Question What's the best way to check if your CPU is faulty besides replacing it ?

[sleepaftertoyle]

My PC has been randomly rebooting for months now, ever since I upgraded my CPU. The first months I would get a solid green screen, now it remains black for a few seconds then restarts. I have replaced both SSD, Motherboard and PSU. I do not have any spare parts to try out and I can't afford to keep buying components if I'm not sure they're going to fix things. These crashes happen completely randomly, no matter what I'm doing. They can happen twice a day or twice a month. The only time I can get it to crash consistently is running the 24 hour test in Burnintest. I have tried all sorts of stress tests and only that one can reproduce the crash. And it always crashes within 3-4 minutes of starting the test. I have tested the RAM many, many times with memtest, never detected any errors. Yesterday I ran it with only one RAM stick and it crashed too. So I went to memtest once again to see if it was faulty but again, nothing. I am currently using Windows 10 but the same crash happened with Linux Mint. Not sure if Linux can give me more info on those crashes, I couldn't find much either. BIOS is mostly all default, nothing is overclocked and things like XMP were never manually enabled. I have taken it to several shops now and no one was able to tell me what was wrong. Is there a way to fully diagnose this without replacing everything? I have a few HWInfo logs that were saved while the pc crashed. Not sure if they add any info, I can upload them somewhere.

Specs: Asus Tuf b550 Wifi II
16GB TridentZ RAM
AMD Radeon 5600XT
Ryzen 5 5500
Cooler Master MWE Gold v2 650

 

[CountMike]

My PC has been randomly rebooting for months now, ever since I upgraded my CPU. The first months I would get a solid green screen, now it remains black for a few seconds then restarts. I have replaced both SSD, Motherboard and PSU. I do not have any spare parts to try out and I can't afford to keep buying components if I'm not sure they're going to fix things. These crashes happen completely randomly, no matter what I'm doing. They can happen twice a day or twice a month. The only time I can get it to crash consistently is running the 24 hour test in Burnintest. I have tried all sorts of stress tests and only that one can reproduce the crash. And it always crashes within 3-4 minutes of starting the test. I have tested the RAM many, many times with memtest, never detected any errors. Yesterday I ran it with only one RAM stick and it crashed too. So I went to memtest once again to see if it was faulty but again, nothing. I am currently using Windows 10 but the same crash happened with Linux Mint. Not sure if Linux can give me more info on those crashes, I couldn't find much either. BIOS is mostly all default, nothing is overclocked and things like XMP were never manually enabled. I have taken it to several shops now and no one was able to tell me what was wrong. Is there a way to fully diagnose this without replacing everything? I have a few HWInfo logs that were saved while the pc crashed. Not sure if they add any info, I can upload them somewhere.

Specs: Asus Tuf b550 Wifi II
16GB TridentZ RAM
AMD Radeon 5600XT
Ryzen 5 5500
Cooler Master MWE Gold v2 650

Temperatures and voltages ?

 

[sleepaftertoyle]

Temperatures and voltages ?

Temperatures are fine, at the absolute most it goes to 80-85 Celsius but only when I run programs like Cinebench, it doesn't reach that while gaming. As for voltages which ones do you mean? I can post some screenshots of the HWinfo file but there are way too may categories.

 

[LOGAN54321]

You didn't mention anything about video card ! I guess you don't have another one.

...that one can reproduce the crash. And it always crashes within 3-4 minutes of starting the test.

But a repair shop should have (more than) one. They should also have an AM4 cpu with which could test your system, but let's say that for some reason they only have others. If it only takes 3-4 minutes to crash(using the test from quote), then in 10 minutes or so , they can put their card in and run the test!
If they have an AM4 cpu, it's not that hard , nor does it take a lot of time to swap it and run the test. (if they are asking some ridiculous amount of money, then it's " a different story")
I am saying this because if i understood correctly, you don't have anybody from whom to borrow components.(otherwise borrow a video card, and run the test yourself)

 

[Misgar]

ever since I upgraded my CPU.

Did you keep the old CPU after upgrading? If so, fit the old CPU back into the mobo.

Assuming the system didn't crash with the old CPU, if it crashes when you start using the old CPU again, you've probably bent a pin in the socket.

