Memtest86 failure

Llama · 9 2024 05:58

Hi,

I understand that this isn’t a kernel question, but I can’t think about the proper place to put it. I’ve been trying to upgrade RAM on PRO B550M-P GEN3 motherboard from 16GB to 32GB by adding 2 8GB DDR4 plates. The upgrade isn’t a disaster: BIOS and the running Manjaro system see the memory, everything seems normal.

The memtest86 fails:

Is it really just a faulty memory? Is there a possibility that something hasn’t been done quite right by me?

The memtest86 on the original 16GB memory set always ran OK.

Mirdarthos · 9 2024 06:02

Hi @Llama,

Personally, I wouldn’t trust it. Easy as that.

Llama · 9 2024 06:02

Wouldn’t trust what?

Mirdarthos · 9 2024 06:03

The RAM.

linux-aarhus · 9 2024 06:44

Remove the original - replace with the newly aquired - rerun memtest

If memtest does not pass - the ram is faulty - you can trust that.

How memtest works is by writing a specific pattern to a memory address - then read it back - if it does not get the exact same pattern back the memory address is faulty.

robin0800 · 9 2024 10:56

Have you got the right memtest? There are two one for MBR and one for UEFI which one do you need?

Llama · 9 2024 19:52

I didn’t know that there are two kinds of memtest, thanks! Actually, the Manjaro installation is clean and quite recent. I suppose the installer put the right one in place. What’s more, the test used to work on the original 16GB memory set.

zhongsiu · 9 2024 21:58

You shouldn’t really mix RAM. I don’t mean capacity, I don’t even mean brand. I mean if they did not come in the same box, the same kit, you are mixing RAM. Every kit has different characteristics, different timings they can do.
If you have Corsair RAM, you never know what you get. They use everything. Micron, Samsung, Hynix. Often even within the same model and ratings. I got 2 kits of Corsair 3000 Mhz @ CL15. One was Hynix AFR, the other was Samsung C-die IIRC. Very different once you start playing with them. But even at stock, I had to play with them just to get em to work nicely together. I was mixing RAM.

You have 2 options. Buy a kit with the amount of RAM you need (easiest).
Or play with both kits, see what timings they do. I don’t mean just primary timings. You need to know just about every timing. And where they overlap so both kits can do those timings. Very timeconsuming as you can imagine.

If you’ve never done it, expect it to take months. I spent like 1-2 years OCing DDR4. 3 different kits, 2 different CPUs. 8-12 hours a day. I learned a lot. I did it mostly for the fun and learning. RAM OC doesn’t really make much difference, generally less than 5% more perf. On AMD. I got Micron Rev E RAM, 3600 Mhz @ 18-22-22. Found an OC post with timings and all running it at CL 16. Helped me a lot. Still, it took me like a month to get there. I had to play with resistances. That is like 7 settings. And I settled for 4 different values on each. That took me days to test out. That is lots of different variations to test. So it runs at 16-16-19 or so. Trcdrd doesn’t go low on Micron. At 1.55v. Been rocking it for 2-3 years now. Something that will throw a wrench in the machinery is BIOS updates/firmware. Those can invalidate any OC you have done, making it unstable. I avoid those versions of firmware.

I should also add: I did none of this on Linux. The tools just aren’t there.

Kobold · 9 2024 22:58

RAM his highly adjustable… there are different ways to make it stable.

Increase Voltsettings a little from your unstable Ram or increase the Timings CL (slower access times).

Arrababiski · 9 2024 23:38

Did the new plates have the same specs as the former ones? Have you previously changed any memory adjustment in the BIOS?

Clean the module socket with a vacuum cleaner (don’t get too close) and a small brush. Dust on the socket can cause wrong contacts.

Re-test all, but only one memory module each time to know which one is faulty.

Also download a previous version of Memtest to try again and discard a software bug. Systemrescuecd 10 has v6.x:

Llama · 10 2024 06:29

I replaced the 2x8GB original ones with the 2x8GB questionable new ones, in the same slots. memtest86 is running fine. Probably, as zhongsiu suggested, I’m mixing RAM

Curiously, the replacement still required a CMOS reset. Somebody inside knows the RAM is different.

