Question Major stability issues with high end build ?

[Dillonr93]

Hello all. I have a high end pc that I've had for just about a year now. Ever since it was first built, it has had issues with booting, issues with freeze framing/stuttering and issues with games crashing. I'll list my specs here and then afterwards go into a more detailed breakdown of the issues, and what Ive done so far to try to fix.

Listed below are my specs.
MOBO: Asus Prime Z690-A
CPU: Intel i9 12900 kf
GPU Zotac RTX 3090Ti
Ram: Corsair vengeance 32GB DDR5 6000mhz
Storage: several Samsung 980 pro m.2 drives
Psu: EVGA Supernova 1000w G6
CPU cooler is a Noctua air cooler.

I'll start with the the freeze framing/stuttering. The freeze framing and stuttering happens frequently and it happens on the desktop, while gaming, and even in the bios. Im fairly certain that the cause of this is the same thing thats making my games crash frequently. Im thinking it has something to do with my cpu or ram. Im suspecting those two as the culprits for the boot issues as well. When it comes to booting, my pc will typically power cycle once or twice before it posts. If I mess around with the internals, or the bios or even if there is a power surge, the booting process takes even longer. It will power cycle even more, or hang at the splash screen or will get stuck on automatic repair.

My motherboard has LED indicators for the CPU, DRAM, VGA and for the Boot drive. The LEDs light up briefly as it checks the integrity of the components during boot. If there is an issue with any of those components, the led will stay lit up on the afflicted component, and the pc will not post. Interestingly, every time i have boot issues, the LED is stuck on a different component. It might be stuck on DRAM on one boot, and it may get stuck on VGA or Boot on another power up. I also can not get my RAM to operate on any frequency other than 4000mhz.

So far i have tried many many many different solutions and have not had any success. Currently, I have spent 3 entire days trying to fix these issues, and as of now, I cant get the pc to post at all. I managed to somewhat fix the freeze framing/stuttering issue by disabling Intel speedshift. Doing that completely eliminated the freeze framing, but i still have frequent micro stuttering, particularly noticable while gaming and when in the bios. When looking at things in the bios, I noticed that my RAM wasn't running at full speed. I know 12th gen intel has issues with DDR5 RAM. I've spent a considerable amount of time messing with RAM and XMP settings in the bios. What I found is that I have to have the XMP settings on very specific settings in order for my PC to boot correctly. When the bios is set to default settings, I have boot issues. Default bios sets literally every parameter to "Auto" and it does not actually show me what setting that auto has selected. I can not have the RAM frequency set to anything but 4000mhz.

Of particular concern, XMP only works if the memory VDD and VDDQ voltages are set to 1.9v Im worried that can cause some damage long term. When I have this specific XMP profile set up, the pc boots exactly how it should. Quick and seamless. Things load quickly on the desktop, however I still get stuttering and I have RAM related crashes when gaming. I have messed with all sorts of bios settings, from cpu boot states to RAM voltages. After setting my cpu boot state to turbo, my cpu is now perpetually stuck in turbo mode. Even if I change it back, or default the bios settings, or even reset the cmos, my cpu remains in turbo mode.

Another interesting thing is that Im unable to test my RAM with memtest86. Everytime I run memtest, it gets stuck gathering the system data. Other benchmarking softwares dont show any system issues. Originally I thought I had a psu issue. I had an older one and had just upgraded to the supernova in hopes of resolving the issue. That did nothing. I have reseated the CPU a few times. I know that there is no issues with my GPU. I have switched my RAM to the other slots, and Ive tried booting with just one stick of RAM and have tried that with both sticks. Nothing has solved these issues. Im at my wits end here. Im beginning to believe that either my CPU, Mobo, or Ram is faulty. Im well past being able to RMA any parts and this is my last effort at finding a solution before I just buy a new mobo and processor. Any help is appreciated, this has been a major headache.

 

[Dillonr93]

Some updates after more trouble shooting. I got memtest to run. No errors. Prime95 tests show no errors. Pc is only posting occasionally now. Another odd thing I discovered is a whole bunch of discrepancies between what the bios is showing and what monitoring programs like Cpuid or Asus armory crate are showing. As stated before, my bios has the RAM voltage set to 1.9 and frequency at 4000mhz. CPUID and Memtest show the RAM operating between 4000-4800. Voltages on CPUID shows 1.35v. Im really starting to believe this is an issue with the motherboard.

 

[Karadjgne]

What ram? Exactly.

