On Thu, 20 Aug 2009 08:26:45 -0500, McKown, John wrote:

>
>Thanks. I was under the impression that the microcode load in a 
>zAAP and a zIIP was different from the microcode load in a CP. I 
>guess that I was thinking that because there definately is a 
>difference (minor) in the microcode load in an IFL and a CP. That's 
>what stops z/OS from IPL'ing on an IFL (check stop as I recall 
>from "messing around"). Of course, the argument then becomes if 
>a difference in the microcode is a "hardware difference" or not. 
>And if a zAAP and a zIIP had the identical microcode to each 
>other or a CP, how would z/OS know which was which? That 
>was my attempted point, which was poorly stated.

If I understand correctly, there is no microcode on a z10 processor.  There
is millicode, which differs from microcode in this regard - it is
implemented using a superset of z/Architecture instructions.  The processor
can run in a special mode in which it executes the millicode to emulate an
instruction that was not implemented in hardware.  Millicode can also be
used to alter the behavior of an instruction, for example if it is
determined that there is an error in the hardware implementation.

Unfortunately, the IBM journals are no longer available free.  There was an
excellent article about millicode a while ago.  I see now that individual
articles can now be purchased for $1.99.

You might want to read the IBM System z10 Enterprise Class Technical
Introduction, available at
http://www.redbooks.ibm.com/redpieces/abstracts/sg247515.html

In there, it says, "The z10 EC implements 894 instructions, of which 668 are
implemented entirely in hardware."  That is, there is no millicode (or
microcode) to implement those 668 instructions.  The Charles Webb
presentation two years ago on what was then called the z6 processor said the
same thing.

By the way, you might want to be careful with "messing around" with trying
to IPL z/OS on IFL, zIIP or zAAP.  You might find yourself in violation of
the T's & C's of those specialty engines by doing that.

-- 
Tom Marchant

----------------------------------------------------------------------
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

Reply via email to