Arrababiski · 10 2024 08:38

Has your Bios a screen like this to see if they are any difference between memory modules?

May be there’s an option “Defaults” where you can select “Load optimized defaults” to let the computer adapt to the hardware you have.

Then install all the memory modules, run this command and copy here the output:

sudo dmidecode -t memory

May be you have an outdated Bios and updating to the last can solve the problem?

Llama · 12 2024 05:11

The BIOS is updated. Maybe I can’t see the option, maybe there’s none. Here’s what won’t pass the tests.

Erim · 12 2024 05:56

It could be that you need to change the order so that the slot DIMMA1 and DIMMA2 contain one brand and DIMMB1 and DIMMB2 contains the other brand.

Have you tried that before?

Arrababiski · 12 2024 06:15

That’s right. Motherboards use to have coloured memory sockets. Always install the same modules in the sockets of the same colour because each pair work in “dual channel mode” to improve performance.

memory

https://www.lenovo.com/us/en/glossary/dual-channel-memory/

Can any computer use dual-channel memory?

Not every computer supports dual-channel memory. To benefit from this technology, your motherboard must have two memory slots that support dual-channel configuration, and you need to install a matching pair of random access memory (RAM) modules.

And @Llama, don’t forget to run sudo dmidecode -t memory and copy here the output.

Edit: searched a photo of the mainboard and all sockets are black. They are numbered DIMMA1, DIMMA2, DIMMB1 and DIMMB2. If you have only one memory module must be in A2. If you have a pair, in A2+B2. If you have four, first pair in A2+B2, second pair A1+B1. Quite weird module order…

Llama · 13 2024 08:40

Thank you very much! I haven’t been systematic enough from the beginning, unfortunately. This is what I gathered so far:

Apacer (a) has device frequency 2666 MHz
Kingston (k) has device frequency 2133 MHz

Native configuration: A1:-, A2:Apacer, B1:-, B2:Apacer; memtest OK
Curiously: A1:-, A2:Apacer, B1:-, B2:Kingston; memtest OK

A1:Apacer, A2:Apacer, B1:Kingston, B2:Kingston
won’t even reach as far as BIOS; the box hangs.

Cases where memtest fails:

One: A1:Kingston, A2:Apacer, B1:Kingston, B2:Apacer; memtest failed
Two: A1:Kingston, A2:Kingston, B1:Apacer, B2:Apacer; memtest failed

For One and Two I ran

sudo dmidecode -t memory

a1_k,a2_a,b1_k,b2_a--runs-test-failed.txt:

# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.8 present.

Handle 0x0011, DMI type 16, 23 bytes
Physical Memory Array
	Location: System Board Or Motherboard
	Use: System Memory
	Error Correction Type: None
	Maximum Capacity: 128 GB
	Error Information Handle: 0x0010
	Number Of Devices: 4

Handle 0x0018, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0017
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL A
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Kingston
	Serial Number: 9638FF14
	Asset Tag: Not Specified
	Part Number: KVR21N15S8/8        
	Rank: 2
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x001B, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001A
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL A
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Unknown
	Serial Number: 01723455
	Asset Tag: Not Specified
	Part Number: D12.2324WH.003      
	Rank: 1
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x001E, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001D
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Kingston
	Serial Number: 9638FF12
	Asset Tag: Not Specified
	Part Number: KVR21N15S8/8        
	Rank: 2
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x0021, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0020
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Unknown
	Serial Number: 01723455
	Asset Tag: Not Specified
	Part Number: D12.2324WH.003      
	Rank: 1
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

a1_k,a2_k,b1_a,b2_a--runs-test-failed.txt:

# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.8 present.

