XP 4GB Limit, but how much do you get to use...

Windows XP (32 bit) can handle / address / map (whatever the correct term
is) 4GB of 'memory'. However if you install 4GB of physical RAM in a system,
it is not all accessible to the user as the operating system uses some of
addresses in the 4GB space for hardware mapping, so cannot simultaneously
map the same address to RAM.

Is there a way of calculating the amount of RAM that would actually be
available to the user, given a known hardware configuration?

I don't expect this is as simple as: 4GB - graphics card memory - 128MB per
PCI card, but there must be some kind of formula along these lines. Anyone
help out here?

Thanks.

"GT" <ContactGT_rem_ove_this_@hotmail.com> wrote in message
news:00253dcf$0$2161$c3e8da3@news.astraweb.com...
> Windows XP (32 bit) can handle / address / map (whatever the correct term
> is) 4GB of 'memory'. However if you install 4GB of physical RAM in a
> system, it is not all accessible to the user as the operating system uses
> some of addresses in the 4GB space for hardware mapping, so cannot
> simultaneously map the same address to RAM.
>
> Is there a way of calculating the amount of RAM that would actually be
> available to the user, given a known hardware configuration?
>
> I don't expect this is as simple as: 4GB - graphics card memory - 128MB
> per PCI card, but there must be some kind of formula along these lines.
> Anyone help out here?
>
> Thanks.
>

I could be wrong but I believe its 4gb total memory - so if you have a 512mb
graphics card then deduct that amount from system RAM
and you'd have 3.5gb usable. If your card is 256mb then you'd have 3.75gb

GT wrote:
> Windows XP (32 bit) can handle / address / map (whatever the correct term
> is) 4GB of 'memory'. However if you install 4GB of physical RAM in a system,
> it is not all accessible to the user as the operating system uses some of
> addresses in the 4GB space for hardware mapping, so cannot simultaneously
> map the same address to RAM.
>
> Is there a way of calculating the amount of RAM that would actually be
> available to the user, given a known hardware configuration?
>
> I don't expect this is as simple as: 4GB - graphics card memory - 128MB per
> PCI card, but there must be some kind of formula along these lines. Anyone
> help out here?
>
> Thanks.
>

This document has an example at the back. It evaluates a 915G based Intel
motherboard, but the PCI graphics card and PCI Express graphics card used are
unstated. Presumably the cards must have small memories on them. Since the
915G doesn't support remapping, as far as I could tell from looking at
the datasheet, WinXP x64 version was not able to change the available
memory readout on their example motherboard. They should have chosen some
different hardware to test, to make the results more worthwhile.

http://web.archive.org/web/200704130...b/4GB_Rev1.pdf

Paul

"Paul" <nospam@needed.com> wrote in message news:g7pet8$a4m$1@aioe.org...
> GT wrote:
>> Windows XP (32 bit) can handle / address / map (whatever the correct term
>> is) 4GB of 'memory'. However if you install 4GB of physical RAM in a
>> system, it is not all accessible to the user as the operating system uses
>> some of addresses in the 4GB space for hardware mapping, so cannot
>> simultaneously map the same address to RAM.
>>
>> Is there a way of calculating the amount of RAM that would actually be
>> available to the user, given a known hardware configuration?
>>
>> I don't expect this is as simple as: 4GB - graphics card memory - 128MB
>> per PCI card, but there must be some kind of formula along these lines.
>> Anyone help out here?
>>
>> Thanks.
>
> This document has an example at the back. It evaluates a 915G based Intel
> motherboard, but the PCI graphics card and PCI Express graphics card used
> are
> unstated. Presumably the cards must have small memories on them. Since the
> 915G doesn't support remapping, as far as I could tell from looking at
> the datasheet, WinXP x64 version was not able to change the available
> memory readout on their example motherboard. They should have chosen some
> different hardware to test, to make the results more worthwhile.
>
> http://web.archive.org/web/200704130...b/4GB_Rev1.pdf
>
> Paul

