The newly discovered vulnerability is Intel only, so it could ultimately mean another possible downgrade in performance for those who have stuck with them after Spectre and Meltdown.
It was said last year at this time that this genre of vulnerability was something barely explored at all. Unfortunately, it looks like the end of the rabbit hole may still be a ways off.
However, I tend to think that these “vulnerabilities” are more theoretical than practical. The information that you are looking for has to be in memory at the instant you go looking for it … and you have to somehow know where it is.
Oh you mean like usernames & passwords, bank data - yeah sure nothing problematic with that at all…
These double quotes are not a good standpoint for security. A theoretical vulnerability can be exploited, it’s just a matter of time and effort.
In fact, the article linked is an extension of a more theoretical attack (row hammering) into a more practical one (to determine the address). I’ll keep more comments until I read the article.
Normally, we assume that processors provide a perfectly safe platform for our applications. If this is not true, things can end up in total chaos. Scale of the original security flaw need not matter.
Intel chips kinda sucks already compared to AMD/Nvidia so I dont even use anything intel-based afaik
I wonder if other brands really have less vulnerabilities or they weren’t explored because the brands are less popular than Intel
The team who discovered this actually tried the same thing on chips from AMD and ARM and failed, so this really is an Intel problem.
My view is that Intel has the most sophisticated processors and thus the most vulnerabilities. Simple systems are usually much more secure ones.
I wonder what are the security risks of all the magic within Ryzens – the perceptron predictors, for example. So far, there were some reports, but they turned out to be pretty fake.