New Ryzen Is Running Solid Under Linux, No Compiler Segmentation Fault Issue

[bridgman]

Thanks. I didn't know about the MCE-specific thread. Will go through it and pass along any new information.

[nagisa]

I sent in my chip for RMA at the beginning of this week and still am waiting for some update since the "we will send you a tracking code once replacement is sent" I got on tuesday. While I'm sure it is me being overly impatient (I had hoped to have the replacement by today), I find it dubious that such a delay is necessary.

[cyring]

About CPU temperature, latest it87 module could be force loaded to querry sensors.

[nagisa]

About CPU temperature, latest it87 module could be force loaded to querry sensors.

This chip only applicable to some motherboards and not the others. For other motherboards you may have to force-load a nct6775 module instead.

Ideally AMD would provide somebody (e.g. me or even better -- groeck) with the specsheets with information on how to read the CPU sensors (or implement that themselves even) so we didn't have to rely on a reverse-engineered implementation of a module that was intended for an entirely different version of the chip.

[Funks]

Good luck trying to find 1725 or newer at local stores. I checked our Local Fry's and Central Computers - only old ones. I doubt Build 25+ chips hit the stores yet - unless you buy a thread ripper (which are UA1727+).

I've been logging all the reports I've been getting

https://docs.google.com/spreadsheets...it?usp=sharing

The only way to get a 1725+ chip currently is through RMA, and that's a slow two week process, AMD needs to get cranking on these RMA's. If it's not a new stepping, manufacturing fix, my assumption is that they bolstered up their QA process WEEK25 on.

[Chewi]

I had things running smoothly with my workarounds in place but then I updated to 4.12 and the segfaults came back. I'll give it another shot soon but I may have to bisect. The good news is that I was looking through the commit log and saw a freeze fix that sounds like it'll fix my other issue so I won't have to enable CONFIG_RCU_NOCB_CPU_ALL any more.

[Apache14]

Fwiw I have just gotten a replacement 1700 from AMD (to replace due to MCE issues and segfaults) the new one "seems" fine and was made week 30 this year (UA 1730) and the old broken one was (I think) week 7 (UA1707)

Looks like (from information on AMD forums) anything newer than 1725 looks to be "fixed"

[sdack]

A manufacturing problem is still a CPU bug as it's a problem with the physical CPU. Perhaps you meant it's not an architecture bug?

Not sure where you want to go with this, but I'm guessing you're looking for whom you can blame. So if this is indeed the same stepping then the problem could have been caused by GlobalFoundries, who make the CPUs for AMD. AMD would only be the middleman or messenger for this issue. Not sure if you're going to like this or if you rather just want to keep beating on AMD.

[Funks]

I had things running smoothly with my workarounds in place but then I updated to 4.12 and the segfaults came back. I'll give it another shot soon but I may have to bisect. The good news is that I was looking through the commit log and saw a freeze fix that sounds like it'll fix my other issue so I won't have to enable CONFIG_RCU_NOCB_CPU_ALL any more.

Just RMA it, no need to waste even more time on voodoo workarounds.

[khnazile]

Well, fuck. Getting the CPU RMAed would be really hard here in Russia, and I'm not sure if it would work for OEM CPU.