Talk:CSC/ECE 506 Spring 2014/7b ks

Changes begin with the section entitled, "TLB coherence problem in multiprocessing," nicely dovetailing with one of the other rewrites, which only modified the Background section. The new prose is generally good, but there are several typos, which the next revision could fix.

The former page referred to implementations from the 1980s and 1990s, primarily. You simply removed those references. It would have been better to have updated them, by mentioning the most recent system to use them.

Lazy devaluation could have been explained more completely ... as could some of the other approaches in that section. "This approach works well unless processes run in parallel in the same address space" is a sentence in need of further elucidation.