I was reluctant to open up the issue that Alan Fargusson has addressed, but
since he has done so . . .

Historically, IBM mainframe applications were executed in AMODE(24) address
spaces containing at most 2**24 = 16,777,216 bytes; and there are in fact
many, far too many, such applications srill in heavy use, all plagued by
what in the jargon are called virtual-storage constraint-relief (VSCR)
problems.

They were succeeded by AMODE(31) applications, which make address spaces of
at most 2**31 = 2,147,483,648 bytes available.  This a 127-fold increase
(2*31/2**24 = 2**7 = 128), but for many applications it too is proving to be
inadequate.

AMODE(64) now makes address spaces of at most 2**64 bytes =
18,446,744,073,709,551,616 bytes available.  This is a qualitatively
different kind of increase, a 1.86-billion-fold one that for the foreseeable
future is likely to provde adequate storage for almost any application.

Moreover, compiled C code can already make very effective use of this space
and the new z/Architecture instructions made available with it.

Thus, while the economic case for converting existing 31-bit applications to
64-bit ones may or may not be persuasive (Often it is not), the case for
developing new ones in anything bit a 64-bit environment is a dubious one.

John Gilmore
SystemCraft LLC

_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963

Reply via email to