Never seen an intel system 'power cycle', I have seen plenty that fail post so will either try 2x again with the same settings (3x attempts is default) and then revert back to any prior bios that actually passed post. If there is no prior, it reverts to default settings.

1.9v on anything other than PLL is nuts. VDD & VDDQ should not exceed 1.6v, ever, and generally prefer closer to 1.35v to 1.45v at best.

 

[Dillonr93]

What ram? Exactly.

Never seen an intel system 'power cycle', I have seen plenty that fail post so will either try 2x again with the same settings (3x attempts is default) and then revert back to any prior bios that actually passed post. If there is no prior, it reverts to default settings.

1.9v on anything other than PLL is nuts. VDD & VDDQ should not exceed 1.6v, ever, and generally prefer closer to 1.35v to 1.45v at best.

CORSAIR Vengeance 32GB (2 x 16GB) 288-Pin PC RAM DDR5 6000 (PC5 48000) Desktop Memory Model CMK32GX5M2D6000C36.​

The system truly power cycles, and it does not revert any bios settings upon failure. When looking at hardware monitoring software, It shows that my RAM is not actually receiving 1.9v. It shows It's receiving 1.35v. If I let the Bios optimize the RAM overclock, it sets the voltage to 1.90 and it has even tried to set it to over 2v at times.

 

[Karadjgne]

Update bios?
During post, (power on self test), if the cpu detects any voltage errors or communication errors with essential components, it reboots by itself. It's not a power cycle as such but an assumption that a driver didn't read right or is locked up or conflicting etc. It'll reboot 2x more times because default attempts is 3. It only reverts to a prior bios setting if it cannot reboot cleanly.

It's also possible it's not an actual bios setting but a cmos setting, but if there's that much discrepancy, it's possible either the cpu or the motherboard is glitched and not reading voltages correctly.

 

[Dillonr93]

Update bios?

I haven't done that yet but I was thinking that might be the fix. I had to go out of town for work today and wont be back untill friday. I'll give it a shot on Friday and report back with my findings.

 

[Misgar]

I booted up an old computer (2600X + 3,000MHz RAM) yesterday with what I believe was MemTest86 (not MemTest86+) and it took nearly 5 minutes before the test started. Afterwards, the test ran fine. Try the alternative version to the one that failed to start up.

When I installed Armoury Crate on a 7950X rig in December 2022 I had instability problems, but I cannot recall exactly what happened until I check my notes. I'm not using Armoury Crate now.

I've become accustomed to long memory training times on my 3800X DDR4 and 7950X DDR5 rigs when I've made changes to the BIOS, but after training, subsequent booting is usually faster.

Your problems do sound memory related. Is your specific memory in the motherboard QVL? I'm not paranoid about sticking to the QVL but sometimes it might be wise to avoid RAM that's not in the list.

DDR5 prices have been dropping and it might be worth trying a different memory kit. If you have a sympathetic supplier, you should be able to RMA the new kit if it doesn't make any difference.

The video editing apps I use don't benefit appreciably from faster memory so I saved money and bought a 2x32GB (64GB) 4800MHz kit last year. If you're not too worried about 6,000MHz XMP, consider slower RAM. It might be more stable.

 

[Dillonr93]

I booted up an old computer (2600X + 3,000MHz RAM) yesterday with what I believe was MemTest86 (not MemTest86+) and it took nearly 5 minutes before the test started. Afterwards, the test ran fine. Try the alternative version to the one that failed to start up.

When I installed Armoury Crate on a 7950X rig in December 2022 I had instability problems, but I cannot recall exactly what happened until I check my notes. I'm not using Armoury Crate now.

I've become accustomed to long memory training times on my 3800X DDR4 and 7950X DDR5 rigs when I've made changes to the BIOS, but after training, subsequent booting is usually faster.

Your problems do sound memory related. Is your specific memory in the motherboard QVL? I'm not paranoid about sticking to the QVL but sometimes it might be wise to avoid RAM that's not in the list.

DDR5 prices have been dropping and it might be worth trying a different memory kit. If you have a sympathetic supplier, you should be able to RMA the new kit if it doesn't make any difference.

The video editing apps I use don't benefit appreciably from faster memory so I saved money and bought a 2x32GB (64GB) 4800MHz kit last year. If you're not too worried about 6,000MHz XMP, consider slower RAM. It might be more stable.

I strongly believe this is a memory issue. I just checked the QVL and my specific RAM is indeed on the list. Im hoping that a bios update will fix this. The newest bios update adds more RAM compatibility and adds support for Intel 13th gen.

