I've had this random graphics stutter that would never ever show up in any benchmark at all, but weirder than anything else was that it also wouldn't show up in recordings made on that machine.<p>To save you the story of weeks-long debugging efforts - the problem turned out to the monitor. It was a Dell Ultrasharp 25", and (as I now know) they all had an issue where the monitor would freeze roughly every 20 seconds for a frame, because the monitor was advertising the refresh rate to windows as 59.95Hz but it was actually refreshing at 60Hz internally - the timing would get out of sync pretty much exactly every 20 seconds, causing the monitor to "pause" and freeze - which looked like stutter when playing games. The solution was using some kind of program(forgot the name now) which basically overrode the refresh information being sent from the monitor and forced windows to refresh at exactly 60Hz and no less - then it worked fine. But of course if you used it with a PS4 then you didn't have any way to do that and just had to suffer this stutter every 20 seconds.
It looks like with a series of changes they were able to mitigate the issue <i>almost</i> completely, but it's still there and the motherboard is obviously deficient in some way.<p>They mentioned they put the ram in 2 years ago and they've had this issue for 2 months. I suspect maybe it started when their OS's started using their TPM's but it's possible their board has just broken randomly. Likely on a hardware level.<p>I do see value in diagnosing the issue and making a post about mitigation's. I love people who do this so others with the issue can find it and get their boards refunded/RMA'd before their warranties expire. But once the tpm/ftpm was identified as being the issue and it was hardware/firmware related, they need to get a new board, not more mitigations (hopefully by getting a refund/RMA so the board manufacturer doesn't get a pass on this). I don't see how someone working in the tech space (we are VERY privileged salary wise) would tolerate having obviously faulty electronics.
> The fTPM clearly affects the system stability, even under an OS where it’s not in use.<p>This is probably a code that works in SMM mode, and it has a priority even over ring 0 code. SMM is one of the worst CPU features, it is basically a backdoor in your system that can do anything and you cannot even see it. Even if you run Linux you have no control over SMM code, it can hang your system and you cannot even debug it.<p>Also this shows a weakness of an "open" PC model where poorly tested hardware and low quality firmware from different vendors are assembled together. I cannot imagine this happening on an iPhone for example.
I knew this was going to be due to the AMD fTPM stuttering. I've had it on my machine since I switched back to AMD in 2019. The BIOS updates haven't helped, and the discrete TPM I tried just made the system unbootable. I'm just used to it now.<p>Honestly, as much as AMD's processors are often great value for money, I have never had a completely stable glitch-free AMD system.
This is a known issue with a fix that was reported a while back:<p><a href="https://www.tomshardware.com/news/amd-issues-fix-and-workaround-for-ftpm-stuttering-issues" rel="nofollow">https://www.tomshardware.com/news/amd-issues-fix-and-workaro...</a>
In the past, I'd buy Intel + Nvidia to hope to avoid these sorts of weird troubleshooting cases. This occurred because I had a first-gen AMD Athlon and ran Win2k, where support for Thunderbird CPUs/mobos was not complete, resulting in random crashes that required a mix of BIOS, Windows, and driver updates to completely resolve.<p>Nowadays, I have no idea if that advice (Intel + Nvidia) is still sound. There are way more AMD users than before, and computers are more complicated than they were 20 years ago.
I had issues once with my gaming setup (Arch Linux, Ryzen 5 3600, rtx 2060 6gb, MSI b350m gaming pro, 16gb ddr4 1600).<p>The only thing I did was to set the governor from powersave, which came by default configured, to ondemand. Graphic stutters immediately went away and performance got close to that on windows.<p>I suggest anyone having performance issues under Linux to check their governor.
This is the kind of thing that makes me chuckle every time I see a comment like ‘just build your own PC - they work better and save money’.<p>Yeah, both can be true, but when it goes wrong, it can be very difficult and expensive to troubleshoot.
I’ve had an audio crackle that happens only sometimes on my PC for years now.<p>Windows 10, Z170-A motherboard with driver for built in audio<p>Tried different cables, inputs (optical vs aux etc) installing different drivers, old drivers, new drivers, nothing fixes it<p>Super annoying when I want to listen to spotify or YouTube or anything
I was hoping to read how he debugged his system and found the rout cause. Instead I just get some ramblings from a guy who just tried different things from stuff he read somewhere.
Wow this guy is really selling the AMD+Linux experience. Improved from 2 crashes per hour to once per 4 hours? Nobody I know would tolerate either of those.
Reminds me of the days when you had to optimize your IRQ's<p>Big reason why lot of audio engineers use an external audio box - latency is super noticeable with audio and motherboards are rarely going to cut it
I am scared of upgrading from my i7-6700 system because it is so rock solid - it never crashes and just runs for days without weirdness or Windows 10 imploding on itself.
The random crashing at low usage sounds very similar to an issue I had with my previous 5950x. There seems to be a real problem with 59{00,50}x’s where you either need to disable PBO (I think that’s the name) or bump voltages.
After finding out it was a silicon lottery thing I sent my cpu right back to Amazon and the replacement has been perfectly fine from first boot (outside of figuring out the upper memory clock for 4 dimms)
There was a twitter thread recently about how Mac people never built their own computers so don’t know what things can be like. But I have built my own computers and I <i>hated</i> dealing with issues like this. There’s a reason I don’t build my own computers anymore.
I ended up having extremely weird stuttering issues where the mouse would suddenly take several seconds to move across the screen on a high-end 64 core threadripper machine. Turning off mouse trails in Windows completely solved the problem.
>The mainboard reverts to fTPM if the hardware TPM is removed. It doesn’t have a way to disable the TPM.<p>Can anyone verify this? Sounds fucked up.
> so I quickly reinstalled<p>Dude needs to stop quickly reinstalling things. Instead they need to slowly work through what is going on in the operating system that could be causing the observed issue. What does the Windows system information program say when it is run as an administrator? Is secure boot enabled? Is the pcr7 configuration on "binding possible"? What about kernel dma protection or virtualization based security?