PC not posting & BIOS flash stuck

[at4nas]

Hello dear people,

I recently built my first PC and noticed that the EZ Debug LEDs light up for the CPU and DRAM. After some troubleshooting, I thought that a BIOS flash may help, but now it's stuck on flashing forever.

Here is what I did:

• - reseated all the parts, and tried each RAM stick by itself in each slot --> no change;
• - pressed the CMOS button, unplugged, and then held the power button until all the electricity got discharged --> worked on an open bench, which allowed me to use the PC, set it up and restart it several times throughout the process, but then when I assembled it and put it back in an upright position, the issue came back;
• - checked to see if there was pressure on any part that could buy any chance be causing this --> I didn't see anything that could be a cause for concern;
• - followed MSI's official instructions on how to prepare a USB drive to perform a BIOS flash --> it's been more than half an hour, and it is still flashing (blinking every second or so, with no changes to the speed of blinking or anything else)...

The PC parts are:
- Motherboard: MSI B650i Edge WiFi Gaming
- CPU: AMD Ryzen 7 9800X3D
- GPU: AMD Radeon 9070 XT Mercury XFX OC
- RAM: Corsair Vengeance RGB 96GB (2X48GB) DDR5 PC5-44800C32 6400MHz (CMK96GX5M2B6400C32)

My main concern is that despite everything I've messed something up, and the motherboard may get bricked. My other concern is that the CPU and DRAM lights will continue to stay lit up whenever I try to boot the PC, preventing the PC from booting normally, and I have no idea what's causing it.

At this point, I would gladly take the PC to a technician and have them diagnose the issue and do whatever is necessary, but I can't, since the flashing is still ongoing, and reading from other forum posts, it may go on like that for who knows how long...

I read somewhere that if the BIOS flashing is taking that long, it may not even have started, and it's just stuck looping -- wonder whether there's any credibility to that statement.

Is there a way to interrupt the BIOS flash, stop it without bricking the motherboard? Should I wait it out? What should I do to prepare in the event that it happens and I have to RMA the motherboard to MSI? Any other suggestions on what I can try to find a solution to any of these?

Thank you so much for taking the time to read and for trying to help out, whoever you are.

 

[ericj159802e9]

followed MSI's official instructions on how to prepare a USB drive to perform a BIOS flash --> it's been more than half an hour, and it is still flashing (blinking every second or so, with no changes to the speed of blinking or anything else)...

I take it you are using the Flash BIOS procedure (the procedure without entering the BIOS). Is that correct? It that is correct and the blinking does not stop after ten minutes the flash was not successful and can be aborted.

The Flash BIOS procedure is quite pciky about USB sticks it will work with and that's the problem you are facing. You will have to try multiple USB sticks to figure out which one works until you find the one that works. In short, round up as many USB sticks with a maximum size of 32 GB from people around you as you can.

Les us know how things go.

 

[at4nas]

Thank you, ericj!! Feels like a weight has been lifted off my chest.

I'll line up all the drives I have that meet the criteria, and try again. Will let you know, once I have something to report.

When I still had access to the PC while on an open bench, I managed to update the BIOS to the latest version.

Do you reckon that doing it again would help with the other issue (CPU and DRAM lights, EZ Debug LED), or would it be better to try something else?

 

[laportein151f02d0]

worked on an open bench, which allowed me to use the PC, set it up and restart it several times throughout the process,

I am going to guess what this means. If you reached Click Bios on your bench or even got Windows to start, then you put your components back into your case, that is where your problem is, with CPU RAM ezdebug leds lighting up. Could be a short circuit somewhere, improper wiring. You do not have to write a book, but a few paragraphs could help. We are picturing in our minds what is going on based on the information you give us. Base on the above quote from you, your build was working on the bench.

 

[ericj159802e9]

worked on an open bench, which allowed me to use the PC, set it up and restart it several times throughout the process, but then when I assembled it and put it back in an upright position, the issue came back

Sharp observation from @laportein151f02d0 which I had seen but forgot about.

The problem may be with the mounting of the CPU cooler. When the pressure is not evenly distributed on all four screws the motherboard will curve slightly and some pins may not connect to the CPU. Make sure to tighten the four screws evenly. Also tighten them stepwise: one by one and put the pressure on in steps (don't tighten a screw in one go). Let us know if that helped.

 

[citay]

If it's still all installed in the case, then as the first little test, you could put the case on the right side, so the CPU cooler sits upright again, as it was in the open bench test. Then, you can take out things which are not necessary to get past those LEDs, like the GPU. Lastly you can take out the board again and check if it works as you previously had it. You can also check for bent pins in the socket, also see here and the following posts.

It's my long-standing suspicion that a lot of the cases of "CPU+DRAM LEDs" have to do with suboptimal contact between CPU and board somehow. It's alarming how often this issue happens on MSI AM5, often times for no apparent reason.