Was the Ryzen 5 5500 brand new in a sealed AMD box, or second hand and potentially suspect?

If second hand, was the CPU housed in an official AMD plastic clamshell, or wrapped in static-inducing bubble wrap/non-ESD bag? A mis-handled CPU could have died from static discharge.

 

[sleepaftertoyle]

You didn't mention anything about video card ! I guess you don't have another one.

But a repair shop should have (more than) one. They should also have an AM4 cpu with which could test your system, but let's say that for some reason they only have others. If it only takes 3-4 minutes to crash(using the test from quote), then in 10 minutes or so , they can put their card in and run the test!
If they have an AM4 cpu, it's not that hard , nor does it take a lot of time to swap it and run the test. (if they are asking some ridiculous amount of money, then it's " a different story")
I am saying this because if i understood correctly, you don't have anybody from whom to borrow components.(otherwise borrow a video card, and run the test yourself)

I don't have any spare parts besides the components I already replaced unfortunately. And yeah frankly I don't know what exactly they're doing. I suggested swapping out components but I have no idea if they did. I repeatedly mentioned the program I used to reproduce the crash too. I already tried the system with one RAM stick and it crashed. Today I will test it with the other one and if it also crashes then I don't think it's the RAM probably.

In the report I noticed the Core 0 and Core 1 C6 Residency are at 0 at about the time I started the stress test. Is this normal? I have a few more screenshots here:

https://imgur.com/a/oRTFeV6

View: https://imgur.com/a/oRTFeV6

 

[JeffreyP55]

My PC has been randomly rebooting for months now, ever since I upgraded my CPU. The first months I would get a solid green screen, now it remains black for a few seconds then restarts. I have replaced both SSD, Motherboard and PSU. I do not have any spare parts to try out and I can't afford to keep buying components if I'm not sure they're going to fix things. These crashes happen completely randomly, no matter what I'm doing. They can happen twice a day or twice a month. The only time I can get it to crash consistently is running the 24 hour test in Burnintest. I have tried all sorts of stress tests and only that one can reproduce the crash. And it always crashes within 3-4 minutes of starting the test. I have tested the RAM many, many times with memtest, never detected any errors. Yesterday I ran it with only one RAM stick and it crashed too. So I went to memtest once again to see if it was faulty but again, nothing. I am currently using Windows 10 but the same crash happened with Linux Mint. Not sure if Linux can give me more info on those crashes, I couldn't find much either. BIOS is mostly all default, nothing is overclocked and things like XMP were never manually enabled. I have taken it to several shops now and no one was able to tell me what was wrong. Is there a way to fully diagnose this without replacing everything? I have a few HWInfo logs that were saved while the pc crashed. Not sure if they add any info, I can upload them somewhere.

Specs: Asus Tuf b550 Wifi II
16GB TridentZ RAM
AMD Radeon 5600XT
Ryzen 5 5500
Cooler Master MWE Gold v2 650

CPU's rarely die. It is usually the motherboard. You will need to test the CPU somehow. Possibly a friend, acquaintance or my last and never choice, PC repair shop.

 

[sleepaftertoyle]

CPU's rarely die. It is usually the motherboard. You will need to test the CPU somehow. Possibly a friend, acquaintance or my last and never choice, PC repair shop.

This all started happening once I upgraded the CPU so could it be possible it was broken from the start? Also, I already replaced the motherboard. Unless I am the unluckiest person in the world, it would be weird if two different mobos had the exact same problem

Another update, I checked the other RAM stick and it crashed too. But for some reason both of them crashed on the same slot, the one closest to the cooler. I'll keep testing them on the other slot for now. But again, considering I already changed the motherboard what could it be? I had a Gigabyte x370 Gaming K7 before.

 

[JeffreyP55]

This all started happening once I upgraded the CPU so could it be possible it was broken from the start? Also, I already replaced the motherboard. Unless I am the unluckiest person in the world, it would be weird if two different mobos had the exact same problem

Another update, I checked the other RAM stick and it crashed too. But for some reason both of them crashed on the same slot, the one closest to the cooler. I'll keep testing them on the other slot for now. But again, considering I already changed the motherboard what could it be? I had a Gigabyte x370 Gaming K7 before.

