OT - What is the proper term for "K" notation?

[Charles Mills]

This question has nothing to do with mainframes (other than that I am trying
to name an option for a mainframe program) but I know there are some
ultra-precise word jockeys here.

What is the correct term for K, M or G type notation? If I had integers 1234
and 4560000, what would you call it if they were formatted as "1.234K" and
"4.56M"? OTOH, what would be the contrasting term for "normal" notation:
"1234" and "4560000"?

What would you call a program option to do things one way or the other? I
want ______ notation output or ______ notation output.

It's effectively a kind of floating point notation, but is there a more
precise term?

Thanks all,
Charles

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to list...@listserv.ua.edu with the message: INFO IBM-MAIN

[John McKown]

K is ISO for Kilo meaning times 1000.
Ki is ISO for Kibi meaning times 1024

ref:
http://en.wikipedia.org/wiki/Binary_prefix
http://en.wikipedia.org/wiki/ISO/IEC_80000
http://www.nist.gov/pml/pubs/sp330/index.cfm
http://www.nist.gov/pml/pubs/sp811/index.cfm

--
This is a test of the Emergency Broadcast System. If this had been an
actual emergency, do you really think we'd stick around to tell you?

Maranatha! <><
John McKown

[Elardus Engelbrecht]

Charles Mills wrote:

>...ultra-precise word jockeys here.

...have already discussed 1001 times on IBM-MAIN and posted/refered in IBM-MAIN this:

http://en.wikipedia.org/wiki/Binary_prefix#IEC_standard_prefixes

There you will learn about kibi and friends.

Enjoy. ;-)

Groete / Greetings
Elardus Engelbrecht

[Don Williams]

Perhaps http://en.wikipedia.org/wiki/Binary_prefix will answer your
question.

[Charles Mills]

No, no one is answering the question I tried to ask. Sorry if I was unclear.

I am NOT asking "what is the difference between kilo and kibi?" or "is it right to refer to 1024 as 1K?" or anything like that.

I am asking what you CALL that KIND of notation.

If my program outputs numbers as 1234 and 4560000 but your program outputs the same values as 1.234K and 4.56M, what would you call the *format* that your program uses? Your program outputs numbers in ______ notation. Mine OTOH outputs numbers in _____ notation.

Perhaps "powers of 1000 notation." Any term more compact than that that could be used as a control statement option?

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Elardus Engelbrecht
Sent: Thursday, May 02, 2013 12:21 PM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

Charles Mills wrote:

>...ultra-precise word jockeys here.

...have already discussed 1001 times on IBM-MAIN and posted/refered in IBM-MAIN this:

http://en.wikipedia.org/wiki/Binary_prefix#IEC_standard_prefixes

There you will learn about kibi and friends.

[Charles Mills]

Or phrasing the question differently:

consider the integer 45600000. It may be expressed as

4.56 x 10**7 or 4.56E7 in scientific notation; or as

45.6 x 10**6 in engineering notation; or as

45.6M in _______ notation; or as

45600000 in _________ notation.

Can anyone fill in those blanks?

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Charles Mills
Sent: Thursday, May 02, 2013 1:24 PM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

No, no one is answering the question I tried to ask. Sorry if I was unclear.

I am NOT asking "what is the difference between kilo and kibi?" or "is it right to refer to 1024 as 1K?" or anything like that.

I am asking what you CALL that KIND of notation.

If my program outputs numbers as 1234 and 4560000 but your program outputs the same values as 1.234K and 4.56M, what would you call the *format* that your program uses? Your program outputs numbers in ______ notation. Mine OTOH outputs numbers in _____ notation.

Perhaps "powers of 1000 notation." Any term more compact than that that could be used as a control statement option?

[DASDBILL2]

More than you might really want to know; i.e., it is not yet officially defined, standardized, or enforced.



Perhaps the term is whatever we use to describe the following notation:

65 mph (miles per hour)

80 gal. (gallons)

5 kw (kilowatts)

88 μ  (microns)



Maybe these are all "abbreviated" notations.