I tried yesterday and today - can't open that document - link not working.
GT

GT wrote:
> "Paul" <nospam@needed.com> wrote in message news:g7pet8$a4m$1@aioe.org...
>> GT wrote:
>>> Windows XP (32 bit) can handle / address / map (whatever the correct term
>>> is) 4GB of 'memory'. However if you install 4GB of physical RAM in a
>>> system, it is not all accessible to the user as the operating system uses
>>> some of addresses in the 4GB space for hardware mapping, so cannot
>>> simultaneously map the same address to RAM.
>>>
>>> Is there a way of calculating the amount of RAM that would actually be
>>> available to the user, given a known hardware configuration?
>>>
>>> I don't expect this is as simple as: 4GB - graphics card memory - 128MB
>>> per PCI card, but there must be some kind of formula along these lines.
>>> Anyone help out here?
>>>
>>> Thanks.
>> This document has an example at the back. It evaluates a 915G based Intel
>> motherboard, but the PCI graphics card and PCI Express graphics card used
>> are
>> unstated. Presumably the cards must have small memories on them. Since the
>> 915G doesn't support remapping, as far as I could tell from looking at
>> the datasheet, WinXP x64 version was not able to change the available
>> memory readout on their example motherboard. They should have chosen some
>> different hardware to test, to make the results more worthwhile.
>>
>> http://web.archive.org/web/200704130...b/4GB_Rev1.pdf
>>
>> Paul
>
> I tried yesterday and today - can't open that document - link not working.
> GT
>
>

It works for me right now.

You can have a look at this page, and see if any of the links on here
work. I don't understand why web.archive.org does this. I've had failures
like this before as well (give someone a link and it doesn't work).

http://web.archive.org/*/http://dlsv...b/4GB_Rev1.pdf

Paul

"Paul" <nospam@needed.com> wrote in message news:g7s4j1$p3h$1@aioe.org...
> GT wrote:
>> "Paul" <nospam@needed.com> wrote in message news:g7pet8$a4m$1@aioe.org...
>>> GT wrote:
>>>> Windows XP (32 bit) can handle / address / map (whatever the correct
>>>> term is) 4GB of 'memory'. However if you install 4GB of physical RAM in
>>>> a system, it is not all accessible to the user as the operating system
>>>> uses some of addresses in the 4GB space for hardware mapping, so cannot
>>>> simultaneously map the same address to RAM.
>>>>
>>>> Is there a way of calculating the amount of RAM that would actually be
>>>> available to the user, given a known hardware configuration?
>>>>
>>>> I don't expect this is as simple as: 4GB - graphics card memory - 128MB
>>>> per PCI card, but there must be some kind of formula along these lines.
>>>> Anyone help out here?
>>>>
>>>> Thanks.
>>> This document has an example at the back. It evaluates a 915G based
>>> Intel
>>> motherboard, but the PCI graphics card and PCI Express graphics card
>>> used are
>>> unstated. Presumably the cards must have small memories on them. Since
>>> the
>>> 915G doesn't support remapping, as far as I could tell from looking at
>>> the datasheet, WinXP x64 version was not able to change the available
>>> memory readout on their example motherboard. They should have chosen
>>> some
>>> different hardware to test, to make the results more worthwhile.
>>>
>>> http://web.archive.org/web/200704130...b/4GB_Rev1.pdf
>>>
>>> Paul
>>
>> I tried yesterday and today - can't open that document - link not
>> working.
>> GT
> It works for me right now.
>
> You can have a look at this page, and see if any of the links on here
> work. I don't understand why web.archive.org does this. I've had failures
> like this before as well (give someone a link and it doesn't work).
>
> http://web.archive.org/*/http://dlsv...b/4GB_Rev1.pdf
>
> Paul

That link doesn't work either, but if I strip out everything before the /*/
(the web.archive part of the link), then it works fine direct to asus.com!

