I wonder why MVCP or MVCS should require millicode.
Is it because the length and key are in registers?
Modern processor designers face a dilemma. Instructions in hardware
(Bob R uses the term "in silicon") can be fast but once the processor is
released they are immutable; you can't change their operation without
replacing the chip at some fantastic expense. Instructions in millicode
are essentially software, loaded when the processor is PORed, so they
can be changed by an EC (Engineering Change) by replacing the millicode
disk. So the silicon instructions tend to be the more basic and most
frequently executed instructions, and others are in millicode.
I am sure this is not the only criteria for the millicode decision, but
it is the most basic.
BTW, for those who don't know, millicode consists of the silicon
instructions from the normal zSeries instruction set, plus some
'millicode-only" silicon instructions, so executing a millicoded
instruction is more or less like executing a subroutine in the
instruction processor.
Bob R has told the story of an earlier generation millicoded machine (I
totally forget which, or which instruction was involved) where a bug was
found in the silicon instruction after GA, so they replaced it with a
millicoded version, but it ran a LOT slower than the silicon version and
impacted performance of certain applications.
--
Bruce A. Black
Senior Software Developer for FDR
Innovation Data Processing 973-890-7300
personal: [EMAIL PROTECTED]
sales info: [EMAIL PROTECTED]
tech support: [EMAIL PROTECTED]
web: www.innovationdp.fdr.com
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
