Intel 8400/B360 Has many problems - same with H370 + QVL RAM

[seansmr]

I am not really sure if I should have put this is kernel and hardware, since it is not a specific question. I am not even sure if anyone can help. However, short of switching to Windows or AMD I have no ideas.
My question is: Should I be able to solve these problems or should I buy AMD instead?

I have been using Gentoo since 2005 and I have had a system that worked perfectly for over a decade, until the latest hardware failed. This has had its CHOST changed twice. I also have a desktop, Core2 that is 9 years old, but failing, but Gentoo still works great on it. Gentoo on a Raspberry Pi, works better than Raspbian.

Now I have an Intel 8400 on an ASUS B360M motherboard and I have had no end to troubles with it. I have never been able to use it fully, I have to be careful with what I do, of stuff doesn't work and at worst I lose unsaved work and require a power off. I have never had a situation where I lost any data on a system with Gentoo installed.

Before I spend any time on each issue, I just wanted to know if there were known problems with the CPU or hardware.

What I really want to do is buy 2 more of the same for work and copy the installation in order to save time. But I can't put a system this unreliable at work.

The problems are so disparate that I can't begin to think of a solution. From most significant to least, they are:
1. Kernel panic when access a Windows server using a GUI. I can usually copy files back and forth on the command line, but it may crash, but I cannot access Windows shares with any GUI like Libre Office or Vivaldi, kernel panic is assured. Even when it works I get plenty of errors in dmesg. There is no problem when accessing a Samba server.

2. Cannot use any Sandisk USB device in the USB 3 slots, they don't exist, it is like nothing is plugged in, nothing in dmesg or lsusb. USB 3 works fine with any other brand of USB 3 device. Sandisk usb 3 devices work in the usb 2 ports. xhci is in the kernel

3. One particular wired mouse just fails after working fine for a while, with the pointer going random places on the screen, or not working altogether. It generates constant messages in dmesg even when it does work. This mouse works just fine on any other computer.
That is 3 things that I cannot do at all.

4. Constant general protection faults or segfaults in dmesg that don't seem to affect the performance, they usually cite glibc or some part of xfce4. This has stayed constant through upgrades since April.

I have never had windows installed so I can't compare. I have had OpenSUSE installed, but didn't use it enough to compare except trying the USB 3, If I recall correctly, may have worked rarely but usually not.

I have tried many kernel versions, including ones several versions ahead of the current stable, this has made no difference on any of the 4 issues above. I haven't tried anything recently and the highest kernel that I tried was 4.17.x

I have avoided buying anything new while I still had the problems, but the need has become quite urgent and time is short now.
Does anyone else with similar hardware have perfectly working system?

Thanks if anyone can help.

[Jaglover]

[seansmr]

Sorry, I was rushed.

I suppose occasional segfaults, like 12. There was 1 today, but I didn't use it at all, it just sat on for the day.

How can a new system have bad RAM, maybe I will try a new slot.


Thanks for the response

[NeddySeagoon]

seansmr,

Pick one problem to work on. It will avoid us mixing things up.

If you suspect RAM, boot into memtest86 and run a few cycles.
Unless you run it in place of the kernel, the information it produces is not useful.
Detected failures do not always indicate a RAM problem. Much more of your system is used that just the RAM.
Post the results.

There are some useful CPU stress test programs too. They all make your CPU work hard, so it tests your CPU, cooling and motherboard PSU.

I suspect you are unlucky and its hardware related in some way or problems like yours would be all over Google, like the early days of AMDs Ryzen.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[ct85711]

I agree, the segfaults, can be a indicator of bad ram. One thing to remember on that, is that the ram may have been fine before, then suddenly die on you. If it is a bad stick of ram, it is going to cause problems regardless what OS you are running (may also be a factor on some of the other problems). Sadly, it is one of those things that tends to act up randomly and doesn't really get better (possibly worse).

Any hardware issues, is going to exist in other OS too, so if you think it is software, try a livecd for a while and see if it exhibits the issue too, if so it is a good chance a hardware issue and not software. In that case, reinstalling (or installing a different OS completely) won't help at all.

As far as switching from intel to AMD, you will need to recompile your entire system back to generic before you switch. Most times, it isn't too bad if you are sticking with the same brand (going from intel to intel).

[pjp]

I agree with others, run memtest86. More specifically, I'd recommend to run it from SystemRescueCD or an ISO to make sure the existing OS isn't part of the problem.

[seansmr]

Thanks very much for the replies.

I will try memtest86 from the SystemRescueCD.

It seems to be the consensus that I have an issue with my particular hardware.
I did think that a problem with the product model was unlikely, but I did not think of the possibility of hardware problem, as suggested by all. I was wondering about Gentoo on the latest hardware, but mostly I was busy at work.

So the question now becomes this:
Could I confidently buy a Core i5 8n00, a different ASUS B360 motherboard, copy my current install over, and rely on that as a dev server at work? It will likely need to go 24/7

[NeddySeagoon]

seansmr,

[seansmr]

Run memtest86. It stops running every time. No errors reported, nothing happens on the screen, just a frozen system.
Ran it 2 times and it stopped 2:38 and 2:39
It reports being in block 4096-6144. Is there any way to set the kernel to not use RAM above 4096?

I slowed the memory down and tested again, twice, it stopped again. Didn't do the math, but I think the longer times reflect the slower memory and that it is stopping at the same point. However, I don't know much about memory testing yet.

No errors reported, I don't know what this means. Bad ram, or ram that doesn't work with my motherboard?

