Buggy AMD CPUs

[Hans Åberg]

> On 6 Nov 2017, at 22:20, Torbjörn Granlund <tg at gmplib.org> wrote:
> 
> Later, it turned out that the Ryzen CPU was buggy.  AMD decided to NOT
> recall broken CPUs NOR make a full and unambiguous disclosure about the
> issue.  They offered to replace CPUs on a case-by-case basis, while
> inventing the newspeak term "performance marginality issue" for the bug.
> But I decided it was not worth the time to go through AMDs tribulations.
> I however turned on the GMP Ryzen system but made the GMP test system
> aware of the broken hardware (this makes it retry everything thrice).
> 
> Today, I decided to purchase a new Ryzen and simply bin the old one,
> again to save time.  But to my utter surprise, AMD still sells the old
> broken CPUs and I was unlucky enough to get one of those.

Some issues are discussed here:

https://en.wikipedia.org/wiki/Ryzen#Known_issues
https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response
https://www.extremetech.com/computing/254750-amd-replaces-ryzen-cpus-users-affected-rare-linux-bug