Or Google/Wikipediaize for "International System of Units."  I suspect that here in the USA we can call it anything we want, since Wikipedia states "Only Burma , Liberia and the United States have not adopted SI units as their official system of weights and measures", where SI means " Le Système international d'unités ".



The use of "k" and "K" in the computer field is discussed here:  http://en.wikipedia.org/wiki/Binary_prefix


Bill Fairchild
Franklin, TN

[tbillette]

I would say:
I want decimal prefix notation output or binary prefix notation output.

Thierry

[John Gilmore]

See the National Institute of Standards and Technology (NIST) website:

http://physics.nist.gov/cuu/Units/binary.html

They are called binary and decimal prefixes (sic). The more commonly
used, (10^3)^n, notation is formally called SI decimal-prefix
notation. The binary, (2^3)^n, notation is called IEC binary
prefix-notation. IEC will shortly become SI too.

Strictly speaking nude values like 1.234K are not well formed. They
specify a magnitude but not a unit, kilobyte or kibibyte, kilocalorie
or kibicalorie, kilogram or kibigram, kilometer or kibimeter, etc.,
etc.

John Gilmore, Ashland, MA 01721 - USA

[Don Williams]

K, Ki, are scaling factors, therefore I would call that type of formatting scaled format. I expect most people would understand what I meant.

[J R]

ITYM: (2^10)^n, notation =
=
> Date: Thu, 2 May 2013 14:31:45 -0400
> From: jwgl...@GMAIL.COM

> Subject: Re: OT - What is the proper term for "K" notation?

> To: IBM-...@LISTSERV.UA.EDU

[Tony Harminc]

On 2 May 2013 14:31, John Gilmore <jwgl...@gmail.com> wrote:

> Strictly speaking nude values like 1.234K are not well formed. They
> specify a magnitude but not a unit, kilobyte or kibibyte, kilocalorie
> or kibicalorie, kilogram or kibigram, kilometer or kibimeter, etc.,
> etc.

To the question 'If I had integers 1234 and 4560000, what would you
call it if they were formatted as "1.234K" and "4.56M"?', I was about
to say "broken" or "wrong". But "nude" expresses the notation and its
problem perfectly, and I recommend it with only a small :-) needed.

To 'OTOH, what would be the contrasting term for "normal" notation:
"1234" and "4560000"?', I would say simply "decimal", though that of
course runs the risk of being taken as being in opposition to binary
or hexadecimal (or any number of other less likely bases).

There are two other aspects that may bear thinking about:

The letters K and M are not universally understood to be either of
(10**3 and 10**6) or (2**10 and 2**20). In particular, financial
people and executive types to this day often use M to mean thousand,
and use MM for million. Doubtless these derive from Roman numerals
(though MM would be 2000), but like the Romans they are not quite dead
yet, and if it really matters you should probably avoid them.*

If you want names for your notations, you should consider also
precision and rounding. Perhaps your program will just do what it
perceives to be the sensible thing, but once you give users the choice
by some keyword of 1.234K vs 1234, when should they expect 1.2340K vs
1.23K or even 1K?

Tony H.

* There are actually three overlapping and incompatible schemes used
for money: that used by accountants and other "money men", that used
by engineers and other technical people, and that used by tabloid
newspapers in headlines:

SI Financial Tabloid Multiplier

k (kilo) M (Roman) G (grand) 10**3
M (mega) MM (Roman) M (million) 10**6
G (giga) MMM (rare) B (billion) 10**9
T (tera) - Tr (trillion) 10**12

[Joel C. Ewing]

I get it. The question is not about what scaling factor is being
specified, or even the name of a specific measurement units designation,
but a more general-level question of what kind of name would one give to
the concept of a scaled number representation like "23.5K" or "23.6M",
if for example you were writing a computer output routine to print a
number in that format, rather than printing it as an integer format, or
a fixed point value format, or floating point value with exponent format.

