I have following 16 memory modules Samsung PC3L-12800L-11-12-C0 (M386B4G70DM0-YK04Q) which are each 32GB. Total memory is 512GB.
The problems I describe are at boot only.
Then I have 3 computers. Computer1: HP z420v2 with one Xeon E5-2660v2. I can boot with 8x 32GB and there are no issues. Computer2: HP z420v2 with one Xeon E5-2643v2. I can boot with 8x 32GB and there are no issues. Computer3; HP z820 with 2x Xeon E5-2667v2. I can boot with 14x 32GB and there are no issues, however if I boot with 16x 32GB then there are issues.
Describing the issue is with Computer3:
I tried different memory configurations and found out that it is best to start with all black slots. Using any dimm out of 8 dimms in a white slot typically gives problems. So I start populating the black dimms. (8x 32GB) Then I continue populate 2 white slots for CPU 0 and the computer boots up normally. (10x 32GB) Then I continue populate 2 white slots for CPU 1 and the computer boots up normally. (12x 32GB) Then I continue populate 2 last white slots for CPU 0 and the computer boots up normally. (14x 32GB)
When I want to populate that last 2 white slots for CPU 1 (dimm 4 and dimm 5) then at boot from BIOS I get 942-training memory error with code 1701.
To check if the memory module is broken, I switch with other memory modules, no matter how many times I switch, the problem remains in slot CPU 1 dimm 4 which makes it look like a slot problem.
But if I populated the white slots in a slightly different order, then the problem slot changed the location (which looks like it is not a slot problem). For example, the "problem slot" might have moved from slot CPU 1 dimm 4 to slot CPU 0 dimm 4. And when populating the slots, I always put 2 dims at the time and the 2 dimms I put on each side of same CPU for perfect symmetry/mirroring. For example dimm 4 and 5 are both white and closest to the CPU, these I might populate last.
Computer1 and Computer2 proved that the memory modules are intact. Switching memories in Computer3 proved the modules are intact. Changing population strategy showed that it is not a slot problem.
I tried other memories, they are 8GB modules. Using 16x 8GB modules in Computer3 there are no issues. But with 16x 32GB modules in Computer3 I get 942-training memory error with code 1701 and instead of 512GB RAM I end up with only 448GB RAM.
It seems error reporting is showing wrong location or then that dimms influence other dimms, some sort of error propagation, causing this mess. Any suggestions on what I can do?
I have not "software tested" the dimms and I do not know such software. I can install Windows if needed, right now my z820 has CentoOS 8 installed. BIOS has done its memory test over night, still only at 67% of completion, I suspect this test will give me nothing.
Maybe the memory module is wrong for z820 motherboard? But z420 takes them. (Overall, I find systems with 2 CPU's always tricky)
there is one other possibility, some server grade network cards are incompatible with systems other than what they were designed for due to how they access the ic2 bus,..... one of the symptoms of installing such a card is loss of memory slots
Be alert for scammers posting fake support phone numbers and/or email addresses on the community. If you think you have received a fake HP Support message, please report it to us by clicking on "Flag Post".