Right. I agree with you there. As I said, I did contact the DIMM manufacturer
and get the exact timing and voltage to use with this memory as well as this
motherboard. They also said that each DIMM is embedded with the SPD which
tells the BIOS which timing and voltage to use. In this case, they are cas
latency 5, ras cas 5, row precharged delay 5 and voltage 1.8. I even hard-set
these in the BIOS to override the SPD with the same results. To me, this
eliminates the SPD as the problem. The BIOS had read it properly and set it
for the DIMMS in the first place.

And, I contacted AMD, the manufacturer or the Phenom 9500 in order to
investigate any issues there - with the memory controller or something else.

ASUS, Kingston and AMD all say that the BIOS is properly reading the SPD in
the DIMMS and utilizing all the RAM. They each say this is an OS issue.
Microsoft will not discuss the issue with me for less than $59 despite the
fact that I paid full price for both Microsoft Os' in use on the computer.
When one pays $399 for an operating system, they should get better treatment.

This is the first incident I ever brought here. All previous issues that
MIGHT be Microsoft-related I handled alone, with friends, or with Microsoft
in the first 90 days. For one thing, although I have been using computers
almost daily since 1978, I am not a good typist and it takes a long time to
write these messages and remove most of the errors.

I had hoped to run into someone with my same motherboard, cpu, memory and
Windows Vista Ultimate SP1 64-bit edition, to see how they make it all play
nicely together. That would have been great, but since it didn't happen, I
get to deal with real life on life's terms.

Frankly, all I am left to do is to examine the issue from all possible
sides, and try those solutions which are the least destructive and seem most
likely to work.

"Colin Barnhorst" wrote:

The problem is timings while accessing the memory. The BIOS is simply
reporting the presence of devices compatible with the mobo. The BIOS is not
predicting success of an installed OS.

For mobo manufacturers and cpu makers, it is a real estate problem. That is
why I call it the "fourth damn dimm" problem sometimes. The electrical
distance out and back from the memory controller to the furthest dimm
determines the maximum frequency that can be supported and also sustain
stability. There is a point at which the memory controller is overwhelmed
or cannot keep up. Slightly increasing the dram voltage (no more than 0.1v)
can help in a marginal situation as can lowering the dram speed if the
highest rated ram for the mobo is in use. I think you are already using 667
ram which is a very comfortable dram speed for current boards so I don't see
that as an issue. But the memory controller is integrated on the Phenom so
while the cpu is compatible with the board and 8GB of ram is compatible with
the board, is your model Phenom AND 8GB of your ram compatible on your mobo?
That is the question. It is never the individual component specs but the
combination that is such a headache to work out at times.

"MrKit" wrote in message
...
On the surface I would agree with you about ASUS. They can and do make
mistakes or misrepresent things in order to sell their products sometimes,
or
to keep from fulfilling their duties as spelled out in their warranty.
However, I did talk to the people at Kingston, who tested the motherboard
model at 8GB with their memory. They are the originators of the claim that
the board could run at 8GBs using this particular memory model.

It is possible, I suppose, that the memory controller on the Phenom is the
problem, but what I need is a definate test to make absolute certain that
this is the cause. If it is, I can then attack it from there, fixing the
problem or returning/replacing the cpu. However, this would be too
expensive
unless there was solid, verifiable evidence that the memory controller is
the
problem.

I know nothing of memory controllers on cpus. I just first heard about
them
a few days ago. So, I don't know how they work or anything. However, it
just
seems to me that if the BIOS accepts and reports the 8GBs, then doesn't
than
mean the motherboard and cpu have also accepted it? I am willing to
contact
the cpu manufacturer. I just need to learn more about the memory
controller
on the cpu, and whether there is a clear test to determine whether it is
the
culprit.

"Charlie Tame" wrote:

Colin Barnhorst wrote:
Curious, that is specific to 32bit Vista. The OP's issue is an
unstable
computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may not
be able to handle the configuration. I have read that there are issues
like that with some phenoms. However, the OP insists it must be Vista
because the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us





Yes, Asus have got it wrong before. My old machine says it can take up
to 2MB but install that and it reports something off the wall like 50
bytes or something and crashes. New BIOS does not fix, it is a hardware
addressing defect apparently.

