AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |
Back to Blog
![]() Most likely, AMD CPUs are unaffected for the same reason ARM designs which precede the A75: they just took a slightly different approach to memory speculation.Ĭlass action lawsuits aren't the way to do this - strong consumer law is. It would have been the (a) the correct thing to do (b) hurt intel. ![]() ![]() Hell, if AMD had a clue about this risk, they would have exposed it years ago. There is zero evidence to support that AMD or anyone else had a clue that memory speculation across privilege levels could be used to feed into a particularly efficient micro-architectural side channel and adopted a speculative execution strategy to minimize it. I wonder if the prefetchers can't be exploited too. Simultaneous multi-threading (aka Hyperthreading) is another, by the way. The very existence of caches and branch predictors (as shown by Spectre) is one. Second, all high performance CPUs have many forms of speculation, which have the potential for covert side channels. Most of the world was f**king unaware of this attack vector until it came out. The fact that Apple runs a closed system and that the ARM Cortex-A75 hasn't shipped yet are irrelevant to this: both teams made similar choice as intel, not because Meltdown is irrelevant but because they were obviously unaware of this attack vector. I just showed you that there are at least two other design teams that took a similar decision: Apple and ARM. You were asserting that intel's CPU designers took an exceptional decision that favoured performance over security. I was not addressing the potential impact of Meltdown in intel vs Apple vs ARM on users. ![]()
0 Comments
Read More
Leave a Reply. |