Perhaps you could just call it a "scaled integer value" where "M" is an
indication of scaling factor, but I can't recall ever hearing someone
attempt to give such an external notation a formal name. The usage of
"M" in a context like "MB" is as a prefix, and the standards of which I
am aware only formally define its usage when combined with a unit of
measurement, not stand-alone. The latter format (45600000) I would
simply call "integer" or possibly "unscaled integer" if the other is
called "scaled". The term "scaled integer" does appear in the context
of some languages like COBOL, but it applies to the implied 10**n
multiplier in an internal data representation, not to scaling specific
to input or output of values.

One can certainly find examples of applications, particularly in the
interactive world, where output values are dynamically scaled so as to
display the most significant digits while still using a limited number
of characters as the values get increasingly large:
as in displaying "900 B" as "900 B", but "212,123 B" as "212.1 KB",
"616,212,123 B" as "616.2MB", etc.. It would make sense to be able to
generalize such a numeric format in the absence of a specific
measurement unit, but I'm not sure what I would call it other than
"dynamically scaled integer". There surely must a programming language
somewhere with direct support for such an output format. I'm just not
personally aware of any.
Joel C. Ewing

On 05/02/2013 12:34 PM, Charles Mills wrote:
> Or phrasing the question differently:
>
> consider the integer 45600000. It may be expressed as
>
> 4.56 x 10**7 or 4.56E7 in scientific notation; or as
>
> 45.6 x 10**6 in engineering notation; or as
>
> 45.6M in _______ notation; or as
>
> 45600000 in _________ notation.
>
> Can anyone fill in those blanks?
>
> Charles
>
> -----Original Message-----
> From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Charles Mills
> Sent: Thursday, May 02, 2013 1:24 PM
> To: IBM-...@LISTSERV.UA.EDU
> Subject: Re: OT - What is the proper term for "K" notation?
>
> No, no one is answering the question I tried to ask. Sorry if I was unclear.
>
> I am NOT asking "what is the difference between kilo and kibi?" or "is it right to refer to 1024 as 1K?" or anything like that.
>
> I am asking what you CALL that KIND of notation.
>
> If my program outputs numbers as 1234 and 4560000 but your program outputs the same values as 1.234K and 4.56M, what would you call the *format* that your program uses? Your program outputs numbers in ______ notation. Mine OTOH outputs numbers in _____ notation.
>
> Perhaps "powers of 1000 notation." Any term more compact than that that could be used as a control statement option?
>


--

Joel C. Ewing, Bentonville, AR jce...@acm.org

[Charles Mills]

You do get it! <g> Your second sentence is a perfect exposition of what I was trying to ask. Your last paragraph is a perfect exposition of the problem I am solving with the "K" notation.

Thanks all, especially JG.

"Scaled" seems to be pretty good. Not sure what the opposite is? "NoScaled"? "Unscaled"?

INTFMT(SCALED|NOSCALED)

Decimal does not really cut it because it's base ten in any event, and a decimal point is absent in any event. Binary would be confusing, I think.

I hear the people complaining about "nudity" (ahem) but the units are already specified. Giving an example, devoid here of all context

BytesIn = 25.7K, BytesOut = 286.4M

Yeah, I suppose it might say just plain In = 25.7KB, but, as we say, "the program doesn't work that way." (Also, due to other constraints it MUST appear in a "string" format message, not in tabular form like most mainframe reports.)

And yes, I am doing it "with my own code" -- there is no built-in support in the language I am using (C++).

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Joel C. Ewing
Sent: Thursday, May 02, 2013 3:30 PM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

I get it. The question is not about what scaling factor is being specified, or even the name of a specific measurement units designation, but a more general-level question of what kind of name would one give to the concept of a scaled number representation like "23.5K" or "23.6M", if for example you were writing a computer output routine to print a number in that format, rather than printing it as an integer format, or a fixed point value format, or floating point value with exponent format.

