The creator of the Linux OS, Linus Torvalds, vented his frustration with AMD's fTPM implementation (h/t Phoronix). Tersely characterizing the AMD fTPM as "crud," the outspoken Finnish developer appeared to suggest the best thing to do with this motherboard flash-based TPM solution would be to disable it, at least as far as using it for hardware random number generation is concerned. Torvalds noted that the CPU RDRAND instruction for the same task doesn't cause stuttering, so it could be used instead.
We have previously covered news of AMD's fTPM (firmware trusted platform module) implementation causing stuttering in Windows apps and games, and the official purported fix/workaround. Please carefully heed AMD's warning about the workaround of switching an active system from fTPM to dTPM. However, it seems like AMD's BIOS patches haven't managed to iron out all the fTPM stuttering behavior on Windows or Linux.
We wrote our last article about AMD fTPM-induced stuttering on Windows in March 2022. However, subsequently launched AMD CPUs and platforms appear to have inherited the same or similar issues. People with the newest AM5 systems in 2023 are still complaining about stuttering, though without further investigation, we can't be sure whether there is some other root cause. System stuttering can be a huge problem when working with media files since it can interrupt video conferencing and infuriate gamers (among other things).
Circling back to the Linux creator's diatribe concerning AMD's fTPM, he began with his conclusion: "Let's just disable the stupid fTPM HWRND thing," was his opening assertion. Various reasons were put forward to this end, but the crux of the issue seems to be that the fixes we have seen and reported upon "apparently didn't turn out to be true."
In a further lengthy comment on the state of the AMD fTPM, Torvalds appeared to throw shade at motherboard BIOS coders before putting forward a key observation about the CPU-based RDRAND vs. the fTPM-based HWRND. "So RDRAND (and RDSEED in particular) can be rather slow, but I think we're talking hundreds of CPU cycles (maybe low thousands)," wrote Torvalds. "Nothing like the stuttering reports we've seen from fTPM."
It would be interesting to hear from readers with AMD CPU-based systems about whether the official fixes or workarounds helped or the extent of any stuttering issues still present (Windows & Linux users).