If the board was working before and you updated the BIOS to the latest version, then it having stopped working in a mounted position has nothing to do with the BIOS. However, if the Flash BIOS Button tried to flash it now and failed, it might have additionally put the BIOS into a corrupted state, which may or may not also trigger those LEDs (although, normally one would expect the CPU LED only then). So it would also be good to get a successful BIOS flash under the belt now. Of course, not necessary if some of the above makes it boot to BIOS again, then the BIOS doesn't need to be flashed.

 

[at4nas]

Thank you laportein, ericj, and citay for the insights! I have made some attempts and can now report my findings.

• - removed all components but the CPU, motherboard, and PSU --> CPU+DRAM LEDs light up still (even though there's no RAM?);
• - checked the motherboard for bent pins in the CPU socket --> no bent pins, as can be seen on these photos;
• - checked the screws holding the motherboard to see if any unevenly distributed pressure may have bent it --> noticed a bend on the motherboard, and after tuning the screws as instructed, I could not see a meaningful difference --> I took these photos of the motherboard from the GPU side and from the cooler side, so that you could verify its condition.

• Could it be that the cooler and GPU may have caused the motherboard to bend like that from the time the PC has spent flat while on open bench?

Since I had to remove the thermal paste of the CPU to check the pins, I decided to leave the CPU without thermal paste, with the cooler on, and powered the PC on to see which EZ Debug LEDs would light up. It was the CPU+DRAM combo again. After about 5 minutes, the LEDs did not change, and I noticed the cooler had warmed up. Once I removed it and touched the CPU, it was warm, going on almost hot. I reapplied the thermal paste, and powered the PC on again. The CPU cooler still felt warm to the touch after several minutes.

Is it the case that the CPU is running at full throttle? If so, is that normal behaviour considering the PC is not even booting?

I have yet to try and flash the BIOS again. Should I still try and do it?

I am very confused as to what is causing this. Below is a recap of everything I have done until now, in as much detail as I can recollect. If you can surmise as to what I may have done wrong, and what else I could try, please do.

Upon assembly, I had an issue with the PC not powering on, and had a technician look at it. He found out the issue was the power button, which was causing the PC to boot every 10 clicks or so. But whenever the button would work, the PC would post. Once I got a replacement power button from the case manufacturer, the PC would boot, and I could enter the UEFI/BIOS.

From there, I tried to install CachyOS from a USB stick. I got the usual menu with installation options, but whichever installation option would not begin, and the following message would appear:

Booting 'CachyOS'
error: loader/efi/linux.c:grub_arch_efi_linux_boot_image:227:cannot load image.
Failed to boot both default and fallback entries.
Press any key to continue.

After trying all USB ports, flashing the image on the stick once again, and being greeted with the same message, I decided to go with Windows. It worked from the get-go.

The night before that I had tried to OC the RAM to 6400MHz from the default 4800MHz by choosing the only profile available, which was 'SK Hynix 6000MHz' or something similar, and manually changing the clock speed from 6000MHz to 6400MHz without touching anything else. The PC refused to boot several times, so I removed the power cord from the socket, and left it like that overnight. When I tried to install Windows the following morning, I did not encounter any errors. The RAM clock speed had reverted to 4800MHz. Throughout this time, the PC was upright, and completely assembled.

Once Windows was installed, I attempted to OC the RAM to 6400MHz again, and it failed, again. I don't remember what exactly happened and what I did after that, but after several attempts at booting, the RAM had reverted to 4800MHz. I may have removed the RAM and placed it back again, but I'm not sure -- apologies.

With the PC working again, I then tried to OC the RAM again, but this time, without changing anything manually. The PC booted into Windows after several minutes of the RAM being tested, and it was now running at 6000MHz. I used it for several hours without issues, restarting whenever necessary. At this point, I think I had already placed the PC flat, and had opened it up, without removing any parts except for some of the fans.

Next, I turned off the PC, removed the power chord from the socket, put the fans back, closed the case, and put the PC upright. I'm not sure if it was then that it stopped booting, or 1-2 restarts later.

Reading this summary myself, I now think I may have done something by OC-ing the RAM like that. My CPU is AMD, and this is an XMP kit, but from what I understand, unless I tinker with specific settings, it should still work, and it did indeed work, though not at full capacity. Perhaps this and the motherboard being slightly bent would explain things...

Ever since assemblying the PC, and putting it upright, it has not been able to post. I even tried different sockets and cables to see if it would change anything. For the moment, I'm out of ideas, and slightly distressed. Unless we manage to find a solution, I am considering leaving the PC with a technician again, with the hope that they will manage to find and fix whatever is causing this.

 

[citay]

- removed all components but the CPU, motherboard, and PSU --> CPU+DRAM LEDs light up still (even though there's no RAM?);

Did you put the PC on the side first? If we think that it could have something to do with the forces that the cooler exerts on the CPU and the socket, then orienting the cooler parallel to gravity (instead of perpendicular to it) should distribute those forces evenly and perhaps achieve something.

