0
Okay guys, I really hope someone smart will hint me towards the right solution. I got a HP Z420 Workstation and I've upgraded the memory. I had 4x4GB before and I bought additional 4x4GB. All 8x4GB are the same: "Samsung 2Rx8 PC3-12800E-11-11-E3". So there are 8 DIMM slots on my main board. Those DIMM slots are dual-channel. DIMM slots 1+3+6+8 are colored black (channel 1), DIMM slots 2+4+5+7 are colored white (channel 2). My old 4x4GB bars were inside the black channel 1 DIMM slots. I inserted the new 4x4GB bars into the white channel 2 DIMM slots. When running memtest, I end up with the message:
929-Fatal MCA error.
HA error detected CPU 0 DIMM Slot 5 or 6
Memory read error - DB parity error
After that I tried different configurations to track down the bug and I've reached a dead end where I need input from smart people :)
Here's a picture of my test summary with description below:
Column B: old bars inside the black channel 1, new bars inside the white channel 2. Fails with the message that I have posted earlier.
Column C: testing only the old bars with memtest, it works
Column D: testing only the new bars with memtest, it works. that "4 / 78%" means that I let memtest run until pass 4 at 78% before I stopped it. I forgot to document that at column C.
important interim conclusion: the old and the new memory bars both work! So there seems to be a problem when I mix them.
Column E: mixing 3 new and 1 old (only black / channel 1): works!
Column F: mixing old and new in dual channel slot 5+6: fails! (forgot to document the slots from the error message though, that's why there's only a red field without description)
Column G: mixing old and new in dual channel slot 3+4: fail!
Column H: mixing old and new in dual channel slot 1+2: fail!
interim thoughts: mixing old and new in dual channel always fails, so could it be that I need to put the old ones and the new ones into their own dual channels? shouldn't make a difference since all 8 bars have the same specs but it looks like it does.
Column I: swapping the bars between slot 2 and 3 from column H. So now the new bars are in the same dual channel: works! yay, my theory seems right.
Column J: group the new bars in a dual channel and group the old bars in a dual channel: fails againnnnnnn omg I want to cry ;_;
I have run out of ideas, I don't see any pattern in the error messages. Does anyone have a good idea / comment? I mean look at column C and D: all RAM bars are fine! then I mix then in columns F to H and they fail, but as soon as I mix them the "right way" it works (column I). But then again it doesn't (column J). What's going on here, I'm so clueless :/
user3182532
Posted 2017-08-02T12:05:03.920
Reputation: 111
Have you considered the problem is the motherboard not the memory? – Ramhound – 2017-08-02T12:32:45.470
Whenever you populate slots 2 or 4 (belonging to the same dual channel pair) you get errors - this looks like a dead channel. Try to populate ONLY 2 and 4, if it boots you can rule this out. – Eugen Rieck – 2017-08-02T12:32:48.583
@Eugen: column F contradicts that idea, where 2 and 4 are empty but the error persists :/ and I don't think that loading only 2 and 4 is possible since those are channel 2 slots which are supposed to be only loaded if channel 1 is loaded, right? – user3182532 – 2017-08-02T12:40:55.920
@user3182532 Column F is a "no boot" condition, telling us nothing about the channel. – Eugen Rieck – 2017-08-02T12:41:50.913
what do you mean it's a "no boot" condition? it did boot and I was able to start memtest? edit: oh you mean that configuration doesn't tell us anything about slot 2 and 4? but column I does! – user3182532 – 2017-08-02T12:42:49.990
Then you should say so more clearly in the Question! With most dual-channel boards, this would be a "no boot". – Eugen Rieck – 2017-08-02T12:44:02.927
why would column F be a "no boot" with most dual channel boards? – user3182532 – 2017-08-02T12:51:27.260