If you read the entire article you will find lots of discussion about
requiring 64 bit Vista in some situations and at lease one reference to 64
bit Vista with 8GB installed.
"Colin Barnhorst" wrote in message
...
Curious, that is specific to 32bit Vista. The OP's issue is an unstable
computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may not be
able to handle the configuration. I have read that there are issues like
that with some phenoms. However, the OP insists it must be Vista because
the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us

If you a retail copy of the OS then you are entitled to free installation
support, period. The only other possibility is an OEM pack, and you would
be out of luck there. The purchaser of an OEM pack is responsible for
supporting the customer for whom he installed the software.

I would simplify. I would get XP off for the time being. I would focus on
Vista because it is the one that can access 8GB of ram. XP can't anyway.
You can always reinstall XP from an image to a second partition or drive and
do the usual repair of the dual-boot startup from the MS KB article on
installing XP second, or by using VistaBoot Pro. It is really hard to
troubleshoot a computer as it is and simplification is priceless. In fact,
it is an art form.

"MrKit" wrote in message
...
Right. I agree with you there. As I said, I did contact the DIMM
manufacturer
and get the exact timing and voltage to use with this memory as well as
this
motherboard. They also said that each DIMM is embedded with the SPD which
tells the BIOS which timing and voltage to use. In this case, they are cas
latency 5, ras cas 5, row precharged delay 5 and voltage 1.8. I even
hard-set
these in the BIOS to override the SPD with the same results. To me, this
eliminates the SPD as the problem. The BIOS had read it properly and set
it
for the DIMMS in the first place.

And, I contacted AMD, the manufacturer or the Phenom 9500 in order to
investigate any issues there - with the memory controller or something
else.

ASUS, Kingston and AMD all say that the BIOS is properly reading the SPD
in
the DIMMS and utilizing all the RAM. They each say this is an OS issue.
Microsoft will not discuss the issue with me for less than $59 despite the
fact that I paid full price for both Microsoft Os' in use on the computer.
When one pays $399 for an operating system, they should get better
treatment.

This is the first incident I ever brought here. All previous issues that
MIGHT be Microsoft-related I handled alone, with friends, or with
Microsoft
in the first 90 days. For one thing, although I have been using computers
almost daily since 1978, I am not a good typist and it takes a long time
to
write these messages and remove most of the errors.

I had hoped to run into someone with my same motherboard, cpu, memory and
Windows Vista Ultimate SP1 64-bit edition, to see how they make it all
play
nicely together. That would have been great, but since it didn't happen, I
get to deal with real life on life's terms.

Frankly, all I am left to do is to examine the issue from all possible
sides, and try those solutions which are the least destructive and seem
most
likely to work.

"Colin Barnhorst" wrote:

The problem is timings while accessing the memory. The BIOS is simply
reporting the presence of devices compatible with the mobo. The BIOS is
not
predicting success of an installed OS.

For mobo manufacturers and cpu makers, it is a real estate problem. That
is
why I call it the "fourth damn dimm" problem sometimes. The electrical
distance out and back from the memory controller to the furthest dimm
determines the maximum frequency that can be supported and also sustain
stability. There is a point at which the memory controller is
overwhelmed
or cannot keep up. Slightly increasing the dram voltage (no more than
0.1v)
can help in a marginal situation as can lowering the dram speed if the
highest rated ram for the mobo is in use. I think you are already using
667
ram which is a very comfortable dram speed for current boards so I don't
see
that as an issue. But the memory controller is integrated on the Phenom
so
while the cpu is compatible with the board and 8GB of ram is compatible
with
the board, is your model Phenom AND 8GB of your ram compatible on your
mobo?
That is the question. It is never the individual component specs but the
combination that is such a headache to work out at times.

"MrKit" wrote in message
...
On the surface I would agree with you about ASUS. They can and do make
mistakes or misrepresent things in order to sell their products
sometimes,
or
to keep from fulfilling their duties as spelled out in their warranty.
However, I did talk to the people at Kingston, who tested the
motherboard
model at 8GB with their memory. They are the originators of the claim
that
the board could run at 8GBs using this particular memory model.