- checked the motherboard for bent pins in the CPU socket --> no bent pins

Agreed, looks good.

- checked the screws holding the motherboard to see if any unevenly distributed pressure may have bent it --> noticed a bend on the motherboard, and after tuning the screws as instructed, I could not see a meaningful difference --> I took these photos of the motherboard from the GPU side and from the cooler side, so that you could verify its condition.

That's normal. In the past, the boards used to have much more serious bends around the socket. Of course, there's a slight chance that MSI AM5 boards have some odd design flaw where, if used with certain coolers or whatever, it can lead to CPU+DRAM LEDs because this is what causes a contact problem in the socket. But then it would also be "by design" because we also see this CPU+DRAM LEDs problem a lot from other people, so it wouldn't be unique to your system. And this bend in general is something that could be observed for as long as there were boards with a similar socket design. Even in the Pentium 4 days you could have some quite significant bending going on around the socket.

Could it be that the cooler and GPU may have caused the motherboard to bend like that from the time the PC has spent flat while on open bench?

For sure the cooler mounting will cause a bit of bending, the torque is rather high for good thermal contact. But this is nothing to be feared, it's normal for such a design.

Is it the case that the CPU is running at full throttle? If so, is that normal behaviour considering the PC is not even booting?

It means the CPU is getting powered. If it was running at full throttle, then the cooler would get hot very quickly (if the fan isn't extracting some heat).

I have yet to try and flash the BIOS again. Should I still try and do it?

Yes, as mentioned, because we cannot rule out a corrupt BIOS from the earlier attempt. So it's good to flash it successfully now (5-8 minutes of flashing).

From there, I tried to install CachyOS from a USB stick. I got the usual menu with installation options, but whichever installation option would not begin, and the following message would appear:

Seems like an issue with Secure Boot. The BIOS nowadays is set up mainly for Windows 11, with advanced security mechanisms in place to prevent unauthorized access during early boot and so on. If you use a Linux-based OS and they don't have a good workaround for this, sometimes you have to disable things like Secure Boot in the BIOS, then the other OS can work properly.

The night before that I had tried to OC the RAM to 6400MHz from the default 4800MHz by choosing the only profile available, which was 'SK Hynix 6000MHz' or something similar, and manually changing the clock speed from 6000MHz to 6400MHz without touching anything else.

Yeah well, this kit is optimized for Intel platforms, and AM5 CPUs have a weaker IMC (integrated memory controller). Still, you should have an "A-XMP" profile of DDR5-6400 available, just by enabling A-XMP in the BIOS. You were probably looking under "MemoryTryIt!" or something.

But for the OS installation, you can even keep XMP/EXPO disabled, because you want the system to be rock stable during that. Once the installation is done, then you can install some test software like OCCT, and then you can enable XMP/EXPO and do some testing.

By the way, DDR5-6000 is the "sweet spot" already, this speed is relatively easy to achieve while extracting most of the performance from DDR5, and doesn't require much tinkering. If you want to get some quick and easy improvements, then with DDR5, you can simply set the "tREFI" (Refresh Interval) timing to 32768, because tuning this value gives the most benefit for DDR5. It gives a nice improvement, because the RAM spends less time refreshing itself and more time performing actual RAM commands. Usually this is no problem.

Once Windows was installed, I attempted to OC the RAM to 6400MHz again, and it failed, again. I don't remember what exactly happened and what I did after that, but after several attempts at booting, the RAM had reverted to 4800MHz. I may have removed the RAM and placed it back again, but I'm not sure -- apologies.

The RAM failed memory training at DDR5-6400, because one "weakest link" of the memory system couldn't handle that speed to the satisfaction of the BIOS routines. If XMP DDR5-6400 doesn't work, then you can keep XMP enabled, but set "DRAM Frequency" to DDR5-6000 by hand before pressing F10 to save&exit. This will apply XMP timings at DDR5-6000.

Reading this summary myself, I now think I may have done something by OC-ing the RAM like that. My CPU is AMD, and this is an XMP kit, but from what I understand, unless I tinker with specific settings, it should still work, and it did indeed work, though not at full capacity. Perhaps this and the motherboard being slightly bent would explain things...

This being an XMP kit primarily for "Intel 700 Series, Intel 800 Series" (as per Corsair), it's not ideal to use on AM5, where an EXPO kit is usually the safer bet. You will see that for the EXPO kits, you can't have certain capacity / speed combinations, they tend to be slower than the Intel-optimized XMP kits. That's because, as mentioned, the IMC on AM5 CPUs generally is weaker, so they can't strive for the same goals on this platform, it would have a lower chance of working.

