Discussion:
Dell Precision M4300 laptop, 2 identical sticks of 2GB RAM, each passes individually, together they fail
Add Reply
Java Jive
2018-07-27 14:25:27 UTC
Reply
Permalink
Raw Message
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.

A Google search didn't find much that was useful. Any suggestions as to
what may be wrong?

TIA
Mark Lloyd
2018-07-27 15:09:21 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.
A Google search didn't find much that was useful.  Any suggestions as to
what may be wrong?
TIA
It looks like both sticks are OK, but that second memory slot is defective.
--
Mark Lloyd
http://notstupid.us/

"Micro$oft, the company that makes spreading malware easy."
Ken
2018-07-27 16:20:24 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.
A Google search didn't find much that was useful.  Any suggestions as to
what may be wrong?
TIA
Like Mark said, you have something wrong with the slot. Look closely at
the contacts for bent or damaged ones. If no damage is seen, clean the
contacts with alcohol and see if that doesn't help.
Paul
2018-07-27 16:45:50 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.
A Google search didn't find much that was useful. Any suggestions as to
what may be wrong?
TIA
There was a problem with the bottom slot here as well.

https://www.dell.com/community/Laptops-General/M4300-won-t-start-with-RAM-installed/td-p/4150560

*******

See if the BIOS has any setting for clearing
DMI/ESCD. The BIOS "remembers" the DIMM inventory
and other hardware items. Any time the hardware
installation changes, the DMI/ESCD is rewritten.
There used to be some kind of problem with that
years ago. It's not a common problem today, as it
doesn't come up in conversations any more.

Tools like DMIExplorer could be used to dump or
review the information.

From a technical perspective, it's not critical
to system operation, but it's part of the POST
sequence.

The DMI/ESCD is normally a "one-shot" thing. You
set it, and the next time the machine starts,
the POST clears the request. Normally the process is
fully automated, but sometimes it gets into an update
loop (updates each and every time the machine
starts). Then, perhaps, the one-shot BIOS request
feature, might knock it out of that loop.

In some cases, the problem isn't exactly what you'd think.
There can be a bug in the preparation of the memory map.
On one chipset, this involved some register for USB
ports, versus the Top_Of_RAM definition. The screen
of the computer would report "USB Overcurrent".
Reducing the quantity of RAM would make the problem
go away. If testing such a thing in your lab,
you'd try a mismatched pair (a 1GB and a 2GB stick)
and see what happens. If memtest passes, the system
behaves normally, it would then be telling you
there might not be a problem with the slot after all.

Dual channel problems show up with older AMD processor
designs. Two DIMMs had to be "put across from one another",
and the DIMM busses worked like tandem locomotives.
One slot is considered the master on the AMD design,
the other is a slave, and two DIMMs work like a 128
bit or a 144 bit wide DIMM in a sense. Filling the
wrong slot first, doesn't work and the system
doesn't start. Mis-matched DIMMs might start up
in single channel mode. These issues were caused
by minimalist memory controller designs on the
first processors of that generation, and gradually
over time, AMD became as good at dual channel design,
as Intel Flex Memory. Today, nobody thinks twice before
filling an AMD memory slot. (My AMD laptop has mis-matched
DIMMs.)

https://www.dell.com/support/home/ca/en/cabsdt1/Drivers/DriversDetails?driverid=R207161

"Dell Precision M4300 System BIOS, A13

4. Fixed Wi-Fi sniffer function with 4GB memory and Dell wireless.
"

That's the first release I could find, which mentions
amount of memory, and a failure condition.

*******

https://arstechnica.com/civis/viewtopic.php?f=9&t=106411

"The Dell Upgrades website claims the Precision M4300 will
support up to 8 GB of ram but my machine refuses to boot
with 2 4GB sticks. It just sits there for hours at the
Dell boot screen with less than half an inch to go on
the progress bar.

If I replace one of the 4GB sticks with one of the original
1GB stick, it will boot right up and show all 5GB in the
bios and windows.
"

"BIOS revision is A10 and currently Dell shows an update to A12"

"Edit: Woohoo!! continuum gets a cookie!"

Too bad the images attached, aren't available for review.

This could be solved by another version of BIOS,
at a guess. I don't normally recommend BIOS flashing,
as there's a risk you could always brick the thing.
Brickage wouldn't happen, if the BIOS designers followed
the rules on "boot block". But a number of commercial BIOS
zap boot block and main BIOS block, and if the refill of
the main BIOS block fails, with the erased boot block,
it can no longer bootstrap itself.

If it does brick, then the issue is whether the BIOS chip
is socketed or there is a programming header next to it.
PLCC or DIP flash chips, some are socketed. Those have
no programming header next to the socket. Newer designs
use 8 pin (serial) Flash chips, for which a 7 pin programming
header might be adjacent. A programmer dongle tied to USB
bus on the Technician computer, allows flashing up the
broken chip and fixing a brickage incident.