It is possible, I suppose, that the memory controller on the Phenom is
the
problem, but what I need is a definate test to make absolute certain
that
this is the cause. If it is, I can then attack it from there, fixing
the
problem or returning/replacing the cpu. However, this would be too
expensive
unless there was solid, verifiable evidence that the memory controller
is
the
problem.

I know nothing of memory controllers on cpus. I just first heard about
them
a few days ago. So, I don't know how they work or anything. However, it
just
seems to me that if the BIOS accepts and reports the 8GBs, then doesn't
than
mean the motherboard and cpu have also accepted it? I am willing to
contact
the cpu manufacturer. I just need to learn more about the memory
controller
on the cpu, and whether there is a clear test to determine whether it
is
the
culprit.

"Charlie Tame" wrote:

Colin Barnhorst wrote:
Curious, that is specific to 32bit Vista. The OP's issue is an
unstable
computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may
not
be able to handle the configuration. I have read that there are
issues
like that with some phenoms. However, the OP insists it must be
Vista
because the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us





Yes, Asus have got it wrong before. My old machine says it can take up
to 2MB but install that and it reports something off the wall like 50
bytes or something and crashes. New BIOS does not fix, it is a
hardware
addressing defect apparently.

Yes, it points out that x64 is required to access all of 4GB of ram. That
is the workaround.

But the issue involving not seeing all of 4GB does not bear on x64. It is a
problem peculiar to the 32bit clients. The KB is complex. It first
addresses a Windows 32bit issue only, the effect of the BIOS reserving
memory for devices. You'll notice through the whole first half that all the
references are specifically 32bit Windows.

The subject changes midway, beginning with the Workaround. At that point it
discusses the requirements for x64 to see 4GB with 4GB installed. Most
64bit computers meet these requirements now. I did have one AMD64 x2 box
that needed the BIOS memory remapping option turned off or I would only see
3.5GB. But the Workaround does not help the 32bit Windows user. Notice
that the last bullet in the Workaround states a 64bit edition of Windows is
required.

Then comes the whole business about PAE which is pretty irrelevant to how
much memory can be accessed by the OS. It is relevant to program space, but
by this time the KB has suffered severe mission creep and the PAE part
should have been merely a link to the same info elsewhere.

PAE gets some techies excited but PAE can never be fully implemented in a
Windows client now. Its use for extending memory addressing also requires a
carefully controlled computing environement. All hardware and software must
be PAE aware. The use of non-PAE compliant software would bring a system to
its knees. That's why you only see support written into server editions of
Windows. When was the last time you checked Ad-Aware to see if it was PAE
compatible?

"Curious" wrote in message
...
If you read the entire article you will find lots of discussion about
requiring 64 bit Vista in some situations and at lease one reference to 64
bit Vista with 8GB installed.
"Colin Barnhorst" wrote in message
...
Curious, that is specific to 32bit Vista. The OP's issue is an unstable
computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may not
be able to handle the configuration. I have read that there are issues
like that with some phenoms. However, the OP insists it must be Vista
because the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us

Everything you say is true. I posted the link since their was some
information in it that applied to the subject of this thread and which might
be of benefit to MrKit. I certainly was not trying to imply that the link's
content would solve the Op's problem.
"Colin Barnhorst" wrote in message
...
Yes, it points out that x64 is required to access all of 4GB of ram. That
is the workaround.

But the issue involving not seeing all of 4GB does not bear on x64. It is
a problem peculiar to the 32bit clients. The KB is complex. It first
addresses a Windows 32bit issue only, the effect of the BIOS reserving
memory for devices. You'll notice through the whole first half that all
the references are specifically 32bit Windows.

The subject changes midway, beginning with the Workaround. At that point
it discusses the requirements for x64 to see 4GB with 4GB installed. Most
64bit computers meet these requirements now. I did have one AMD64 x2 box
that needed the BIOS memory remapping option turned off or I would only
see 3.5GB. But the Workaround does not help the 32bit Windows user.
Notice that the last bullet in the Workaround states a 64bit edition of
Windows is required.

Then comes the whole business about PAE which is pretty irrelevant to how
much memory can be accessed by the OS. It is relevant to program space,
but by this time the KB has suffered severe mission creep and the PAE part
should have been merely a link to the same info elsewhere.