Speaking of capacity, 96 GB total in what looks to be a gaming-focused system (considering the choice of CPU and GPU) is complete overkill. Much of the RAM will just lay dormant, and using such a high-capacity kit just makes it that bit harder to run problem-free at nice speeds. Whereas with the kits using the smaller 16 GB or 24 GB modules, which are so-called single-rank, it is much less stressful for the memory system. Most games don't need more than 16 GB RAM total (because they also have to run on consoles), a handful of games use more, but with a 32 GB kit, you're usually golden, no need to get more. It will just drain your wallet for no benefit, unless you also have professional workloads which require tons of RAM.

Having said all that, simply attempting to run at XMP will not damage anything on the board. Either it passes memory training or not.

As for the bend, as I said, this is rather normal for decades now, although I will not completely rule out that MSI messed something up in the general design of these boards where physical forces like the cooler mounting mechanism's torque can trigger the CPU+DRAM LEDs issue sometimes. Getting these CPU+DRAM LEDs is by far the most annoying thing I read about with MSI AM5, because there is no easy way to solve this.

If you were to give this to a technician, then they should test with a known good AM5 CPU. Because technically, a bad CPU could cause these LEDs too, since the memory controller also sits inside the CPU, hence the CPU+DRAM LEDs could make sense for that too.

First, I would perhaps focus on getting a good BIOS flash now.

 

[autotech21200]

If it runs good at 6000 for the ram that is where I would leave it. I use 2x32 gig sticks and at 6000 it runs great.

 

[at4nas]

Wow, citay, thank you so much for the comprehensive and lightning-fast response!! Insightful comments and helpful advice all around. You seem to have encyclopaedic knowledgeable about this stuff.

Did you put the PC on the side first? If we think that it could have something to do with the forces that the cooler exerts on the CPU and the socket, then orienting the cooler parallel to gravity (instead of perpendicular to it) should distribute those forces evenly and perhaps achieve something.

Yes. I put it on the side - the motherboard flat, with the cooler on top of it, parallel to gravity. The photos I took were from a different position, though, so I understand it may have seemed confusing.

Agreed, looks good.

Thank you for confirming the pins' condition. It's reassuring.

That's normal. In the past, the boards used to have much more serious bends around the socket. Of course, there's a slight chance that MSI AM5 boards have some odd design flaw where, if used with certain coolers or whatever, it can lead to CPU+DRAM LEDs because this is what causes a contact problem in the socket. But then it would also be "by design" because we also see this CPU+DRAM LEDs problem a lot from other people, so it wouldn't be unique to your system. And this bend in general is something that could be observed for as long as there were boards with a similar socket design. Even in the Pentium 4 days you could have some quite significant bending going on around the socket.

For sure the cooler mounting will cause a bit of bending, the torque is rather high for good thermal contact. But this is nothing to be feared, it's normal for such a design.

Huh, so the bend itself is normal, but it's still possible that it may be affecting this specific motherboard brand and/or model... Note taken.

It means the CPU is getting powered. If it was running at full throttle, then the cooler would get hot very quickly (if the fan isn't extracting some heat).

Oh, okay, got it. Thanks for clarifying this!

Yes, as mentioned, because we cannot rule out a corrupt BIOS from the earlier attempt. So it's good to flash it successfully now (5-8 minutes of flashing).

I saw your reply, so I went ahead and performed a BIOS flash. The USB I used was a single 16 GB, FAT32, Basic MBR partition formatted at the default allocation bus size. The file was the latest BIOS from the official MSI website, renamed as MSI.ROM, and put by itself in the drive directory. I pressed the BIOS Flash button and held it for several seconds. Red, static light appeared. Once I unpressed it, it began flashing. Several minutes later, I returned, and there was no light flashing anymore. I unplugged the PC power chord, unplugged the USB drive, plugged the PC power chord again, and pressed the power button. I was greeted with a red-orange EZ Debug LED combo, CPU+DRAM, yet again.

Seems like an issue with Secure Boot. The BIOS nowadays is set up mainly for Windows 11, with advanced security mechanisms in place to prevent unauthorized access during early boot and so on. If you use a Linux-based OS and they don't have a good workaround for this, sometimes you have to disable things like Secure Boot in the BIOS, then the other OS can work properly.

I see, so Secure Boot it was, after all! I was aware that I may have to turn it off for Linux installations, so I looked at some instructions, but I couldn't find it. Perhaps I've missed something.

Yeah well, this kit is optimized for Intel platforms, and AM5 CPUs have a weaker IMC (integrated memory controller). Still, you should have an "A-XMP" profile of DDR5-6400 available, just by enabling A-XMP in the BIOS. You were probably looking under "MemoryTryIt!" or something.

But for the OS installation, you can even keep XMP/EXPO disabled, because you want the system to be rock stable during that. Once the installation is done, then you can install some test software like OCCT, and then you can enable XMP/EXPO and do some testing.

By the way, DDR5-6000 is the "sweet spot" already, this speed is relatively easy to achieve while extracting most of the performance from DDR5, and doesn't require much tinkering. If you want to get some quick and easy improvements, then with DDR5, you can simply set the "tREFI" (Refresh Interval) timing to 32768, because tuning this value gives the most benefit for DDR5. It gives a nice improvement, because the RAM spends less time refreshing itself and more time performing actual RAM commands. Usually this is no problem.

