Interestingly, Peter Relson  wrote:

<begin extract>
By the way, despite 50 years of history (or over 30 since BASR became
available), I'm told that BASR is marginally better to use than BALR.
<end extract>

and this is what I find when, after each new z/OS release, I make my
own necessarily crude timing studies.  (Instruction timings and, not a
little surprisingly to me, not just the millicode-implemented ones,
are moving targets on the same platform.)

I have long wished for something better and, in particular, for
something better from IBM.  I think I understand (and I can readily
sympathize with) its apparent reluctance to provide absolute timings,
but a set of index numbers that compared instruction-execution times
in some standard testbed for each release of z/OS (or at some other,
perhaps better canonical interval) would be helpful to those of us
write assembly-language routines.

Currently, there are urban legends, many of them fantastical or long
obsolete, but almost no hard facts available in this area.

John Gilmore, Ashland, MA 01721 - USA

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to