On Saturday, 12/13/2003 at 06:50 EST, Paul Hanrahan
<[EMAIL PROTECTED]> wrote:
> By themsleves the cycle times might not be of interest. The cycle times
in
> conjunction with explanation of fetch operations and how instructions
and
> their addresses are really resolved would probably be very instructive.

Instructive from an academic perspective, maybe, but not from a business
point of view since how it "really" works can (and does) change from
generation to generation.  The reason the architecture has survived for so
long is because its Prime Directive [apologies to G.L.] is to very
specifically hide the underlying implementation from the programs.

> Not complex examples but a few that illustrate why a compare and swap is
> more reliable in a multi processor environment and an MVCL isn't.

Paul, your wording suggests that MVCL isn't reliable under some
conditions.  The word "reliable" in this context must mean "conformance to
specification".  I.e. you are saying that sometimes MVCL doesn't do what
the book says it does.  Details, please.  (Or I've misunderstood what you
were trying to say!)

Alan Altmark
Sr. Software Engineer
IBM z/VM Development

Reply via email to