Okay, more useful, new information! I was not aware about the weaker IMC on AM5 CPUs, that software you mentioned, and that I could tune a single value to get additional benefits for DDR5. Thank you for sharing this!

The RAM failed memory training at DDR5-6400, because one "weakest link" of the memory system couldn't handle that speed to the satisfaction of the BIOS routines. If XMP DDR5-6400 doesn't work, then you can keep XMP enabled, but set "DRAM Frequency" to DDR5-6000 by hand before pressing F10 to save&exit. This will apply XMP timings at DDR5-6000.

So that's what it was... Going forward, I will stick to the settings you recommended.

This being an XMP kit primarily for "Intel 700 Series, Intel 800 Series" (as per Corsair), it's not ideal to use on AM5, where an EXPO kit is usually the safer bet. You will see that for the EXPO kits, you can't have certain capacity / speed combinations, they tend to be slower than the Intel-optimized XMP kits. That's because, as mentioned, the IMC on AM5 CPUs generally is weaker, so they can't strive for the same goals on this platform, it would have a lower chance of working.

Speaking of capacity, 96 GB total in what looks to be a gaming-focused system (considering the choice of CPU and GPU) is complete overkill. Much of the RAM will just lay dormant, and using such a high-capacity kit just makes it that bit harder to run problem-free at nice speeds. Whereas with the kits using the smaller 16 GB or 24 GB modules, which are so-called single-rank, it is much less stressful for the memory system. Most games don't need more than 16 GB RAM total (because they also have to run on consoles), a handful of games use more, but with a 32 GB kit, you're usually golden, no need to get more. It will just drain your wallet for no benefit, unless you also have professional workloads which require tons of RAM.

Having said all that, simply attempting to run at XMP will not damage anything on the board. Either it passes memory training or not.

Thank you for confirming that just setting the RAM to XMP wouldn't damage the motherboard! I was worried it may have.

And yes, about EXPO & that capacity... I would have preferred an EXPO kit, but when I was looking for RAM, ironically, this was the best value. Same about the capacity: my intention was to get 64 GB for productivity purposes, and I came upon someone who wanted to get rid of 2 kits of 2x48GB RAM, so I bought them and sold the second kit. That way, the price for this kit ended up costing only slightly more than it would've cost at the start of the rampocalypse. A bit of a roundabout way of saying it felt like a bargain at the time, but now that I know what you have told me, maybe it wasn't.

As for the bend, as I said, this is rather normal for decades now, although I will not completely rule out that MSI messed something up in the general design of these boards where physical forces like the cooler mounting mechanism's torque can trigger the CPU+DRAM LEDs issue sometimes. Getting these CPU+DRAM LEDs is by far the most annoying thing I read about with MSI AM5, because there is no easy way to solve this.

Hmm... Even if MSI AM5 motherboards get this error so often, I wonder if doing something on my end could help, such as switching to a different cooler. This one is a desktop-sized cooler, Peerless Assassin 120 SE. What do you reckon?

If you were to give this to a technician, then they should test with a known good AM5 CPU. Because technically, a bad CPU could cause these LEDs too, since the memory controller also sits inside the CPU, hence the CPU+DRAM LEDs could make sense for that too.

If we find no solution, I will take the PC to a technician, and mention this to them!

First, I would perhaps focus on getting a good BIOS flash now.

Do you think the BIOS flash worked, considering how the flashing had stopped after a couple of minutes? This time, it took less than 5 minutes, I think. Granted, the BIOS version I used was the same as the one I had already updated to when I still had access to the BIOS, but still.

When I was reading about RAM, I came upon a tool called MemTest86, which could eventually check for any errors in the RAM, but I don't know if it'd even be relevant. Firstly, the PC doesn't reach BIOS. Secondly, the RAM LED still lit up despite the RAM not being inserted when I had removed all the components in order to see which LEDs would light up.

The latter makes my mind go in the direction of a bricked motherboard again. I mean, it could be the CPU too, as you said. If we reach a point where nothing else can be done on my end, then probably a technician would be my best bet, because they'd have plenty of PC parts to use as substitutes and accurately diagnose which component is failing the build.

If it runs good at 6000 for the ram that is where I would leave it. I use 2x32 gig sticks and at 6000 it runs great.

Thank you, autotech, for sharing your experience. You're probably right. After the issues I had when I tried to go all the way, I wouldn't want to push it further. So, 6000MHz it is.

 

[citay]

Several minutes later, I returned, and there was no light flashing anymore.

It's better to babysit this process, so you can verify it lasting 5-8 minutes followed by a reboot (you will see when the reboot happens by the EZ Debug LEDs starting to cycle through again). If you want, we can consider the BIOS successfully flashed. But from what you said, I'm not 100% sure yet.