With regard to copying the system over, I get the point.
Buying new hardware will let me swap parts, to see what could be the problem.


Thanks the responses and help.

[NeddySeagoon]

seansmr,

Tells us about the RAM layout in your motherboard.
dmidecode may help.

If your system will run with a single stick of RAM, try that.
I suspect that you will discover that every individual stick alone works, its only when you add more that you see problems.

Its possible for RAM channels or sockets to be faulty too. So if nothing works, try a different RAM socket.
Minimise the variables. If your motherboard insists on 2 or ever 3 sticks of RAM being fitted ensure that they are a matched set and in the same colour sockets.

Look at the labels on the RAM and make sure all the RAM is rated the same speed and voltage. Sometimes you can mix and match sometimes not.
Post the Vendors and part numbers if you can't tell.

You are onto something with memtest but we don't know what yet.
memtest is a nice simple program. Until that runs, there is no point in trying anything else.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[seansmr]

Thanks.

I have 1 stick for some reason.

I changed the slot and ran memtest86 again, it stopped at 2:39 again.
It was #7, block move or copy, forget now.

It's Kingston KVR24N17S8/8

[NeddySeagoon]

seansmr,

Test the RAM in a Blue Socket and a Black socket.
If you get the same results in both, its likely the RAM that's faulty but without a substitution test, its not possible to be sure.
There may be a compatibility problem too.

Your motherboard manual says

[seansmr]

NeddySeagoon,

Thanks for your help on this.
I just read that part in the motherboard manual, but didn't know what it meant.

Slowing it down doesn't help. memtest86 stops at the same place.

My desktop is gone now, since swapping RAM slots. It can't be related. I now have a gray screen, Everything works still. I have a backup.

Are you really at 56N 3W?

[NeddySeagoon]

seansmr,

That's my Lat/Long rounded to the nearest degree.
If you put that location into Google maps, you will see that at that spot, its very wet. :)
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[seansmr]

NeddySeagoon,

I found those coordinates to be highly suspicious.
I find that many relatives are either in, or from a location that rounds to those same 2 numbers.

I have just found this, the qualified vendor list:
https://dlcdnets.asus.com/pub/ASUS/mb/LGA1151/PRIME_B250M-C/PRIME-B250M-C_Memory_QVL_Report.pdf

According to this, the KVR24N17S8/4 and KVR24N17D8/8 is supported, however my KVR24N17S8/8 is not on the list. Is this the end of the issue, or is there any further effort to make?


Thanks

[krinn]

[NeddySeagoon]

seansmr,

The nearest large town to that location is Edinburgh, Scotland.

The RAM QVL is usually out of date. It would be very rare for other RAM, with the same timings, not to work too.
There are 100's of RAM stick vendors but only one or two vendors that make the RAM silicon, so unless you get your RAM from a cut price unknown vendor, that uses reject silicon, it should be OK. Kingston are a well known, well respected company, so its not that.

The available data says that your RAM, left at its default settings, is too fast for your motherboard. In the BIOS, you should be able to slow it down.
There is probably an overclocking menu somewhere.
Don't be fooled by the name, you can underclock too. Your RAM won't mind that.

You are, in effect, overclocking, even though you didn't intend to.

I found E13793_PRIME_B360M-A_UM_web.pdf on the web, which I think is the manual for your motherboard. On page 21 of 28 It shows Advanced Mode Bios settings.
The image is dreadful but I think I can see on the left hand side, some DRAM setting options, in the box labelled Menu Items.
One of them might be DRAM Frequency.
In that menu, choose 2133 MHz, save the setting and exit.
Your system will reboot, using that setting.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[seansmr]

Hi,
Thanks for your response.

I have tried the memory at the lower speed. I just wasn't clear in writing about it.

I set the RAM to operate at 2133 MHz and tried memtest86 again. It stopped at the same point in the test again. I have left it at that speed.


About choosing RAM, it has never been a show stopper in the past. I choose a cpu and motherboard from physical shops and then look for RAM that fits the motherboard. They don't put a QVL in the box. The vendor picked the RAM for me.

[NeddySeagoon]

seansmr,

You can try a lower speed if you like.

Its clearly a hardware fault but without another stick of RAM to test with or another system to try the RAM in, its not possible to say if its actually the RAM at fault.
Are you able to test the RAM in another system?
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[seansmr]

Thank for the help on this.

I have tried it as low as 1600 MHz, it stops just the same, and at the same place, I believe.
I have slowed RAM speed as a solution to old and failing hardware long ago. I will try this one even slower just to see.

I think that this is what I have to do. Go back to where I bought it and have them try it out. Or order my new B360 based system with approved RAM, and swap the RAM and see what happens.

It is good to know that there is an identifiable problem with this system, so I think that I can confidently buy the much needed similar one. I appreciate the assistance.

[seansmr]

So, I have bought a new board, ASUS H370, and was able to get RAM on the QVL list.
I have also made up the latest System Rescue CD and run memtest86 on the new board. It stops just like the other one does, at just under 3 minutes in.

This has got to be a problem with memtest86. When it runs the test it says it is DDR3-0, and the timings that it shows are not what are shown in the BIOS.
Is memtest86 suitable for DDR4?

[NeddySeagoon]

seansmr,

The software writes test patterns and reads them back to verify that they are correct.

If memtest86 changes the memory timings, that would be a bug.
If it reads the SPD ROM and displays gibberish, that's more of a feature.

To try the latest version, in case that's not what SystemRescueCD has, memtest86
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.