Outside of the CPUID crap (P cores vs. e-cores have a slightly different CPUID) there’s also a problem somewhere where different cores, including P-cores or the same set of e-cores, Something They work differently from each other, which can cause similar issues (perhaps a timestamp?).

This feature doesn’t exist in any other x86 microarchitecture, so it was common (until now) to use a value like this as a starter for temporary encoding or to link something to a system for things like DRM. Intel could have solved this differently, for example with a new instruction (or “high bit” in the CPUID), the second error is also ultimately a design flaw by Intel, which may have been noticed only later during development that the software Here it depends.

By the way, I think the main title “All Games Solved” is a bit exaggerated, this is just about the list of Denuvo titles that Intel maintains, there are some rare DRM systems that still cause problems but weren’t on this list, most likely due to no Intel information is shared with these parties, and this is in part not indicated anywhere in the public assembly guide: only CPUID differences between the two core types are included, not for example this difference in timestamp/cache coherence/whatever.

[Reactie gewijzigd door NTAuthority op 10 januari 2022 20:38]