I see, so Secure Boot it was, after all! I was aware that I may have to turn it off for Linux installations, so I looked at some instructions, but I couldn't find it. Perhaps I've missed something.

Tends to be under Settings\Security\Secure Boot. First you have to set the BIOS to Advanced View (F7).

And yes, about EXPO & that capacity... I would have preferred an EXPO kit, but when I was looking for RAM, ironically, this was the best value. Same about the capacity: my intention was to get 64 GB for productivity purposes, and I came upon someone who wanted to get rid of 2 kits of 2x48GB RAM, so I bought them and sold the second kit. That way, the price for this kit ended up costing only slightly more than it would've cost at the start of the rampocalypse. A bit of a roundabout way of saying it felt like a bargain at the time, but now that I know what you have told me, maybe it wasn't.

Hey, if you can get that much RAM at the price from half a year ago, by all means, it was a good investment. However, with these large modules, they won't run so easily at higher speeds on AM5, you saw it for yourself, the Intel XMP profile wasn't to the board or CPU's liking. So these "productivity purposes" would have to be pretty concrete to justify this. Sure, you could just set them to DDR5-6000 and enjoy the bargain. But if your workloads don't need anywhere near this much RAM later, then you got a lot of value sitting in your board for nothing. And in one or two years, that value might decrease drastically, when RAM production capacity has finally been scaled up a bit more.

So if even the initially planned 64 GB were more of a vague idea or a "nice to have", and you didn't have any specific workloads in mind that would actually require that, then at the current RAM situation (insanely high prices and you perhaps not benefiting from this amount), it might be smarter to sell this kit for an appropriate sum, and get a kit of 2x 16 GB or 2x 24 GB DDR5-6000 for example. Because that 2x 48 GB kit you got there, they go for what, 1000-1500 EUR/USD right now, if they're even available? So you could pocket a large sum right now by swapping this with the amount of RAM you actually need.

Of course, you have a different problem to deal with first, but keep this in mind. Yes, this RAM was a huge bargain, but this RAM won't be worth 1500 bucks in two years anymore, and it's unclear if you will have benefited from this amount until then.

Hmm... Even if MSI AM5 motherboards get this error so often, I wonder if doing something on my end could help, such as switching to a different cooler. This one is a desktop-sized cooler, Peerless Assassin 120 SE. What do you reckon?

That's a really popular cooler model because it's cheap but decent. At the most, I would do what ericj159802e9 posted before, play with the mounting pressure, loosening the screws by half a turn for example and then trying to boot again. Repeat a couple times.

If we find no solution, I will take the PC to a technician, and mention this to them!

If they're worth anything, they will know to rule out parts by leaving them out (as you have done for a lot already), or where that is not possible (CPU, PSU), try with known good ones. They can even try with a different cooler, if they have to get to the CPU anyway. In a lot of these cases, the MSI board is the common denominator. It's just so frustrating because we don't know why this happens on MSI AM5. Well, it could also happen for other board brands with a different symptom, but I'm only here on the MSI forums. It's not like this is happening dozens of times a week, but I see perhaps a couple cases each week, so something is not as it should be.

When I was reading about RAM, I came upon a tool called MemTest86, which could eventually check for any errors in the RAM, but I don't know if it'd even be relevant. Firstly, the PC doesn't reach BIOS. Secondly, the RAM LED still lit up despite the RAM not being inserted when I had removed all the components in order to see which LEDs would light up.

Memtest86 is good for finding a faulty module, I like to use it before installing the OS to find any early faults, see my RAM thread. But yeah, right now it's useless. With any of the first three EZ Debug LEDs, the problem has to be solved first, there is no hope of seeing the BIOS.

The latter makes my mind go in the direction of a bricked motherboard again. I mean, it could be the CPU too, as you said. If we reach a point where nothing else can be done on my end, then probably a technician would be my best bet, because they'd have plenty of PC parts to use as substitutes and accurately diagnose which component is failing the build.

Right. If it's narrowed down to the board, it could be something quite simple, that they can fix in ten minutes at MSI, who knows. But we are still in the dark about it, and MSI have not officially acknowledged anything about this issue or mentioned it anywhere, to my knowledge. But when all other components are proven good or not involved, then you are only left with an RMA of the board, I'm afraid. I wish I knew some trick to solve this, it's quite annoying not being able to help.

 

[laportein151f02d0]

I appreciate the detail you have now provided. You were able to enter Click Bios on the first try, bad power switch notwithstanding, congratulations. Then Linux was your operating system of choice. You were unsuccessful with CatchyOS, possibly due to secure boot issues. You switched to Windows and had a fully functioning pc. You tinkered with the ram speed and then things went bad, you can no longer get into Click Bios, and have CPU RAM ezdebug leds lit, but most likely not because of the ram tinkering. You wrote about removing and reinstalling fans, and that is when your computer no longer booted. Something in that process changed. I keep thinking "something is touching that should not be touching" or "something is not touching that should be touching". In your narrative, you had a working system but then made physical changes to your hardware. Can you start over again? Give it another try before turning it over to a tech? You do not need the GPU, you can use the HDMI port on the backpanel. You do not need any storage devices. Don't need fans, just disconnect the wiring. CPU cooler pressure is up to you. I would even remove the motherboard and put it back just to make sure there is nothing underneath it. Good luck and may the green ezdebug light shine upon you.

