You are making excuses for AMD and you're spewing nonsense conjecture. Here are the facts:ChubChub wrote: ↑ KPTI was enabled on Ryzen chips for some reason; essentially that alone makes all of those benchmarks useless (for AMD). KPTI is not needed for AMDs chips (not susceptible to Meltdown), so the decision to enable it is purely to exaggerate the performance hit (KPTI is basically what is screwing Intel's chips). As well, these are beta changes to the compiler and kernel; considerable optimizations are likely in the pipe (and probably will not make it into kernel 4.15), especially once the microcode updates (from both Intel, and AMD) expose new features to mitigate this performance impact.
I am not making ANY excuses for Ryzen; were you reading what I wrote? I was simply defending the fact that the lawsuit is largely based on false information; class action lawsuits are very lucrative for lawyers, so they just need to convince someone with deep-ish pockets that they have a case, and lawyers win regardless.
If I get a patch that kills performance in a way that actually impacts me, I'll be pretty pissed; I bought a workstation CPU with a certain level of performance (this machine actually generates cash for me by way of hosting various specialized VMs), and I expect it to stay there, or possibly get better (which it conveniently has over the last 8 months). Unlike you, I don't blindly love one company, and hate the other, regardless of performance in my use-case. However, as it sits, Meltdown impacts me 0% after the MS patch (which I was happy to see; Microsoft originally was not going to exclude the AMD CPUs from performance-sapping changes), and the Spectre updates are looking like it will largely not impact me either (that one benchmark where the GCC compiler had a large marked positive impact on performance tells me there is a lot of opportunity for restoring some of that precious I/O).
You really should not blindly look at benchmarks without paying attention to the methodology (benchmarking an Intel system with an SSD versus an AMD system with an NVMe drive as an example of KPTI-level unfairness); when someone pulls obvious shenanigans, you should realise they're probably doing it for a reason, and expect that reason is nefarious.
- AMD said that their chips are completely unaffected by Meltdown and Spectre (not true)
- AMD said there would be no performance degradation on their chips (not true)
How is that so hard for you to understand? It was even their CEO saying this. And you're trying to pretend that they weren't being deceptive? Whether it affected their stock price or not, it was incredibly dishonest.
Now you're trying to discredit the benchmarks. Straight out of the playbook. I can't wait for the dust to settle on this one. I can guarantee you that the AMD chips are going to take a performance hit due to Spectre. Like I said, it is the most serious bug in terms of performance impact. And that information comes straight from Microsoft. You can argue with that all you want.