PAE gets some techies excited but PAE can never be fully implemented in a
Windows client now. Its use for extending memory addressing also requires
a carefully controlled computing environement. All hardware and software
must be PAE aware. The use of non-PAE compliant software would bring a
system to its knees. That's why you only see support written into server
editions of Windows. When was the last time you checked Ad-Aware to see
if it was PAE compatible?

"Curious" wrote in message
...
If you read the entire article you will find lots of discussion about
requiring 64 bit Vista in some situations and at lease one reference to
64 bit Vista with 8GB installed.
"Colin Barnhorst" wrote in message
...
Curious, that is specific to 32bit Vista. The OP's issue is an unstable
computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may not
be able to handle the configuration. I have read that there are issues
like that with some phenoms. However, the OP insists it must be Vista
because the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us

It's a good link. I usually snip all but the 32bit stuff. Frankly I'll be
glad to see the issue die of old age someday when all new computers come
with 64bit Windows. Most users just don't have the computer background to
sort it out.

"Curious" wrote in message
...
Everything you say is true. I posted the link since their was some
information in it that applied to the subject of this thread and which
might be of benefit to MrKit. I certainly was not trying to imply that
the link's content would solve the Op's problem.
"Colin Barnhorst" wrote in message
...
Yes, it points out that x64 is required to access all of 4GB of ram.
That is the workaround.

But the issue involving not seeing all of 4GB does not bear on x64. It
is a problem peculiar to the 32bit clients. The KB is complex. It
first addresses a Windows 32bit issue only, the effect of the BIOS
reserving memory for devices. You'll notice through the whole first half
that all the references are specifically 32bit Windows.

The subject changes midway, beginning with the Workaround. At that point
it discusses the requirements for x64 to see 4GB with 4GB installed.
Most 64bit computers meet these requirements now. I did have one AMD64
x2 box that needed the BIOS memory remapping option turned off or I would
only see 3.5GB. But the Workaround does not help the 32bit Windows user.
Notice that the last bullet in the Workaround states a 64bit edition of
Windows is required.

Then comes the whole business about PAE which is pretty irrelevant to how
much memory can be accessed by the OS. It is relevant to program space,
but by this time the KB has suffered severe mission creep and the PAE
part should have been merely a link to the same info elsewhere.

PAE gets some techies excited but PAE can never be fully implemented in a
Windows client now. Its use for extending memory addressing also
requires a carefully controlled computing environement. All hardware and
software must be PAE aware. The use of non-PAE compliant software would
bring a system to its knees. That's why you only see support written
into server editions of Windows. When was the last time you checked
Ad-Aware to see if it was PAE compatible?

"Curious" wrote in message
...
If you read the entire article you will find lots of discussion about
requiring 64 bit Vista in some situations and at lease one reference to
64 bit Vista with 8GB installed.
"Colin Barnhorst" wrote in message
...
Curious, that is specific to 32bit Vista. The OP's issue is an
unstable computer running 64bit Windows on 8GB.

The integrated memory controller in the phenom quad he is using may not
be able to handle the configuration. I have read that there are issues
like that with some phenoms. However, the OP insists it must be Vista
because the mobo maker couldn't have gotten it wrong.

Oh well.

"Curious" wrote in message
...
Maybe the information in the following link will help:

http://support.microsoft.com/kb/929605/en-us

Colin Barnhorst wrote:

Then comes the whole business about PAE which is pretty irrelevant to
how much memory can be accessed by the OS. It is relevant to program
space, but by this time the KB has suffered severe mission creep and the
PAE part should have been merely a link to the same info elsewhere.

PAE gets some techies excited but PAE can never be fully implemented in
a Windows client now. Its use for extending memory addressing also
requires a carefully controlled computing environement. All hardware
and software must be PAE aware. The use of non-PAE compliant software
would bring a system to its knees. That's why you only see support
written into server editions of Windows. When was the last time you
checked Ad-Aware to see if it was PAE compatible?

Actually, it is not about program space (virtual address space), because
that does not change in pae mode. It only affects (bad) drivers that
takes for granted that nothing will ever exists above 4G. Normal
applications never sees anything more than 32bit addresses, and there
cannot be affected.