 

[ericj159802e9]

You switched to Windows and had a fully functioning pc. You tinkered with the ram speed and then things went bad, you can no longer get into Click Bios, and have CPU RAM ezdebug leds lit, but most likely not because of the ram tinkering.

This had me thinking of what I had just read about "complete BIOS flush" and may be worth trying in this situation:

When you update your BIOS, are you making sure to do a complete BIOS flush (power down, switch off or unplug Power Supply, hold down the BIOS Reset button on the back (middle button) for 10-15 sec, then go into the BIOS and for good measure, load bios defaults then save and restart before going into your BIOS. Early on, MSI had some issues with retaining some old settings in different fields and it required the BIOS to be reset during updates. Not sure if that screwball behavior has returned.

 

[at4nas]

Thank you all for replying once again! I feel encouraged to push through because of your continuous suggestions.

It's better to babysit this process, so you can verify it lasting 5-8 minutes followed by a reboot (you will see when the reboot happens by the EZ Debug LEDs starting to cycle through again). If you want, we can consider the BIOS successfully flashed. But from what you said, I'm not 100% sure yet.

Tried again for good measure. This time, I noticed the flash drive prompted the red light to appear, but several seconds later, it turned off. I guess it didn't actually start the BIOS flash process, which means the previous try was unsuccessful as well. Good thing we tried again then!

Next, I tried with another drive. I didn't have drives under 32 GB at hand, so I opted for a 64 GB Sandisk Ultra Dual Drive. This time, once I held the BIOS Flash button for several seconds, there was no static light -- it immediately started flashing normally (~1 flash/second), and then it accelerated (~2 flashes/second). Not even 3 seconds pressing the BIOS Flash button had the other parts turned on as well (I thought the PC was supposed to turn on after the BIOS flash had finished?). The whole PC continued running like that for almost exactly 5 minutes. It then turned off, and several seconds later, it turned on again; this time, the red light indicating the BIOS Flash running was no longer on.

Currently, the PC remains on, and the EZ Debug LEDs indicate an issue with the DRAM+CPU, as up until now. Should I do anything else from this state? Can I turn it off?

Tends to be under Settings\Security\Secure Boot. First you have to set the BIOS to Advanced View (F7).

Thank you! I will look for it once I get to the BIOS, hopefully.

Yes, this RAM was a huge bargain, but this RAM won't be worth 1500 bucks in two years anymore, and it's unclear if you will have benefited from this amount until then.

...You're absolutely right, and I admit that I hadn't thought about this -- perhaps in part to avoid the unpleasant feeling if the answer is no. That said, now that you've put it in my face, it'd be a good time to take your words into consideration.

At the most, I would do what ericj159802e9 posted before, play with the mounting pressure, loosening the screws by half a turn for example and then trying to boot again. Repeat a couple times.

I tried doing this with the cooler parallel to gravity. I didn't do it many times, but there was no change. I may try again just in case.

If they're worth anything, they will know to rule out parts by leaving them out (as you have done for a lot already), or where that is not possible (CPU, PSU), try with known good ones. They can even try with a different cooler, if they have to get to the CPU anyway.

I am glad you said this. They had confirmed previously that it's part of their procedure to try out different parts so the culprit is revealed. I'll share our findings here, and let them do their tests.

In a lot of these cases, the MSI board is the common denominator.

Honestly, as long as we find out what the issue is, I'm totally fine with an RMA, even if it takes another month to get this PC running. If the motherboard issues are common as you say -- several cases a week here in the forums -- then I'm optimistic they'll address them quietly by sending new parts without fuss, and by improving the tech in future motherboards as well... but who knows.

But yeah, right now it's useless.

Then, I will keep the MemTest ready for action, and we'll see. Thanks for confirming!

But when all other components are proven good or not involved, then you are only left with an RMA of the board, I'm afraid.

I can only hope that's the case and not something more complex pertaining to more than one component. Don't mind the waiting; only the grating feeling I've made a mess of these expensive parts.

I wish I knew some trick to solve this, it's quite annoying not being able to help.

As long as we keep prodding, something eventually comes up! Or so experience has taught me. You seem to seek actively, so as long as you keep it up, I'm sure you'll get an answer sooner or later.

I appreciate the detail you have now provided.

And I appreciate your earlier instructions on what you needed, because thanks to that I decided to not shy away and be as detailed as possible.