 

[Karadjgne]

Ram is made up of chips stuck to a pcb. There's only a dozen or so actual ram OEMs, like Samsung, Micron, Nanya, SkHynix etc. They make the chips for everyone. So if you see a Corsair LPX 3200, it's the same SkHynix chips as a good chunk of Gskill 3200 linup, and Patriot, and Kingston, Team, and various others.

QVL is qualified vendor list, not qualified ram list, it's backwards to common perception. It simply means that the mobo Vendor used a kit and it worked, saying that the Vendor's mobo functioned using whatever specs the ram had. It's qualifying the mobo, not the ram.

Whether any particular kit actually functions on a mobo is pot-luck, because each individual kit comes from a different silicon batch, so just because the similar kit you may own was tested on the prototype board, does not guarantee that the actual kit you own works on the actual mobo you own.

It's like Dodge says my wife's Challenger does 0-60 in 5.3 seconds, I personally have never been able to duplicate that because it's on a different road, different blacktop heat, different gas, different oil, different weather, different height above sea level, different driver and habits, different car (same model) etc. I get 6.0 seconds best time.

It could simply be that whatever the Secondary and Tertiary timings are are not liked by the cpu, so tweaking some of those may or may not have positive affects.

If you use Typhoon Burner to do a dive into the ram, it'll tell you the OEM, the revision or die etc, then look up OC videos for that particular ram to see what changes should or should not be changed, what values can offer better stability etc, just don't immediately try to copy videos the results, but tailor yours down in stages.

If you do find a level that works good, save the bios as a Profile, take a screenshot, don't lose those numbers, you'll maybe need to plug them back in one day. There's 50 or so values for Secondary and Tertiary timings, don't bank on trying to remember them all....

 

[Dillonr93]

Update bios?
During post, (power on self test), if the cpu detects any voltage errors or communication errors with essential components, it reboots by itself. It's not a power cycle as such but an assumption that a driver didn't read right or is locked up or conflicting etc. It'll reboot 2x more times because default attempts is 3. It only reverts to a prior bios setting if it cannot reboot cleanly.

It's also possible it's not an actual bios setting but a cmos setting, but if there's that much discrepancy, it's possible either the cpu or the motherboard is glitched and not reading voltages correctly.

So today I did the Bios update. It seems to have fixed the major issues so far. Im able to boot without issue at default settings now. My RAM is now running at proper speeds. An interesting thing I noticed was a difference in the memory controller frequency. Before the update, it was operating at around 900mhz. Post update, its now operating at 1200 MHZ. Im still getting some minor stuttering periodically in games, but im thinking that might have something to do with a driver or a program or some type of background task that windows is running.

 

[Dillonr93]

Another update. I can now utilize the full capability of my RAM. Got the xmp set up and stable, running it at the full 6000mhz. No more boot issues at all. I still am having micro stuttering though. That is no longer occuring within the bios, just when doing anything in windows.

 

[Dillonr93]

Final update. I can now utilize the full capability of my RAM. Got the xmp set up and stable, running it at the full 6000mhz. No more boot issues at all. I have eliminated the microstuttering by uninstalling a few programs and by disabling some settings in windows. Everything is working as it should now and very smoothly. Thanks for the help everyone!

 

[Dillonr93]

Hi all just providing an update. I want to start by saying that I never resolved the stutter issue. Every time I thought I fixed it, the stutter would eventually return . I decided for a while that I should just stop wasting everyone's time and just learn to live with it. I still continued to trouble shoot all this time I've been away. After constant troubleshooting for the past few months, I still haven't fixed this issue. None of the ideas posted here have helped, but they were worth a shot, so I appreciate the suggestions. What I have learned is that this issue is most likely hardware related, and it absolutely has to do with the way that the MOBO, CPU and RAM are interacting with each other. I have experimented a lot with overclocking and underclocking/volting. Every over/underclock had an effect on the stuttering, making it slightly better or worse. I can mitigate the stuttering but I can not eliminate it no matter what I try. I tried another new RAM kit. Its slower than the kit I initially bought, but when I installed it, the stuttering stopped. There was no more stuttering until I tried messing with xmp. After messing with xmp, the stutter started again and I couldn't get rid of it. I switched back to my old RAM kit which is faster. Interestingly, the RAM kit im using overclocks incredibly well. The stuttering issue is most mitigated when I have an extreme RAM overclock. I am once again asking for some more ideas here. This issue has persisted for almost 2 years now and has driven me insane. If I can't fix it soon, I'll have to just build a new rig, which I really don't want to have to do.