[prev in list] [next in list] [prev in thread] [next in thread] 

List:       sas-l
Subject:    Re: (MVS) Re: Re: CPU
From:       Ed Gould <edgould () WORLDNET ! ATT ! NET>
Date:       1999-01-29 2:09:16
[Download RAW message or body]

Michal,

I don't have the announcement in front of me but it is similiar to the AMDAHL
 feature of a
few years ago. Of course you had to pay extra (to AMDAHL) to throw the "switch"
 but it
did make your cpu faster. Its not clear to me (anyway)  what the charges will be
 for IBM
to throw the switch, I was not able to read and understand the 40+ page
 announcement.
I am still reading it. This was discussed on another list I am on quite
 lengthily(sp?)

I am also not talking about adding another cpu "on the fly" which is also a new
 feature.

Of course if you have an old AMDAHL then you already have something to wory
 about
:)

Ed


On Thu, 28 Jan 1999 10:45:59 -0500, Michael A. Raithel wrote:

>     Ed Gould posted the following:
>
>
>     >Actually there is a control block that is accessible by
>     >the user (if my memory serves me correctly) that indicates
>     >how much cpu time the task has used. However I would
>     >*STRONGLY* suggest yuou not use it.. Don't forget cpu's
>     >can and do increase speed wise.. what may take 10 seconds
>     >today may take 1 second tommmarow.. whats too much? With
>     >the latest generation of IBM's cpu's they can increase in
>     >spead "dynamicaly".
>     >
>
>     Ed, your memory _DOES_ serve you right, as I pointed out in
>     my posting.  But, I wonder if it is serving you right
>     regarding your warning about CPU speed!  Where on earth did
>     you get information that "the latest generation of IBM's
>     cpu's" ... "can increase in speed 'dynamically'"?
>
>     Are you stating that a batch job that consumes 10 seconds
>     of TCB time on a given mainframe, under OS/390, could
>     consume (say) 1 second when run later on, with no changes
>     in the inputs/outputs/program, and no change in the
>     mainframe--no upgrade to the processors, no upgrade to
>     OS/390?  I don't think so!  Nope!  It would not, should
>     not, could not happen!
>
>     You might be thinking of the recent IBM announcement that
>     new mainframes in the 9672 series can have additional
>     processors turned on dynamically.  In that way, the total
>     MIPS of the mainframe will rise and more work could be done
>     in the machine at one time.  However, the newly enabled
>     processors still run at the same speed as the other
>     processors.  They _DO_NOT_ dynamically increase in speed!
>
>     Ed, my reason for writing this is that I don't want people
>     who may be able to use valuable TCB information to be
>     scared off by an assumption that this information is
>     unstable and therefore unreliable.  I've received some
>     great code (off-line from a talented SAS programmer who I
>     urged to post to the 'L) that uses the PEEK function to get
>     the TCB time. I think that you got some bum information and
>     that you and other SAS-L-ers will find this code very
>     useful.
>
>     Ed, best of luck in all of your endeavors on that most
>     stable and reliable of computing platforms OS/390 (MVS)!
>
>
>     I hope that this suggestion proves helpful now, and in the
>     future!
>
>     Of course, all of these opinions and insights are my own,
>     and do not reflect those of my organization or my
>     associates.
>
>     +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>     Michael A. Raithel
>     "The man who wrote the book on performance" E-mail:
>     maraithel@erols.com
>     Author: Tuning SAS Applications in the MVS Environment
>     +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>     Truth is something you stumble into when you think you are
>     going some place else.  -- Jerry Garcia
>     +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

********************************************************************************
 *************
*
* edgould@worldnet.att.net
*
... The sad thing about Windows bashing is it's all true.
********************************************************************************
 *************

[prev in list] [next in list] [prev in thread] [next in thread]