GT wrote:
> "Paul" <nospam@needed.com> wrote in message news:g7s4j1$p3h$1@aioe.org...
>> GT wrote:
>>> "Paul" <nospam@needed.com> wrote in message news:g7pet8$a4m$1@aioe.org...
>>>> GT wrote:
>>>>> Windows XP (32 bit) can handle / address / map (whatever the correct
>>>>> term is) 4GB of 'memory'. However if you install 4GB of physical RAM in
>>>>> a system, it is not all accessible to the user as the operating system
>>>>> uses some of addresses in the 4GB space for hardware mapping, so cannot
>>>>> simultaneously map the same address to RAM.
>>>>>
>>>>> Is there a way of calculating the amount of RAM that would actually be
>>>>> available to the user, given a known hardware configuration?
>>>>>
>>>>> I don't expect this is as simple as: 4GB - graphics card memory - 128MB
>>>>> per PCI card, but there must be some kind of formula along these lines.
>>>>> Anyone help out here?
>>>>>
>>>>> Thanks.
>>>> This document has an example at the back. It evaluates a 915G based
>>>> Intel
>>>> motherboard, but the PCI graphics card and PCI Express graphics card
>>>> used are
>>>> unstated. Presumably the cards must have small memories on them. Since
>>>> the
>>>> 915G doesn't support remapping, as far as I could tell from looking at
>>>> the datasheet, WinXP x64 version was not able to change the available
>>>> memory readout on their example motherboard. They should have chosen
>>>> some
>>>> different hardware to test, to make the results more worthwhile.
>>>>
>>>> http://web.archive.org/web/200704130...b/4GB_Rev1.pdf
>>>>
>>>> Paul
>>> I tried yesterday and today - can't open that document - link not
>>> working.
>>> GT
>> It works for me right now.
>>
>> You can have a look at this page, and see if any of the links on here
>> work. I don't understand why web.archive.org does this. I've had failures
>> like this before as well (give someone a link and it doesn't work).
>>
>> http://web.archive.org/*/http://dlsv...b/4GB_Rev1.pdf
>>
>> Paul
>
> That link doesn't work either, but if I strip out everything before the /*/
> (the web.archive part of the link), then it works fine direct to asus.com!
>

:-) The "direct to Asus" wasn't working for me yesterday, which is why
I used the archived copy in the first place. The document actually
originates from Intel, but I doubt I could find the source there, even
if it was still available.

Paul

"Paul" <nospam@needed.com> wrote in message news:g7s8pl$b10$1@aioe.org...
> GT wrote:
>> "Paul" <nospam@needed.com> wrote in message news:g7s4j1$p3h$1@aioe.org...
>>> GT wrote:
>>>> "Paul" <nospam@needed.com> wrote in message
>>>> news:g7pet8$a4m$1@aioe.org...
>>>>> GT wrote:
>>>>>> Windows XP (32 bit) can handle / address / map (whatever the correct
>>>>>> term is) 4GB of 'memory'. However if you install 4GB of physical RAM
>>>>>> in a system, it is not all accessible to the user as the operating
>>>>>> system uses some of addresses in the 4GB space for hardware mapping,
>>>>>> so cannot simultaneously map the same address to RAM.
>>>>>>
>>>>>> Is there a way of calculating the amount of RAM that would actually
>>>>>> be available to the user, given a known hardware configuration?
>>>>>>
>>>>>> I don't expect this is as simple as: 4GB - graphics card memory -
>>>>>> 128MB per PCI card, but there must be some kind of formula along
>>>>>> these lines. Anyone help out here?
>>>>>>
>>>>>> Thanks.
>>>>> This document has an example at the back. It evaluates a 915G based
>>>>> Intel
>>>>> motherboard, but the PCI graphics card and PCI Express graphics card
>>>>> used are
>>>>> unstated. Presumably the cards must have small memories on them. Since
>>>>> the
>>>>> 915G doesn't support remapping, as far as I could tell from looking at
>>>>> the datasheet, WinXP x64 version was not able to change the available
>>>>> memory readout on their example motherboard. They should have chosen
>>>>> some
>>>>> different hardware to test, to make the results more worthwhile.
>>>>>
>>>>> http://web.archive.org/web/200704130...b/4GB_Rev1.pdf
>>>>>
>>>>> Paul
>>>> I tried yesterday and today - can't open that document - link not
>>>> working.
>>>> GT
>>> It works for me right now.
>>>
>>> You can have a look at this page, and see if any of the links on here
>>> work. I don't understand why web.archive.org does this. I've had
>>> failures
>>> like this before as well (give someone a link and it doesn't work).
>>>
>>> http://web.archive.org/*/http://dlsv...b/4GB_Rev1.pdf
>>>
>>> Paul
>>
>> That link doesn't work either, but if I strip out everything before the
>> /*/ (the web.archive part of the link), then it works fine direct to
>> asus.com!
>
> :-) The "direct to Asus" wasn't working for me yesterday, which is why
> I used the archived copy in the first place. The document actually
> originates from Intel, but I doubt I could find the source there, even
> if it was still available.
>
> Paul

Computers, eh?