM, the extra
amount represented by page file will need to be larger, not smaller,
than that needed in a machine with big RAM." etc etc etc.
How odd that we aint seen even a single substantiation from you, ever.
Seems like even if someone provides substantiation,
I am the OP.
Way back in the original post it says I run XP.
That is true if we consider the jobs the system is running to be a fixed variable.
We often don't - often if a system isn't running huge jobs, there is no money
(wasted?) put into massive amounts of memory for it, so when a system is
configured per job and has a lot of memory, that is typically an indicatio of
the potential to need more virtual memory space too (but even this will vary
depending on exact jobs and the way the app decides how large a space to
reserve for it's use).
Wrong, as always. The answer can obviously
cover all the possibilitys and so be complete.
Hardly ever in practice.
Can you prove that beyond a shadow of a doubt?
Now that Rod's involved we can't trust much of anything. ;-)
The generic answer is to put your pagefile on the most used
partition of the least used, *modern performance level*,
drive. Put that drive on the PATA controller channel that
is least used simultaneous to any other I/O that might be
occuring while the system is actively making use (reading or
writing) of virtual memory.
kony said:Yep.
trying to divide access between drives and channels is
offset by the performance difference between all drives.
You can get better performance from two new 400GB drives on the same
channel than one 400GB on channel 0 and one 40GB on channel 2, for example.
Almost always in practice.
Seldom does one buy a boatload of identical PATA drives
then later reconfigure them all on the same system, unless
that system is just a filestore instead of actively used XP PC.
Its true even if we dont.
Never ever could bullshit its way out of a wet paper bag.
The real generic answer is to ensure you have enough physical ram
so the page file isnt used because there isnt enough physical ram.
THAT provides a VASTLY better performance config.
kony said:Yep.
otherwise one expects a system that runs larger jobs
to have been properly equipped for these tasks.
Since size of job obviously effects amount of real and virtual memory used,
it would have to be a fixed variable or else "it depends" on that variable.
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.