Perhaps you could just call it a "scaled integer value" where "M" is an indication of scaling factor, but I can't recall ever hearing someone attempt to give such an external notation a formal name. The usage of "M" in a context like "MB" is as a prefix, and the standards of which I am aware only formally define its usage when combined with a unit of measurement, not stand-alone. The latter format (45600000) I would simply call "integer" or possibly "unscaled integer" if the other is
called "scaled". The term "scaled integer" does appear in the context
of some languages like COBOL, but it applies to the implied 10**n multiplier in an internal data representation, not to scaling specific to input or output of values.

One can certainly find examples of applications, particularly in the interactive world, where output values are dynamically scaled so as to display the most significant digits while still using a limited number of characters as the values get increasingly large:
as in displaying "900 B" as "900 B", but "212,123 B" as "212.1 KB",
"616,212,123 B" as "616.2MB", etc.. It would make sense to be able to generalize such a numeric format in the absence of a specific measurement unit, but I'm not sure what I would call it other than
"dynamically scaled integer". There surely must a programming language
somewhere with direct support for such an output format. I'm just not personally aware of any.
Joel C. Ewing

[John Gilmore]

'Scaled', 'scaling', and their cognates have a long history in computing.

Packed-decimal arithmetic results are scaled programmatically in
compiled COBOL code. In assembly language, on the other hand, all
packed-decimal arithmetic is integer arithmetic. The programmer who
uses it for fixed-point computations must scale outcomes himself. He
must, that is, keep track of the decimal points in results himself.
The same things are tgrue mutatis mutandis for binary points in
binary-fixed arithmetic.

Fletcher Jones, the founder of the Computer Science Corporation and
its leader when it was an important technical force in the industry,
had the habit of delivering a well-honed exposition of what he called
"the lost art of scaling" with little or no encouragement.

(I did of course mean 2^10 = 1024 and10^3 = 1000.)

John Gilmore, Ashland, MA 01721 - USA

[Ed Gould]

On May 2, 2013, at 3:08 PM, Charles Mills wrote:

> You do get it! <g> Your second sentence is a perfect exposition of
> what I was trying to ask. Your last paragraph is a perfect
> exposition of the problem I am solving with the "K" notation.
>
> Thanks all, especially JG.
>
> "Scaled" seems to be pretty good. Not sure what the opposite is?
> "NoScaled"? "Unscaled"?

live fish?:)

[Gerhard Postpischil]

On 5/2/2013 4:08 PM, Charles Mills wrote:
> Yeah, I suppose it might say just plain In = 25.7KB, but, as we say,
> "the program doesn't work that way." (Also, due to other constraints
> it MUST appear in a "string" format message, not in tabular form like
> most mainframe reports.)

This reminds me of G format in ForTran, although that accepts only
reals. But I don't recall if they had a special name for it.

> And yes, I am doing it "with my own code" -- there is no built-in
> support in the language I am using (C++).

It may be easier to write the code than to arrive at a name for it <g>.
And you never mentioned whether you want truncation, rounding, banker's
rounding, or ?

Gerhard Postpischil
Bradford, Vermont

[Shmuel Metz , Seymour J.]

In <0a4a01ce474d$c93942b0$5babc810$@mcn.org>, on 05/02/2013

at 11:57 AM, Charles Mills <char...@MCN.ORG> said:

>What is the correct term for K, M or G type notation?

IEC prefixes or IEC decimal prefixes if you need to distinguish them
from Ki, Mi, Gi et al.

>OTOH, what would be the contrasting term for "normal" notation:
>"1234" and "4560000"?

"Unabbreviated"?

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Scott Ford]

Ok guys ..I understand the science here but I learned a K = 1024 bytes not 1000 ...am I too old school .....

Scott ford
www.identityforge.com
from my IPAD

'Infinite wisdom through infinite means'

[Charles Mills]

The code is done LOL. Nope, no options. What you get is what you get. Customers are happy with it. I now have a need to be able to turn it OFF (not customer unhappiness; machine parsing: machines are happier with 7867543225 than with 7.86G). So now I need a name for it and for its negation.

