I was taught to put the EXecuted instruction elsewhere, AND as the comment,
thus:
EX R1,MVCEX MVC TARGET(*-*),SOURCE
...
MVCEX MVC TARGET(*-*),SOURCE
The "*-*" is an eyecatcher to remind you what's being ORed with the
register. And that this isn't a "real" instruction otherwise (well, yeah,
it'll move ONE byte, but).
On Wed, May 21, 2014 at 10:34 AM, Robert Ngan <[email protected]> wrote:
> For simple moves (i.e. non-padded) longer than 256 bytes and less than 4K,
> I've started using MVCOS with R00 set to zero. It's said to be more
> expensive than MVC but does anyone know what the "break even" length is
> (approximately)?
> I'll probable continue using it regardless of the performance penalty since
> it's very simple to set up for a variable length move (for lengths longer
> than 4K, I'll stick with MVCL).
>
> I'd also be interested in a comparison between EX/MVC and MVCOS if anyone
> has that information.
>
> Robert Ngan
> CSC Financial Services Group
>
> IBM Mainframe Assembler List <[email protected]> wrote on
> 2014/05/20 17:38:06:
>
> > From: Gerhard Postpischil <[email protected]>
> > To: [email protected]
> > Date: 2014/05/21 02:49
> > Subject: Re: MVCL
> > Sent by: IBM Mainframe Assembler List <[email protected]>
> >
> > On 5/20/2014 11:44 PM, Robin Vowels wrote:
> > > Sounds like overkill.
> > > MVC with EX would be simpler.
> >
> > It depends very much on the requirement. For example, building a longer
> > string from several different sources; MVCL sets the destination address
> > and residual length after completion, allowing for simple chaining of
> > multiple moves. While EX/MVC can accomplish the same thing, they require
> > a little more arithmetic.
> >
> > Gerhard Postpischil
> > Bradford, Vermont
>
--
zMan -- "I've got a mainframe and I'm not afraid to use it"
