> :Intel screwed up their design of hyperthreading with caches, and as a > :result, local users can steal data from one another. > > Intel did? How's that? This cache issue has been a problem before at > different levels. You're stating that it's the CPU's job to determine > scheduling of what threads are running on the HTT enabled CPU. Do you > want another cache for each 'virtual' cpu? Sounds like you might just > want to go the next step and do a true MP system instead of virtual :). > I'd blame the OS scheduler before Intel with regards to this cache issue.
I admit I am not expert on this issue. I merely brought it up to illustrate a point. However, let me ask you this (as I truly don't know): Did Intel advertize to OS makers that they should never allow two processes of different access rights to use the two virtual CPUs at the same time? If it wasn't documented, then it surely was their fault. If it was documented, then it really does cut down on the benefit of the feature. tim _______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/