Have you been using your user manual so you KNOW which RAM slots to use for different configurations?

TUF GAMING B550-PLUS WIFI II - Support

www.asus.com

 

[sleepaftertoyle]

Have you been using your user manual so you KNOW which RAM slots to use for different configurations?

TUF GAMING B550-PLUS WIFI II - Support

www.asus.com

Yes, they were both always in the correct slots

 

[Misgar]

But for some reason both of them crashed on the same slot, the one closest to the cooler.

The slot closest to the CPU is not the slot recommended for use with a single DIMM, so perhaps that's why you had problems with this configuration.

Alternatively, if one (both) of the 5500's IMCs is faulty, even if the RAM is OK, the test will fail regardless of which socket(s) you use.

I'll keep testing them on the other slot for now.

By this I assume you mean slot B2, furthest away from the CPU, as recommended in the manual (page 1-10).

This all started happening once I upgraded the CPU so could it be possible it was broken from the start?

As I asked earlier, was the Ryzen 5500 purchased brand new in a sealed AMD box, or second hand?

If second hand, there's no telling if the 5500 is good or bad. The previous owner might have overvolted or overclocked it too far, causing electromigration.
https://forums.tomshardware.com/threads/overclocking-amd-ryzen-5500-on-gigabyte-motherboard.3825469/

I had a Gigabyte x370 Gaming K7 before.

Do you still have the old CPU and the Gigabyte X370? If so you could test the RAM again in the old system with Burnintest.

Another update, I checked the other RAM stick and it crashed too.

It's not unknown for two sticks of RAM to fail, especially if both of them were zapped with static at some time in the past. I was gifted over 100 sticks of second hand DDR3/DDR4 memory and 7 of them failed MemTest86. RAM doesn't always stay good for all time. Sometimes it dies.

The only time I can get it to crash consistently is running the 24 hour test in Burnintest.

That's not surprising considering this type of software is designed specifically to reveal weak points. A system might run perfectly OK in most normal scenarios, but still fail when stressed to the "nth" degree.

I suspect some of my "stable" systems would fail Burnintest too, especially if they contain overclocked CPUs, or RAM running high XMP settings. Even at stock BIOS, some Burnintests might prove too much, with less than ideal combinations of RAM, CPU and mobo.

Unless I am the unluckiest person in the world, it would be weird if two different mobos had the exact same problem

Not all negative resuits are indicative of outright failure, if the system will never (or rarely) be subjected to such stress in real life. Are you surprised if a vehicle engine fails during in a 24 hour race, despite the fact it's been through standard stress testing?

As a systems engineer I'm used to running stress tests on equipment at temperatures between -51°C (-60°F) and +55°C (+131°F). I've run drop tests, bump tests, vibration tests, humidity tests, etc., to see if designs fail. Ditto EMC and EMI tests.

You need to consider carefully what tests to run and how to interpret the results. For example, there's no way I'd operate a standard PC at -51°C or +55°C or run a Milspec drop test.

Burnintest may be highlighting real problems with your existing components and if so, you might have to buy some new components. Sorry.

My PC has been randomly rebooting for months now, ever since I upgraded my CPU

This was where it all started to go wrong. Perhaps the 5500 is the culprit after all?

I have replaced both SSD, Motherboard and PSU.

If I'm correct, you haven't (personally) tried any new RAM.

You can continue testing the existing RAM until you're blue in the face, but even if all tests pass, it still might not be best suited to the 5500.

Some RAM kits with XMP SPD timings for Intel CPUs don't work well with AMD CPUs. Hence a few RAM kits come with specific EXPO SPD timings for AMD.

Check the mobo QVL to confirm your exact TridentZ RAM part number is on the list.

N.B. I use QVL as a guide only. I frequently use RAM not on the QVL. Occasionally I buy second hand RAM. Sometimes it works OK, other times not.

Consider buying new RAM. It's one of the few parts you haven't changed. It's a sad fact of life that when things go wrong, it can get expensive (as I know to my cost).

 

[sleepaftertoyle]

The slot closest to the CPU is not the slot recommended for use with a single DIMM, so perhaps that's why you had problems with this configuration.

Alternatively, if one (both) of the 5500's IMCs is faulty, even if the RAM is OK, the test will fail regardless of which socket(s) you use.