I installed Vista initially about 18 months ago. The 90-day tech support has
long since expired. I mentioned that in at least a couple of my earlier
messages.

Both Windows XP Pro 64-bit AND Windows Vista Ultimate 64-bit claim to be
able to use 8GB of memory.

I repeat something else I said more than once before - I will NOT be
removing my XP installation. It works, and I use it for things I cannot do in
Vista. It is much faster than Vista. There are many things that work in XP
and not Vista. Besides all that, it takes too long and is a big hassle to
re-install XP and then call in and beg for an activation when it won't
activate over the internet because so many tech support people in the
beginning told me to uninstall it. I used up the standard number of allowed
XP installs just doing what Microsoft said way back then.

So, since I am unwilling to remove XP, I am only looking for solutions that
do not require me to remove Windows XP, especially since no one has presented
one shred of evidence that XP is causing any kind of problem whatsoever. I
won't destroy something that's working in order to fight with something that
isn't working.

This is computer science, not an art class. Computers are black-and-white.
Art is relative.

"Colin Barnhorst" wrote:

If you a retail copy of the OS then you are entitled to free installation
support, period. The only other possibility is an OEM pack, and you would
be out of luck there. The purchaser of an OEM pack is responsible for
supporting the customer for whom he installed the software.

I would simplify. I would get XP off for the time being. I would focus on
Vista because it is the one that can access 8GB of ram. XP can't anyway.
You can always reinstall XP from an image to a second partition or drive and
do the usual repair of the dual-boot startup from the MS KB article on
installing XP second, or by using VistaBoot Pro. It is really hard to
troubleshoot a computer as it is and simplification is priceless. In fact,
it is an art form.

Colin is the guru of 8GB memory support and addressing in this and several
other newsgroups.
He assumed that when you mentioned XP that you did not mean the 64bit
version of XP.
All versions of Vista can run 64 bit vista and can support more the 4GB of
memory. The 32bit versions can not support more the 4GB for any one process
however.
One difference between memory support in memory support in XP and in Vista
is that when Vista boots it will test the speed of all the memory to insure
that it is all running at the same speed. If it is not then it only
recognizes the slower memory.
"MrKit" wrote in message
...
I installed Vista initially about 18 months ago. The 90-day tech support
has
long since expired. I mentioned that in at least a couple of my earlier
messages.

Both Windows XP Pro 64-bit AND Windows Vista Ultimate 64-bit claim to be
able to use 8GB of memory.

I repeat something else I said more than once before - I will NOT be
removing my XP installation. It works, and I use it for things I cannot do
in
Vista. It is much faster than Vista. There are many things that work in XP
and not Vista. Besides all that, it takes too long and is a big hassle to
re-install XP and then call in and beg for an activation when it won't
activate over the internet because so many tech support people in the
beginning told me to uninstall it. I used up the standard number of
allowed
XP installs just doing what Microsoft said way back then.

So, since I am unwilling to remove XP, I am only looking for solutions
that
do not require me to remove Windows XP, especially since no one has
presented
one shred of evidence that XP is causing any kind of problem whatsoever. I
won't destroy something that's working in order to fight with something
that
isn't working.

This is computer science, not an art class. Computers are black-and-white.
Art is relative.

"Colin Barnhorst" wrote:

If you a retail copy of the OS then you are entitled to free installation
support, period. The only other possibility is an OEM pack, and you
would
be out of luck there. The purchaser of an OEM pack is responsible for
supporting the customer for whom he installed the software.

I would simplify. I would get XP off for the time being. I would focus
on
Vista because it is the one that can access 8GB of ram. XP can't anyway.
You can always reinstall XP from an image to a second partition or drive
and
do the usual repair of the dual-boot startup from the MS KB article on
installing XP second, or by using VistaBoot Pro. It is really hard to
troubleshoot a computer as it is and simplification is priceless. In
fact,
it is an art form.

Curious wrote:

One difference between memory support in memory support in XP and in
Vista is that when Vista boots it will test the speed of all the memory
to insure that it is all running at the same speed. If it is not then it
only recognizes the slower memory.

That is a job for the BIOS. It programs the memory controller, not the OS.