HP EliteDesk 705 G1's BIOS does not see all memory that Windows or CPU-Z sees

[DGroves]

did you take a 8gb ram stick and  use it on socket # 0, power on check the shown ram, powering off, moving the ram stick to the next socket #2, power on, check the shown ram, power off moving the ram stick to the next socket #3 power on check ram, power off moving the ram stick to the next socket #4

if all 4 sockets check good showing 8gb each time, power off place second 8gb in socket #1, and repeat socket tests using other 8gb module and finishing at #3 did this test show 16 gb at each socket?

if so, power off leave the two 8gb modules in the sockets used for the systems first memory bank  and repeat tests using a 4gb module in the remaining 2 empty sockets does it show 20gb on each socket?

if so, power off place 4 gb module in first empty socket of memory bank 2, power on, confirm 20gb, power off swap 4 gb module with other one, power on check ram, if 20gb, power off, and insert last 4 gb module in last open slot power on, if 24 gb ram all is good if not then last socket is bad or mother board is bad

the above sequence of testing rules out the possibility of bad ram, as each module is tested using the known good test results of  the previous  tests, when check in the next ram socket

and each good result showing a ram increase valadates the ram and now the socket as good.

the point where it fails to show a ram increase is where the socket or if you get to the last socket/ram module the motherboard itself is showing a defect

you must test the ram/sockets as listed in the exact order listed above, if you skip any steps then you will not have ruled out the possibility of bad ram/socket/motherboard

[old_geekster]

I had a DFI mothetboard that had 4 memory sockets, but I could only populate two of them at a time.  If I populated all of them the computer wouldn't boot.  I had ever DFI expert in the world trying to determine the problem.  Nothing ever worked.

I agree that the tests you have run eliminate a problem with the sockets or memory sticks.  The only caveat is, sometimes different brands of memory, even though the correct specs, don't play nice together.  Samsung memory is one of those culprits.

[kennethi]

@DGroves, thanks for the suggestion, I might have to try that ...

However, I can't imagine bad RAM, bad sockets, or a bad motherboard returning perfect results in a full run of all memory/stress tests under memtest86 from RAM addresses 0 to 0x63F000000. Or, the ability of Windows and other Windows-based utilities to recognize the 24GB.

[kennethi]

Thank you @old_geekster, in my case the PC boots just fine and pretty much everything except the UEFI/BIOS POST seems to recognize the 24GB of RAM.

& @Paul_Tikkanen, a correction to what I wrote before (original post corrected too) ... Right now, with all four RAM slots populated, whenever I start the computer, the BIOS POST shows this for about 2 seconds ...

Copyright (c) 2014 Hewlett-Packard Development Company L.P.
Version 2.16.1240. Copyright (C) 2013 American Megatrends, Inc.
24576 MB

... then the screen immediately changes to:

Copyright (c) 2014 Hewlett-Packard Development Company L.P.
Version 2.16.1240. Copyright (C) 2013 American Megatrends, Inc.
12288 MB

No error messages are given by the BIOS POST before it then boots into Windows normally.

[kennethi]

HP posted a BIOS update - version 2.30 Rev A (L06) dated March 22, 2018 - for the 705 G1's: https://support.hp.com/us-en/drivers/selfservice/swdetails/hp-elitedesk-705-g1-small-form-factor-pc/... ... It must have been posted in the last few days, because 3 days ago, it was still showing version 2.28 Rev A (L06).

I applied version 2.30, but sadly it did not fix the RAM issue.

I opened case [edited] with HP yesterday, but it was closed today already with a email from the tech writing that I needed to resolve the issue by purchasing HP branded memory! I thought that a manufacturer may not require the use of any specific brand unless the manufacturer proves the item free of charge?

Side note: AGESA is still showing version 1.1.0.5. AGESA version 1.1.0.7 supposedly available since 2015 was supposed to patch a critical CPU vulnerability: https://h30434.www3.hp.com/t5/Business-PCs-Workstations-and-Point-of-Sale-Systems/AGESA-BIOS-upgrade...