By this I assume you mean slot B2, furthest away from the CPU, as recommended in the manual (page 1-10).


As I asked earlier, was the Ryzen 5500 purchased brand new in a sealed AMD box, or second hand?

If second hand, there's no telling if the 5500 is good or bad. The previous owner might have overvolted or overclocked it too far, causing electromigration.
https://forums.tomshardware.com/threads/overclocking-amd-ryzen-5500-on-gigabyte-motherboard.3825469/


Do you still have the old CPU and the Gigabyte X370? If so you could test the RAM again in the old system with Burnintest.



It's not unknown for two sticks of RAM to fail, especially if both of them were zapped with static at some time in the past. I was gifted over 100 sticks of second hand DDR3/DDR4 memory and 7 of them failed MemTest86. RAM doesn't always stay good for all time. Sometimes it dies.


That's not surprising considering this type of software is designed specifically to reveal weak points. A system might run perfectly OK in most normal scenarios, but still fail when stressed to the "nth" degree.

I suspect some of my "stable" systems would fail Burnintest too, especially if they contain overclocked CPUs, or RAM running high XMP settings. Even at stock BIOS, some Burnintests might prove too much, with less than ideal combinations of RAM, CPU and mobo.


Not all negative resuits are indicative of outright failure, if the system will never (or rarely) be subjected to such stress in real life. Are you surprised if a vehicle engine fails during in a 24 hour race, despite the fact it's been through standard stress testing?

As a systems engineer I'm used to running stress tests on equipment at temperatures between -51°C (-60°F) and +55°C (+131°F). I've run drop tests, bump tests, vibration tests, humidity tests, etc., to see if designs fail. Ditto EMC and EMI tests.

You need to consider carefully what tests to run and how to interpret the results. For example, there's no way I'd operate a standard PC at -51°C or +55°C or run a Milspec drop test.

Burnintest may be highlighting real problems with your existing components and if so, you might have to buy some new components. Sorry.


This was where it all started to go wrong. Perhaps the 5500 is the culprit after all?


If I'm correct, you haven't (personally) tried any new RAM.

You can continue testing the existing RAM until you're blue in the face, but even if all tests pass, it still might not be best suited to the 5500.

Some RAM kits with XMP SPD timings for Intel CPUs don't work well with AMD CPUs. Hence a few RAM kits come with specific EXPO SPD timings for AMD.

Check the mobo QVL to confirm your exact TridentZ RAM part number is on the list.

N.B. I use QVL as a guide only. I frequently use RAM not on the QVL. Occasionally I buy second hand RAM. Sometimes it works OK, other times not.

Consider buying new RAM. It's one of the few parts you haven't changed. It's a sad fact of life that when things go wrong, it can get expensive (as I know to my cost).

When I meant closest, I mean closest of the two recommended for a dual stick configuration. I did check the manual and I'm using only those two slots. I did try the other two out of curiosity and it would turn on and off again on a loop.

The CPU is brand new but unfortunately it's been over a year so I don't have anymore warranty with the place I bought it from. Only option I have now is to send it to AMD if it's faulty. And yes I do have my older components. I might check them again but as I said everything was working perfectly, zero crashes whatsoever until I changed the CPU.

I understand what you say about stress tests, but the thing is these keep happening regularly. Now more often than before. I only use this pc for gaming but it has crashed on desktop and while browsing the web too. Burnintest is just what I use to check things faster, otherwise I would have to spend hours and even days until it crashes again.

I checked the QVL list, it seems to work but the XMP/EXPO section is marked as 'N/A'. Either way, I'm pretty sure it was never enabled in the first place. I might have tried enabling it but if I did it did not fix things. Pretty sure it's disabled now though.

Anyway, I'm mentioning again that I have a few reports of HWInfo recording everything up until the system crashed. I posted some screenshots above but if you understand these things I could upload the whole thing? I noticed a few weird numbers but again I don't know what many of those mean. The most notable one is that both the Core 0 and Core 1 C6 residency drop to 0 as soon as I started the stress test. While the other cores did not. These numbers are similar in an earlier file too. Again, not sure what it means or if it means anything at all but I thought it was worth mentioning