You wrote about removing and reinstalling fans, and that is when your computer no longer booted. Something in that process changed. I keep thinking "something is touching that should not be touching" or "something is not touching that should be touching". In your narrative, you had a working system but then made physical changes to your hardware.

Earlier today, it came to me... I know exactly when the PC stopped functioning properly!

I had booted into Windows with the RAM OC-ed at 6000MHz, and I had just opened My Computer... That is when the PC froze. No keyboard or mouse inputs could be registered, or so it seemed from the unmoving image on the screen. I waited a while, but it didn't unfreeze, so I pressed and held the power button until the PC shut down. Once I pressed it again to turn the PC on, it started displaying the EZ Debug LEDs; specifically, our "favourite", infamous DRAM+CPU combo.

Can you start over again? Give it another try before turning it over to a tech? You do not need the GPU, you can use the HDMI port on the backpanel. You do not need any storage devices. Don't need fans, just disconnect the wiring. CPU cooler pressure is up to you. I would even remove the motherboard and put it back just to make sure there is nothing underneath it.

I could try!

Good luck and may the green ezdebug light shine upon you.

Thank you very much!!

This had me thinking of what I had just read about "complete BIOS flush" and may be worth trying in this situation:

I don't have a BIOS Reset button in the back, only a BIOS Flash and CMOS Reset button. Is the person you quoted referring to the CMOS Reset button, by any chance? If so, then, if I understand correctly, the process would be: BIOS Flash finishes, PC remains on, I turn it off from the power button, unplug PSU, hold CMOS Reset button for 10-15 seconds, and then power on the PC, try to enter BIOS, and load the default settings. Am I missing something?

 

[citay]

Currently, the PC remains on, and the EZ Debug LEDs indicate an issue with the DRAM+CPU, as up until now. Should I do anything else from this state? Can I turn it off?

Yeah, you can turn it off. During those five minutes, the Flash BIOS Button LED didn't flash?

 

[ericj159802e9]

I don't have a BIOS Reset button in the back, only a BIOS Flash and CMOS Reset button. Is the person you quoted referring to the CMOS Reset button, by any chance?

That is correct. The BIOS is also being referred to as CMOS. CMOS is the name of the memory technology that is used in the BIOS chip. BIOS and CMOS are used interchangeably: they point to the same thing.

If so, then, if I understand correctly, the process would be: BIOS Flash finishes, PC remains on, I turn it off from the power button, unplug PSU, hold CMOS Reset button for 10-15 seconds, and then power on the PC, try to enter BIOS, and load the default settings. Am I missing something?

That is correct. You are not missing anything.

 

[at4nas]

Yeah, you can turn it off.

Thank you for the confirmation, citay!

During those five minutes, the Flash BIOS Button LED didn't flash?

Do you mean the Flash BIOS Button LED that lights up after a USB drive is inserted and the BIOS Flash button is pressed?

When I was using the old USB drive, I held the BIOS Flash button, and the BIOS Button LED emitted a red, static light, which started flashing as soon as I let the button go. Today, I pressed and held the button, and the BIOS Button LED started flashing immediately, no static state before that.

BIOS and CMOS are used interchangeably: they point to the same thing.

That is correct. You are not missing anything.

Thank you for confirming both things, ericj!

BIOS Flash finishes, PC remains on, I turn it off from the power button, unplug PSU, hold CMOS Reset button for 10-15 seconds, and then power on the PC, try to enter BIOS, and load the default settings

I followed the steps, and hit the same wall of DRAM+CPU LEDs when I powered on the PC.

 

[citay]

You don't have to keep holding the Flash BIOS Button, usually it should work similar to the power button, you just give the start signal and the rest will happen (or not).

Once you pressed the button, it should pretty much happen as you described here:

This time, once I held the BIOS Flash button for several seconds, there was no static light -- it immediately started flashing normally (~1 flash/second), and then it accelerated (~2 flashes/second). Not even 3 seconds pressing the BIOS Flash button had the other parts turned on as well (I thought the PC was supposed to turn on after the BIOS flash had finished?). The whole PC continued running like that for almost exactly 5 minutes. It then turned off, and several seconds later, it turned on again; this time, the red light indicating the BIOS Flash running was no longer on.

First slow flashes a couple times, then around two per second, and that should go on for 5-8 minutes followed by a reboot. So what I was asking was, during those five minutes until the reboot, did you see some (usually red) LED flashing on the board, somewhere around the socket area or the rear I/O?

 

[at4nas]

So what I was asking was, during those five minutes until the reboot, did you see some (usually red) LED flashing on the board, somewhere around the socket area or the rear I/O?

Yes. It happened as you described: slow flashes a couple of times, then two per second, and that went on for 5 minutes. The extra thing that happened was that the other components lit up as well after the couple of slow flashes had ended. I thought it was not supposed to happen until the flashing had ended entirely.

Is this normal behaviour? Should I try to perform the BIOS flash anew?

 

[dvair]

Did it reboot on it's own? If it did then flashing is complete.