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Gerhard Postpischil
Sent: Thursday, May 02, 2013 5:45 PM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

On 5/2/2013 4:08 PM, Charles Mills wrote:
> Yeah, I suppose it might say just plain In = 25.7KB, but, as we say,
> "the program doesn't work that way." (Also, due to other constraints
> it MUST appear in a "string" format message, not in tabular form like
> most mainframe reports.)

This reminds me of G format in ForTran, although that accepts only reals. But I don't recall if they had a special name for it.

> And yes, I am doing it "with my own code" -- there is no built-in
> support in the language I am using (C++).

It may be easier to write the code than to arrive at a name for it <g>.
And you never mentioned whether you want truncation, rounding, banker's rounding, or ?

[Joel C. Ewing]

The "correct" meaning of "K" (kilo) from its Greek origins was "1000".
But, even before the PC weenies took over, "K" was used ambiguously in
the computer industry and required one to understand the context
conventions: At least by the 1960's and S/360 IBM "officially"
(documented in manuals) used "K" to be "1024 "when talking about
binary-addressed memory and to be "1000" when talking about external
storage, which was not binary addressed. For non-binary-addressed
machines (like IBM 1620), "K" still meant "1000" for memory as well; and
on a few machines where octal rather than hex notation was typically
used for memory addresses I have even seen "K" used for 8^3=512 . PC
folks didn't understand those conventions and made things even more
confused by ambiguously using the "1024" definition in inappropriate
contexts that did not involve binary addressing.

In 1998 new international standards were officially adopted to end the
need for any ambiguity by establishing distinct "binary" prefixes to be
used where powers of 2 rather than powers of 10 were meant. Since then
the correct usage is "K" (kilo-) for 1000 and "Ki" (kibi-, short for
"kilo-binary") for 1024, with similar conventions for other multipliers
(http://physics.nist.gov/cuu/Units/binary.html). You will still find
many people who were educated prior to 1998, or taught by those who were
educated before 1998, who are either ignorant of the new standards or
fail to use them consistently and who use "K" ,"M", or "G" in contexts
where "Ki", "Mi", or "Gi" should be used.
Joel C. Ewing

--
Joel C. Ewing, Bentonville, AR jce...@acm.org

[Gerhard Postpischil]

On 5/2/2013 9:55 PM, Charles Mills wrote:
> The code is done LOL. Nope, no options. What you get is what you get.
> Customers are happy with it. I now have a need to be able to turn it
> OFF (not customer unhappiness; machine parsing: machines are happier
> with 7867543225 than with 7.86G). So now I need a name for it and for
> its negation.

Writing an integer in full is referred to as canonical form, thus the
shortened format would be a non-canonical form. You probably want
something shorter, or easier for the non-specialist to understand.


Gerhard Postpischil
Bradford, Vermont

[Elardus Engelbrecht]

Charles Mills wrote:

>Or phrasing the question differently:

Thanks for clarifying your need. You've got all of us in a corner in a rondavel! (round room) ;-D

>consider the integer 45600000. It may be expressed as
>4.56 x 10**7 or 4.56E7 in scientific notation; or as
>45.6 x 10**6 in engineering notation; or as
>45.6M in _______ notation; or as
>45600000 in _________ notation.
>Can anyone fill in those blanks?

What you can do is, say somewhere in your output what the scale of those numbers are 1000, 10 000 or whatever.

45.6 Million of whatever unit
45 600 000 units of whatever (liter/meter/kilogram/dollar, etc)

Or simply 45.6 A and 45 600 000 B where you say somewhere in a footnote that A is million and B is unit x.

On tables, graphs etc, there are sometimes a footnote showing in what units are these numbers.

Say Year 2001 gave 10 and Year 2010 gave 100.
(where numbers are shown in thousands/millions/etc)

Science and mathematics have a format in that 9.999 times 10 to the power of y. There is always one digit to the left of the point.

If you must use powers of ten (or other scale), you can write something like 10^2 or 10^6.

It all depends on what you are showing and whether your readers of your software ca understand.

Good luck!

