SCSIraidGURU wrote:
Applications use both Physical RAM and Paging file. You can disable all
paging and still get programs to function. Applications like many A/V
and CAD apps will not work without paging to swap pages with. The
paging file in Vista does more than just virtual memory.



You don't know what I'm talking about. Read up about physical and
virtual address space, then you'll known.

Charlie Tame wrote:

Yes, I am just saying there are bigger fish to fry and I think there is
nitpicking but no real argument... been there done that (with you) and
it really does not help anyone. We all face a bigger problem with ISPs
now. FWIW I think Microsoft has done a good job trying to stay impartial
and an even better job at keeping the business in friendly territories....

I take my time to respond to your false statement about not being able
to address more than 4GB (mathematically) in PAE mode, and this is what
your return? Don't expect me to respond to anything you have to say in
the future.

Charlie Tame wrote:

But I could NOT use BASIC routines and the
underlying RAM at the same time.

Try and count the number of address pins on a 32-bit CPU that supports
36bit PAE, and tell me if anything needs to overlap

dennis wrote:
Charlie Tame wrote:

Yes, I am just saying there are bigger fish to fry and I think there
is nitpicking but no real argument... been there done that (with you)
and it really does not help anyone. We all face a bigger problem with
ISPs now. FWIW I think Microsoft has done a good job trying to stay
impartial and an even better job at keeping the business in friendly
territories....

I take my time to respond to your false statement about not being able
to address more than 4GB (mathematically) in PAE mode, and this is what
your return? Don't expect me to respond to anything you have to say in
the future.


You seem to have a problem with your attitude Dennis. PAE mode is what I
was talking about, it is an ADDITION to a 32 bit system. As I recall we
were talking 32 bit addressing originally. In general 32 bit has a
physical and mathematical limit of 4GB. Whether MS chose to use it or
not, the fact that PAE could have been used does not change this fact. I
was not being confrontational, rather the opposite in fact, nor did I
make the original statement, however I think for practical purposes
saying that a 32bit OS can only support 4GB is true enough for most
situations, Anything over that in my opinion is an addition or workaround.

Charlie Tame wrote:

You seem to have a problem with your attitude Dennis. PAE mode is what I
was talking about, it is an ADDITION to a 32 bit system. As I recall we
were talking 32 bit addressing originally. In general 32 bit has a
physical and mathematical limit of 4GB. Whether MS chose to use it or
not, the fact that PAE could have been used does not change this fact. I
was not being confrontational, rather the opposite in fact, nor did I
make the original statement, however I think for practical purposes
saying that a 32bit OS can only support 4GB is true enough for most
situations, Anything over that in my opinion is an addition or workaround.

I'm not the one with an attitude. This entire discussion is about PAE
and what is possible in PAE mode. Both XP and Vista includes a PAE
kernel, and in PAE mode, you can use more than 32 bits to define the
physical address. Microsoft just choose not to use any additional bits
to address memory, but it is a choice, not something that is
mathematical impossible.

dennis wrote:
Charlie Tame wrote:

You seem to have a problem with your attitude Dennis. PAE mode is what
I was talking about, it is an ADDITION to a 32 bit system. As I recall
we were talking 32 bit addressing originally. In general 32 bit has a
physical and mathematical limit of 4GB. Whether MS chose to use it or
not, the fact that PAE could have been used does not change this fact.
I was not being confrontational, rather the opposite in fact, nor did
I make the original statement, however I think for practical purposes
saying that a 32bit OS can only support 4GB is true enough for most
situations, Anything over that in my opinion is an addition or
workaround.

I'm not the one with an attitude. This entire discussion is about PAE
and what is possible in PAE mode. Both XP and Vista includes a PAE
kernel, and in PAE mode, you can use more than 32 bits to define the
physical address. Microsoft just choose not to use any additional bits
to address memory, but it is a choice, not something that is
mathematical impossible.


Right, and I think this is where the misunderstanding arose, and why I
said nobody was actually "Wrong" earlier...

I am saying it is impossible if you don't use PAE mode, not that it is
impossible period. I think the OP also intended to say this, knowing
that MS had not used it. I suspect that MS decided that because 64 bit
hardware is now very much available, a serious user with a need for
large amounts of memory would prefer that than to use a mode which is
possibly less reliable due to driver issues etc.

In fact there are probably a number of "Hardware" based solutions to the
4GB limit as well, "Impossible" is probably not quite the right word
ever these days

As Colin pointed out in one of his posts yes you can support more the 4GB of
physical memory with a 32bit OS if you provide different threads/users each
with their own individual logical memory map which is limited to no more the
4GB. 32 bit Server Operating systems have used this technique for years to
support large large numbers of users on large memory systems. Also by
providing separate logical memory maps for different users servers are able
to insure that the programs or data being used by one user can not be
corrupted by another server user.
Dennis is correct when he states that with a 32 bit OS that supports memory
mapping one can develop an application that use more then 4GB of physical
memory by spreading the functionality across multiple processes each of
which is limited to its own 4GB of logical address space.

