As always it's the usual, PR wipes the problem away. They own much of the media with huge ad buys for major tech sites. When AMD has a minor bug they have those sites talk it up like a monumental issue, when they have a major problem the sites talk it down.
They had that chipset issue with sata, then had like two generations where virtualisation didn't work as promised while they still sold it and advertised it as a fully working feature, they had a big usb security issue, this. Meanwhile AMD had one major bug with Phenom which they both fixed and announced before launch, let all reviews come out with the fix in place which reduced performance and in that case if you ran without the fix.... you'd almost never have the problem yet AMD got destroyed over that. You can't win now, when you have money you get to control the message.
With EPYC release AMD couldn't publish proper results using the GCC compiler because specint contract you to only publish Intel compiler results.
Look at the compubench results thing going on now, using a virtual machine people are spoofing an Intel ID on the CPU and getting vastly higher results because the compiler thinks it's an Intel CPU and optimises. Intel does this absolutely everywhere, buys up or pays heavily into well used benchmarks, tie them to Intel compiler, disable optimisation for AMD and proclaim way higher wins in benchmarks than they get in real world performance in many many applications. Until AMD has billions in cash on hand to fight back and fight as dirty as Intel then problems for Intel will get brushed under the rug and downplayed by the media.
