-
×InformationNeed Windows 11 help?Check documents on compatibility, FAQs, upgrade information and available fixes.
Windows 11 Support Center. -
-
×InformationNeed Windows 11 help?Check documents on compatibility, FAQs, upgrade information and available fixes.
Windows 11 Support Center. -
- HP Community
- Desktops
- Desktop Operating Systems and Recovery
- Z420: Memory read error - DB parity error

Create an account on the HP Community to personalize your profile and ask a question
08-08-2017 03:08 AM - edited 08-08-2017 03:09 AM
Ooookay now I REALLY understood what you are referring to haha :)) It's a very subtle misunderstanding:
I said that columns E, F and G support my observation.
My observation is: "old modules in a channel with both slots loaded cause the error"
E = old module in a channel with both slots loaded => error
F = old module in a channel with both slots loaded => error
G = NO old modules in a channel with both slots loaded => NO error
Column G is a test which doesn't include any old modules, but since no error arises this column still supports my observation of "old modules in a channel with both slots loaded cause the error".
Summary: No error in 1+2 if it is new+new (that's column G in the old screenshot as well as in the new screenshot, hope you can see it soon).
08-08-2017 09:33 AM
Now, I can see your new table.
Parity error:
B 5 or 6 OLD NEW
E 1 or 2 OLD OLD
F 1 or 2 OLD NEW
H 5 or 6 OLD OLD
I 1 or 2 OLD OLD
L 5 or 6 NEW OLD
M 3 or 4 OLD NEW
You may have one of the OLD sticks that is "bad".
You keep moving it to different slots.
Can you run MEMTEST with less than 4 sticks inserted?
The vendor's statement is "for optimal performance, use 4 sticks", but is that an absolute requirement to use 4 sticks?
If the computer will run with "less", try the "old" RAM:
* OLD1 in 1, with OLD2 in 3
* OLD1 in 1, with OLD3 in 3
* OLD1 in 1, with OLD4 in 3
* OLD2 in 1, with OLD1 in 3
* OLD2 in 1, with OLD3 in 3
* OLD2 in 1, with OLD4 in 3
* OLD3 in 1, with OLD1 in 3
* OLD3 in 1, with OLD2 in 3
* OLD3 in 1, with OLD4 in 3
* OLD4 in 1, with OLD1 in 3
* OLD4 in 1, with OLD2 in 3
* OLD4 in 1, with OLD3 in 3
As it runs, MEMTEST occupies some of the RAM in slot 1.
So, that's why I have listed OLD1 in 1 with OLD2 in 3,
and also OLD2 in 1 with OLD1 in 3 --
to be sure that "all" the RAM on each chip is tested.
08-09-2017 02:27 AM - edited 08-09-2017 02:29 AM
If I put 2 old modules into slot 1 and 3 (2 x black), then those two channels are "half-filled" because I leave the according white slots empty. But I have already tested all old modules in a half-filled setup! That's column C. It shows that the old modules work perfectly fine when in a half-filled channel ...
How is it possible that these memory modules work perfectly fine in a 1-slot-channel but not in a 2-slot-channel?
The new memory modules are exactly the same and they work in a 2-slot-channel.
08-09-2017 10:05 AM
> If I put 2 old modules into slot 1 and 3 (2 x black), then those two channels are "half-filled" because I leave the according white slots empty.
Yes, that is what I suggested, namely to leave slots 4 through 8 "empty", and run MEMTEST, if your computer will run with only these 2 slots being populated.
Does your computer boot in this configuration?
What did MEMTEST reveal, for each of the combinations that I suggested?
Without actual "experimental" evidence, we can talk "theory" forever, without approaching a solution.
> But I have already tested all old modules in a half-filled setup!
> That's column C. It shows that the old modules work perfectly fine when in a half-filled channel ...
Yes, you clearly have run that "experiment". Now, I am asking you to report the results of further "experiments".
> How is it possible that these memory modules work perfectly fine in a 1-slot-channel but not in a 2-slot-channel?
Those are two different configurations, and something "subtle" is faulty with your RAM, and the fault is exposed only in a specific configuration.
That's why I have asked you to try to run MEMTEST on exactly two sticks at a time, and to report your results.
08-10-2017 03:09 AM - edited 08-18-2017 01:22 AM
Okay finally done with all the configurations of your last post. Result: ALL of them work!
Additionally I've also tested 1-3-6 loaded with OLD modules (but this time not with all permutations of the modules): works fine too.
EDIT: I think continuing this thread doesn't make much sense since nobody with fresh ideas wants to read 2 pages of posts. Maybe I will reopen a thread in the hardware section and describe the current status right at the start.
- « Previous
-
- 1
- 2
- Next »