"dennis" wrote in message
...
Charlie Tame wrote:

But I could NOT use BASIC routines and the underlying RAM at the same
time.

Try and count the number of address pins on a 32-bit CPU that supports
36bit PAE, and tell me if anything needs to overlap

Yes Dennis was correct but so is the 4GB limit if some other technology
is not used. My point was that if "Something else" is not used then the
original statement that there is a 4GB limit was correct.

Curious wrote:
As Colin pointed out in one of his posts yes you can support more the
4GB of physical memory with a 32bit OS if you provide different
threads/users each with their own individual logical memory map which is
limited to no more the 4GB. 32 bit Server Operating systems have used
this technique for years to support large large numbers of users on
large memory systems. Also by providing separate logical memory maps
for different users servers are able to insure that the programs or
data being used by one user can not be corrupted by another server user.
Dennis is correct when he states that with a 32 bit OS that supports
memory mapping one can develop an application that use more then 4GB of
physical memory by spreading the functionality across multiple processes
each of which is limited to its own 4GB of logical address space.

"dennis" wrote in message
...
Charlie Tame wrote:

But I could NOT use BASIC routines and the underlying RAM at the same
time.

Try and count the number of address pins on a 32-bit CPU that supports
36bit PAE, and tell me if anything needs to overlap



Yes he

What keeps getting missed is the reason why a functionality to extend past
the 4GB limit is not available to Windows clients like XP so that us
technology enthusiasts can romp and play in Big Memory Land.

Dennis is saying it could be done and that is true. He is saying therefore
MS chose not to enable it in Windows clients and faults them for not doing
so. That's where I disagree. My take on that point is that MS concluded
that PAE is not the solution they wanted to present to Windows client users
because moving on to 64bit Windows clients was the more robust solution by
far.

There isn't much beyond this point that is getting anywhere in this thread.
The technical woulda, coulda, shoulda stuff really misses the point of why
MS didn't bother to use PAE to address the 4GB limitations in 32bit Windows
clients.

Providing 64bit operating systems is a far better resolution of the issue
than a workaround for 32bit. For that I applaud MS, not fault them.

"Charlie Tame" wrote in message
...
Yes Dennis was correct but so is the 4GB limit if some other technology is
not used. My point was that if "Something else" is not used then the
original statement that there is a 4GB limit was correct.

Curious wrote:
As Colin pointed out in one of his posts yes you can support more the 4GB
of physical memory with a 32bit OS if you provide different threads/users
each with their own individual logical memory map which is limited to no
more the 4GB. 32 bit Server Operating systems have used this technique
for years to support large large numbers of users on large memory
systems. Also by providing separate logical memory maps for different
users servers are able to insure that the programs or data being used by
one user can not be corrupted by another server user.
Dennis is correct when he states that with a 32 bit OS that supports
memory mapping one can develop an application that use more then 4GB of
physical memory by spreading the functionality across multiple processes
each of which is limited to its own 4GB of logical address space.

"dennis" wrote in message
...
Charlie Tame wrote:

But I could NOT use BASIC routines and the underlying RAM at the same
time.

Try and count the number of address pins on a 32-bit CPU that supports
36bit PAE, and tell me if anything needs to overlap



Yes he

Colin Barnhorst wrote:
What keeps getting missed is the reason why a functionality to extend
past the 4GB limit is not available to Windows clients like XP so that
us technology enthusiasts can romp and play in Big Memory Land.

Dennis is saying it could be done and that is true. He is saying
therefore MS chose not to enable it in Windows clients and faults them
for not doing so. That's where I disagree. My take on that point is
that MS concluded that PAE is not the solution they wanted to present to
Windows client users because moving on to 64bit Windows clients was the
more robust solution by far.

There isn't much beyond this point that is getting anywhere in this
thread. The technical woulda, coulda, shoulda stuff really misses the
point of why MS didn't bother to use PAE to address the 4GB limitations
in 32bit Windows clients.

Providing 64bit operating systems is a far better resolution of the
issue than a workaround for 32bit. For that I applaud MS, not fault them.

Nobody is saying that a 64bit OS is not the best solution. It is, also
because the virtual address space is so much bigger than in 32bit. Even
though, most applications are till 32bit and confided to a 4GB address
space when executed in a 64bit environment.

Where we disagree is when you say it is not mathematically possible, and
therefore not a choice Microsoft made. The bits are there, but because
of driver problems, they choose not to use them (this is the reason they
give in their KBs).