Title: Revisiting "Multiprocessors Should Support Simple Memory
Consistency Models"

Mark D. Hill
Univ. of Wisconsin-Madison

Abstract:

Five years ago I published "Multiprocessors Should Support Simple
Memory Consistency Models" [IEEE Computer, August 1998]. It discussed
the memory consistency model interface between low-level software
and hardware (as a opposed to a high-level language (HLL) interface).
It argued that the low-level interface should be sequential consistency
(or possibly processor consistency), regardless of the HLL interface.

In this talk, I will revisit my arguments five years on.