On 5/1/2018 12:59 PM, Paul wrote:
Bill Anderson wrote:
It began happening a few weeks ago -- I'd try to boot but before post
I'd hear four short beeps followed by the post chirp (different sound)
and then nothing.Â* But it was an intermittent problem and I let it go
when rebooting seemed to solve things.

Then yesterday the problem stuck around through several attempted
reboots and I figured I'd better do something.Â* So I looked up the
error beep table on the Asus website and found nothing about four
short beeps. Thanks, Asus.Â* Then I branched out and learned Asus uses
American Megatrends' AMI BIOS,Â* and that four beeps mean System Timer
Failure, which further means something's wrong with memory.

That was when I actually looked at BIOS to see how much memory I had
and I learned my four 8-gig memory sticks were producing not 32 gigs
of memory, but just 8 and change.Â* In other words, only one stick was
working.

So just now I removed all the memory and began replacing sticks one at
a time. At first I couldn't get past the four beeps, but I kept
replacing sticks in the D1 slot until one worked.Â* (The manual says to
put a single stick in D1.) Then I put a stick in B1 as shown in the
manual and went right back to the four error beeps.

But I persevered, removing and re-inserting, and eventually I had two
working, then four, and now I'm booting nicely and BIOS shows total
memory at 32 gigs and change.

Funny things was -- the AIDA 64 Extreme system monitor software always
showed four slots filled with 8-gig sticks.Â* It saw them when BIOS
didn't. Weird.

But things are fine now.Â* I think. Maybe I oughta run Memtest just to
be sure...


Memory is detected two ways at BIOS level.

1) Read config info from SPD PROM on each DIMM.

2) Once the BIOS knows a DIMM is present, it uses "peek & poke"
Â*Â* testing to prove "a RAM is a RAM" and that it can actually
Â*Â* store stuff.

One of the "proof cases" for this, was the day when
some brand of DIMMs, had the wrong SPD chip soldered
to them. The DIMM may have declared it had 256MB on it,
when the physical chips were 128MB. The BIOS (correctly)
did "peek & poke" and measured 128MB, and the system
started just fine and ran with the reduced amount
of memory. Because in fact, that's all the physical memory
that was present on the stick.

This issue seemed to first show up on triple channel memory.
Maybe a 12GB system would be detected as an 8GB system. At
first, people might have tried blaming a "socket contact"
issue for the problem.

Then later, people started randomly adjusting the IMC or
VNorthbridge. And then there were claims that this
was "fixing it".

I don't know if I've ever seen a company web site
(Asus or Intel) making claims as to why this happens.

When the OS is running, hardware identification software
continues to have access to the SPD and can then
claim that 32GB are "installed", even if the BIOS
has tested and chosen to only use a subset, because
"peek & poke" is failing.

"Peek & poke" is a quick check of memory presence and
does not represent a full memory test. That may come
later in the POST, if the user has enabled it.

*******

And yes, doing some memtest is a good idea.
Between that, and something like Prime95, you'll get
a better idea whether it's really working properly or not.

Â*Â* Paul

I think re-seating the memory sticks was what I needed to do. BIOS is
showing 32 gigs of RAM now and so is Win10. I think you must be right
about how software might see the RAM sticks when BIOS couldn't. It's
always something. Thanks, Paul.

--
Bill Anderson

I am the Mighty Favog