Paul
mike
2018-07-28 04:36:30 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.
A Google search didn't find much that was useful. Any suggestions as to
what may be wrong?
TIA
I assume that your laptop claims that it will work with
4GB ram?

I once had an issue with a bad bit in the high memory area of
a ram stick. Turned out that the bad section was used for
display frame buffer and memory test didn't test it.
Java Jive
2018-07-28 15:36:37 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2 identical
sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option, each
stick passes individually in Slot 1 (under the keyboard), but together
one in each slot they fail, and the PC won't boot with a stick only in
Slot 2, behind a cover on the base of the laptop.
A Google search didn't find much that was useful.  Any suggestions as to
what may be wrong?
TIA
Thanks for all the replies, this is a 'put on hold' post because,
tentatively, touching every bit of wood within reach including my head,
I think I've solved it. Will report fully on that when I'm sure either way.

Meanwhile, I can tell what it wasn't. I'd already concluded, in
agreement with the suggestions made here, that the second socket might
be faulty, so, although I'd cleaned it out already with the nozzle of a
vacuum cleaner, I thought I'd take a x10 lens to it, and, lo and behold,
spider silk, completely with remains of microscopically small spider.
You, like me, might be surprised that the silk and the corpse had
survived the vacuum cleaner (it's a powerful one, too, near deafeningly
so), but man is that stuff sticky?! I applied the nozzle again, but it
was still there afterwards. Next I got an old, but clean, toothbrush,
dipped it an IPA* 1 in 4 solution that I used to use to clean LPs, and
cleaned up the socket and the RAM with that. That got rid of most, but
not quite all, of it. So there followed several cycles of IPA
toothbrush, vacuum cleaner, clean dry toothbruch, vacuum cleaner until
finally after about an hour I was satisfied that there were no threads
of silk lying across the terminals. I replaced the RAM, replaced the
battery, connected the power, and ... nothing, it still wouldn't boot!

But at least now I know how to clean spider silk from electronic equipment.

* BTW, that's IsoPropyl Alcohol, not Indian Pale Ale!
Java Jive
2018-07-28 17:39:49 UTC
Reply
Permalink
Raw Message
Post by Java Jive
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2
identical sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2 64
RAM. Running the RAM test from the Ubuntu 16.04 64-bit grub boot
option, each stick passes individually in Slot 1 (under the keyboard),
but together one in each slot they fail, and the PC won't boot with a
stick only in Slot 2, behind a cover on the base of the laptop.
Thanks for all the replies, this is a 'put on hold' post because,
tentatively, touching every bit of wood within reach including my head,
I think I've solved it.  Will report fully on that when I'm sure either
way.
B*gger, and I think you can guess what that means! B*stard got to 99%
of second pass before it froze.

A while ago, I installed a 4G WLAN card, and, although I hadn't noticed
RAM-type problems beginning around that time, since the PC had passed a
RAM test with both sockets populated when I first acquired it, that was
the only thing that I could think of that had changed since. So I tried
removing the card, and lo, it booted with its full compliment of RAM.
Encouraged, I let it run for a while, and wrote the above after it had
completed one successful pass, but, as already described it froze right
at the end of the second.

I'm now thinking temperature is part of the problem. The PC seems much
more likely to fail when it's hot. The RAM test flogs it continuously,
but when started from cold succeeds initially, only to fail once the
thing gets really hot. Then if you immediately restart, it fails more
or less immediately, if it will even boot at all.

Presumably there is something like a poorly soldered joint or hairline
crack in a track on the mainboard, and, as it heats up, eventually a gap
opens and it fails. Short of replacing the mainboard, a serious fag, as
well as being somewhat intricate and error prone, I think it's just
going to have to run with one stick for the foreseeable future.