Handle 0x0011, DMI type 16, 23 bytes
Physical Memory Array
	Location: System Board Or Motherboard
	Use: System Memory
	Error Correction Type: None
	Maximum Capacity: 128 GB
	Error Information Handle: 0x0010
	Number Of Devices: 4

Handle 0x0018, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0017
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL A
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Kingston
	Serial Number: 9638FF14
	Asset Tag: Not Specified
	Part Number: KVR21N15S8/8        
	Rank: 2
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x001B, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001A
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL A
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Kingston
	Serial Number: 9638FF12
	Asset Tag: Not Specified
	Part Number: KVR21N15S8/8        
	Rank: 2
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x001E, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001D
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Unknown
	Serial Number: 01723455
	Asset Tag: Not Specified
	Part Number: D12.2324WH.003      
	Rank: 1
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x0021, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0020
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2133 MT/s
	Manufacturer: Unknown
	Serial Number: 01723455
	Asset Tag: Not Specified
	Part Number: D12.2324WH.003      
	Rank: 1
	Configured Memory Speed: 2133 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

MrLavender · 13 2024 12:12

On a MSI B550 motherboard you populate the DIMM slots in this order (see your motherboard manual)

1 stick - DIMM A2
2 stick - DIMM A2 + B2
4 stick - DIMM A2 + B2 + A1 + B1

So if the Kingston was your original 16GB then you had it right.

The problem is mixing different sticks as others have already suggested. Maybe if you have the technical knowledge to manually adjust DDR timings and voltages you could make it more stable. Maybe.

But the performance will be sub-standard. It looks like neither of those kits supports XMP and AMD Ryzen cpus much prefer 3200 MHz (and especially if you’re using the integrated graphics on that 5600G).

I suggest returning the new RAM and then either

Buy a matching 2x8GB kit (still not 100% guaranteed because as @zhongsiu pointed out many manufacturers use different RAM chips even within the “same” kits, but very likely it will work).
Buy a 32GB kit (ideally something like dual-channel 2x16GB 3200 MHz CAS 16 and then, after testing it at default settings, you can turn on the A-XMP Profile 1 button next to CPU in top left).

Llama · 6 2024 17:35

I got 2x16GB kit, put it into A2, B2. A-XMP Profile buttons are no longer gray. memtest86 is passing. The only trouble: the BIOS (updated), as well as memtest86, sees it as 2x8GB kit. What else do I have to check?

BTW, what’s “dual-channel”?

UPD

The legend of a plate: CMR16GX4M2Z2933C16

DDR4 16GB(2X8GB) 2933MHz

Arrababiski · 6 2024 20:13

Your computer is kidding you.

The only thing that comes to my mind now is run again sudo dmidecode -t memory to see what’s happening with your memory…

Double channel: two data buses, the cpu can read/write data in each memory module at the same time, so memory speed doubles (in theory). Thinking in roads instead of bytes, it’s like having a single or double lane road.

Llama · 7 2024 05:14

sudo dmidecode -t memory:

# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.8 present.

Handle 0x0011, DMI type 16, 23 bytes
Physical Memory Array
	Location: System Board Or Motherboard
	Use: System Memory
	Error Correction Type: None
	Maximum Capacity: 128 GB
	Error Information Handle: 0x0010
	Number Of Devices: 4

Handle 0x0018, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0017
	Total Width: Unknown
	Data Width: Unknown
	Size: No Module Installed
	Form Factor: Unknown
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL A
	Type: Unknown
	Type Detail: Unknown

Handle 0x001A, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x0019
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL A
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2667 MT/s
	Manufacturer: Unknown
	Serial Number: 00000000
	Asset Tag: Not Specified
	Part Number: CMR16GX4M2Z2933C16  
	Rank: 1
	Configured Memory Speed: 2667 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V

Handle 0x001D, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001C
	Total Width: Unknown
	Data Width: Unknown
	Size: No Module Installed
	Form Factor: Unknown
	Set: None
	Locator: DIMM 0
	Bank Locator: P0 CHANNEL B
	Type: Unknown
	Type Detail: Unknown

Handle 0x001F, DMI type 17, 40 bytes
Memory Device
	Array Handle: 0x0011
	Error Information Handle: 0x001E
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 8 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 2667 MT/s
	Manufacturer: Unknown
	Serial Number: 00000000
	Asset Tag: Not Specified
	Part Number: CMR16GX4M2Z2933C16  
	Rank: 1
	Configured Memory Speed: 2667 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V