Groete / Greetings
Elardus Engelbecht

[Scott Ford]

Gerhard,

Like a mangler ..aka manager ...lol

Scott ford
www.identityforge.com
from my IPAD

'Infinite wisdom through infinite means'

[Mike Schwab]

On Fri, May 3, 2013 at 1:11 AM, Elardus Engelbrecht
<elardus.e...@sita.co.za> wrote:
> Charles Mills wrote:
>
>>Or phrasing the question differently:
>
> Thanks for clarifying your need. You've got all of us in a corner in a rondavel! (round room) ;-D

The farmer went crazy in the round barn. Why?
He couldn't find a corner to piss in.

http://www.semissourian.com/blogs/flynch/entry/46666/

--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?

[Charles Mills]

Canonical! I like it:

3. authorized; recognized; accepted.
4. (of a mathematical equation, coordinate, etc.) in simplest or standard form.
-- http://www.thefreedictionary.com/canonical

INTFMT(CANON|SCALED)

Thanks everyone for your help.

Some more helpful than others. Might I humbly observe that some people here are more fond of posting their favorite answer than of reading the question? Why the obsession with kibi, mibi, etc.? (Rhetorical question -- no need to prolong the thread.) There are many, many examples of imprecision in our industry and our world.

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Gerhard Postpischil
Sent: Friday, May 03, 2013 2:11 AM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

On 5/2/2013 9:55 PM, Charles Mills wrote:
> The code is done LOL. Nope, no options. What you get is what you get.
> Customers are happy with it. I now have a need to be able to turn it
> OFF (not customer unhappiness; machine parsing: machines are happier
> with 7867543225 than with 7.86G). So now I need a name for it and for
> its negation.

Writing an integer in full is referred to as canonical form, thus the shortened format would be a non-canonical form. You probably want something shorter, or easier for the non-specialist to understand.

[Scott Ford]

Charles,

Our absolutely right...a lot of mis impressions in this industry ...and lately because of the Net ..there seems to be a lot of no rtfm

Scott ford
www.identityforge.com
from my IPAD

'Infinite wisdom through infinite means'

[Shmuel Metz , Seymour J.]

In <02C2A028-A177-4B92...@yahoo.com>, on 05/02/2013

at 09:16 PM, Scott Ford <scott_...@YAHOO.COM> said:

>Ok guys ..I understand the science here but I learned a K = 1024
>bytes

Back in the 1960's people who used K when they meant 1024 unsderstood
that they were misusing it. In the same epoch I also learned that K
was 512, which was an even worse misuse.

>am I too old school .....

No; were you old schjool then you would use K as 1000 exclusively.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Shmuel Metz , Seymour J.]

In <51832B68...@acm.org>, on 05/02/2013

at 10:13 PM, "Joel C. Ewing" <jce...@ACM.ORG> said:

>The "correct" meaning of "K" (kilo) from its Greek origins was
>"1000". But, even before the PC weenies took over, "K" was used
>ambiguously in the computer industry and required one to understand
>the context conventions: At least by the 1960's and S/360 IBM
>"officially" (documented in manuals) used "K" to be "1024 "when
>talking about binary-addressed memory and to be "1000" when talking
>about external storage, which was not binary addressed. For
>non-binary-addressed machines (like IBM 1620), "K" still meant
>"1000" for memory as well; and on a few machines where octal rather
>than hex notation was typically used for memory addresses I have
>even seen "K" used for 8^3=512 . PC folks didn't understand those
>conventions and made things even more confused by ambiguously using
>the "1024" definition in inappropriate contexts that did not involve
>binary addressing.

Perhaps the absolute worst was the M=1024000 aberration.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Shmuel Metz , Seymour J.]

In <0b9601ce47ec$c6518bd0$52f4a370$@mcn.org>, on 05/03/2013

at 06:55 AM, Charles Mills <char...@MCN.ORG> said:

>Might I humbly observe that some people here are more fond of
>posting their favorite answer than of reading the question?