Bloody nuisance, as that was my development PC, the one that has been
doing all the hard work, compiling, etc. It's like lame duck with only
half the RAM.
Paul
2018-07-28 20:36:54 UTC
Reply
Permalink
Raw Message
Post by Java Jive
Post by Java Jive
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2
identical sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM PC2
64 RAM. Running the RAM test from the Ubuntu 16.04 64-bit grub boot
option, each stick passes individually in Slot 1 (under the
keyboard), but together one in each slot they fail, and the PC won't
boot with a stick only in Slot 2, behind a cover on the base of the
laptop.
Thanks for all the replies, this is a 'put on hold' post because,
tentatively, touching every bit of wood within reach including my
head, I think I've solved it. Will report fully on that when I'm sure
either way.
B*gger, and I think you can guess what that means! B*stard got to 99%
of second pass before it froze.
A while ago, I installed a 4G WLAN card, and, although I hadn't noticed
RAM-type problems beginning around that time, since the PC had passed a
RAM test with both sockets populated when I first acquired it, that was
the only thing that I could think of that had changed since. So I tried
removing the card, and lo, it booted with its full compliment of RAM.
Encouraged, I let it run for a while, and wrote the above after it had
completed one successful pass, but, as already described it froze right
at the end of the second.
I'm now thinking temperature is part of the problem. The PC seems much
more likely to fail when it's hot. The RAM test flogs it continuously,
but when started from cold succeeds initially, only to fail once the
thing gets really hot. Then if you immediately restart, it fails more
or less immediately, if it will even boot at all.
Presumably there is something like a poorly soldered joint or hairline
crack in a track on the mainboard, and, as it heats up, eventually a gap
opens and it fails. Short of replacing the mainboard, a serious fag, as
well as being somewhat intricate and error prone, I think it's just
going to have to run with one stick for the foreseeable future.
Bloody nuisance, as that was my development PC, the one that has been
doing all the hard work, compiling, etc. It's like lame duck with only
half the RAM.
http://www.crucial.com/usa/en/compatible-upgrade-for/Dell/precision-m4300

Maximum Memory:4GB
Standard Memory:1GB removable
Slots:2 (2 banks of 1)

Chipset: Intel PM965

Unfortunately, of the 8 items listed in the extended Crucial list,
the 4GB DDR2 SODIMM they sell, is not in the list. So you cannot
fix it that way. I was thinking, maybe you could replace the
single 2GB SODIMM with a 4GB one, but the chipset doesn't
support it.

I have run into one case, where the VIA chipset company,
didn't actually know the capabilities of their own hardware,
and a chipset with a 1GB slot limit, actually could use 2GB
DIMMs. But that's only happened the one time, and the
BIOS wasn't "tuned" for 2GB DIMMs so the experiment
was overall a failure. The setup had an excessive background
error rate which would have required 256 experiments to
correct :-(

*******

This is the PM965 datasheet.

https://www.intel.com/Assets/PDF/datasheet/316273.pdf

1.1.2 System Memory Support

1-Gbit memory technologies supported [=128MB].

16 chips of 128MB each is 2GB, and that's why it won't take
more than 2GB per slot. To fit that many chips on an
SODIMM, takes FBGAs or two dies inside a single plastic
package.

*******

It doesn't look like there is an easy workaround.

As for bad solder joints, it could be a bad
joint on the DIMM slot, but it could also be
a cracked solder ball underneath the PM965.
At the factory, you can spot such things with
a 2.5D Xray machine, which shoots pictures
from either side of vertical, to review the
balls in perspective view. Some companies
do Xray inspection (two snaps per chip),
for every large BGA, as part of closed
loop quality control. This makes sure they
aren't doing anything stupid on a soldering
profile. A properly tuned soldering setup,
will have around 1 bad ball per 100,000 balls.
With the XRay machine available on the line,
presumably the amount of bad product that
escapes the line, is even lower than that.

The PM965 is conservatively designed. It's a
2500 ball pattern, with a sparse fill and
only 1300 balls populated. This leaves
plenty of room for stress relief as the
chip heats up. From a bar bet perspective,
the pattern I see on page 137 is not the
worst I've ever seen. When you have BGAs
with full matrix and high ball counts, the
balls in the corner sometimes crack. For example,
if all 2500 locations had balls, then the corner
ones would be prime candidates for failure.
I've had at least one corner ball crack on
me in the lab, out of all the proto boards that
crossed my bench. Shit happens, as they say.

Paul
Daniel James
2018-07-29 11:46:40 UTC
Reply
Permalink
Raw Message
Post by Java Jive
As per title, I have an old Dell Precision M4300 laptop with 2
identical sticks of Crucial CT25664AC800 16 FHZ 2GB DDR2 SODIMM
PC2 64 RAM.
Running the RAM test from the Ubuntu 16.04 64-bit grub boot option,
each stick passes individually in Slot 1 (under the keyboard), but
together one in each slot they fail ...
I had a similar embuggerance with a Toshiba, some time ago. It has two
SODIMM slots, and came with 1x1GB and 1x2GB. I upgraded it immediately
by swapping the 1GB SODIMM for a 4GB, giving 6GB in all, and it worked
just fine.

Then, after a couple of years, I had a need to run three Windows VMs at
once (under Linux, of course) for a client project ... which didn't
quite work in 6GB. So, I swapped the original 2GB SODIMM for another
4GB. The machine reported (IIRC) 1GB in total installed!

After a lot of hair-pulling and asking questions online that elicited no
useful answers I found a comment in a forum somewhere that prompted me
wonder whether a BIOS update might be available and might fix the
problem. Sure enough, an update existed -- the release notes said
nothing about improved RAM handling -- so I installed it, and on
rebooting all 8GB were recognized.

So: try a BIOS update.
--
Cheers,
Daniel.
Loading...