You might, but it might be more useful to note that some people here
are more fond of rhetoric than with accuracy.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Kirk Talman]

I am curious. I know and understand that 1234567 = 1234.567K = 1.234567M

But is the notation such that 1234567 = 1205.657Ki? And how would one
write the Mi value to as many places?

And how are fractional parts handled in "binary" notation? The link below
did not say. And the example about 1.44 MB diskette was not helpful.

I am guessing that this is a kind of unnatural blend (cross-breed) between
decimal and "binary prefix" notation.

IBM Mainframe Discussion List <IBM-...@LISTSERV.UA.EDU> wrote on
05/02/2013 02:31:45 PM:

> From: John Gilmore <jwgl...@GMAIL.COM>

> See the National Institute of Standards and Technology (NIST) website:

> http://physics.nist.gov/cuu/Units/binary.html

-----------------------------------------
The information contained in this communication (including any
attachments hereto) is confidential and is intended solely for the
personal and confidential use of the individual or entity to whom
it is addressed. If the reader of this message is not the intended
recipient or an agent responsible for delivering it to the intended
recipient, you are hereby notified that you have received this
communication in error and that any review, dissemination, copying,
or unauthorized use of this information, or the taking of any
action in reliance on the contents of this information is strictly
prohibited. If you have received this communication in error,
please notify us immediately by e-mail, and delete the original
message. Thank you

[Charles Mills]

IMHO xxbi scaled notation makes little sense except in the context of things that have a close relationship to integral powers of 2. It might be accurate to say "my annual salary is 65.37 kibibucks (Ki$ ?)" but it is hardly illuminating.

Charles
Composed on a mobile: please excuse my brevity

[Mike Schwab]

On Fri, May 3, 2013 at 3:01 PM, Shmuel Metz (Seymour J.)
<shmue...@patriot.net> wrote:
> In <51832B68...@acm.org>, on 05/02/2013
> at 10:13 PM, "Joel C. Ewing" <jce...@ACM.ORG> said:
>

<<deleted>>

>> PC folks didn't understand those
>>conventions and made things even more confused by ambiguously using
>>the "1024" definition in inappropriate contexts that did not involve
>>binary addressing.
>
> Perhaps the absolute worst was the M=1024000 aberration.
>

From the 3.5 inch diskette?
The first 3.5 inch diskette was 720KB, with 1440 sectors of 512 bytes
for 737,280 bytes.
The next one was Double Density labeled 1.44MB diskette with 2880
sectors of 512 bytes for 1,474,560 bytes.
Divide by 1000 twice it should be 1.475MB.
Divide by 1024 twice it should be 1.406MB.
Only by dividing by 1024 once and 1000 once do you come up with 1.440MB.
Kind of like mixing metaphors. (2010: Easy as cake (no), Piece of
pie (no), It is Easy as pie, Piece of cake).

--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?

[Scott Ford]

Shmuel,

I always used K, with the understanding it was 1024 ...if you write assembler a base register used to cover 4K ,,4096 bytes ....it's all through IBM manuals as far as I know ...unless your doing baseless.....

Scott ford
www.identityforge.com
from my IPAD

'Infinite wisdom through infinite means'

[Mike Schwab]

When you build a memory chip, the input is X number of address bits,
and you have to return 2 ** X number of unique storage bytes. If the
next chip will allow 1 more bit, you have to hold twice as many
storage locations. So memory chips *MUST* be a multiple of 2.
Examples are 10 address bits, 1,024 locations. 20 address bits,
1,048,576. 30 address bits, 1,073,741,824 locations.

When you build a computer disk, you can to return any number of 512
byte sectors, or 4096 byte sectors for newer drives. If you need to
define it in terms of Cylinders, Heads, and Sectors, you round down to
the next multiple.

On Fri, May 3, 2013 at 3:35 PM, Kirk Talman <rkue...@tsys.com> wrote:
> I am curious. I know and understand that 1234567 = 1234.567K = 1.234567M
>
> But is the notation such that 1234567 = 1205.657Ki? And how would one
> write the Mi value to as many places?
>
> And how are fractional parts handled in "binary" notation? The link below
> did not say. And the example about 1.44 MB diskette was not helpful.
>
> I am guessing that this is a kind of unnatural blend (cross-breed) between
> decimal and "binary prefix" notation.

--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?

[Mike Schwab]

On Fri, May 3, 2013 at 5:08 PM, Scott Ford <scott_...@yahoo.com> wrote:
> Shmuel,
>
> I always used K, with the understanding it was 1024 ...if you write assembler a base register used to cover 4K ,,4096 bytes ....it's all through IBM manuals as far as I know ...unless your doing baseless.....
>

1 KM (kilometers) is 1024 M (meters)?

--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?

[J R]

And, if you were offered a job that paid $100K, would you expect to receive $102,400?

=
=
> Date: Fri, 3 May 2013 18:08:28 -0400
> From: scott_...@YAHOO.COM

> Subject: Re: OT - What is the proper term for "K" notation?

> To: IBM-...@LISTSERV.UA.EDU

[Scott Ford]

That would be nice ...

Scott ford
www.identityforge.com
from my IPAD

'Infinite wisdom through infinite means'

[Joel C. Ewing]

But that "K=1024" usage was ONLY because in the context of Assembler
coding you were dealing with processor memory addresses in hardware
instructions, and memory on the S/360 (and its successors) was addressed
by binary values of 24 bits (later 31 bits). So the S/360 PoOp and ASM
manuals, only in discussing memory addresses and blocks of memory, for
brevity used "K" and "M" for close powers of 2. Those manuals were also
very careful to explicitly state that was the convention being used --
precisely because it was not the conventional usage in other contexts,
even within IBM and S/360 (like as Mike recently pointed out, for metric
physical length measurements in IBM manuals, or for DASD).

Other S/360 manuals which dealt with storage measurement in contexts
other than processor memory or blocks of processor memory, most notably
those having to do with DASD and DASD external storage capacity in
bytes, used "K" and "M" with the conventional meanings of powers of 10.

Somewhere along the line a confused notational convention became even
more confused as succeeding generations failed to understand the
underlying rational for the distinct conventions, and in some cases
perhaps even failed to recognize there were two usage conventions.
Usage also got further confused with FBA disk devices with capacity
defined as n*b Bytes, where "b" was a power of 2 and "n" was not.
Where "b" was 512, this made "1024" a convenient multiplier to use for
"K", but some also chose 1000*1024 as a more natural multiplier for M
since the binary dependency ended at 2^9. ... and then you had a
similar issue with describing capacity of Page Data Sets (and now also
Linear Data Sets and PDS/E's) on CKD DASD, where the DASD architecture
has no relation with binary but the physical blocks represent "4K"
processor memory pages, so you have a non-binary-related quantity of 4
KiB blocks.

Those who understood the problem talked about resolving this confusion
in various incompatible ways for over three decades before the binary
prefix standard was finally adopted in 1998. Now if needed, we even have
precise notation for 1000*1024 as "KKi" (which makes it more precise,
but still aesthetically displeasing).
Joel C. Ewing

On 05/03/2013 05:08 PM, Scott Ford wrote:
> Shmuel,
>
> I always used K, with the understanding it was 1024 ...if you write assembler a base register used to cover 4K ,,4096 bytes ....it's all through IBM manuals as far as I know ...unless your doing baseless.....
>
> Scott ford
> www.identityforge.com
> from my IPAD
>
> 'Infinite wisdom through infinite means'
>
>
> On May 3, 2013, at 3:58 PM, "Shmuel Metz (Seymour J.)" <shmue...@PATRIOT.NET> wrote:
>
>> In <02C2A028-A177-4B92...@yahoo.com>, on 05/02/2013
>> at 09:16 PM, Scott Ford <scott_...@YAHOO.COM> said:
>>
>>> Ok guys ..I understand the science here but I learned a K = 1024
>>> bytes
>> Back in the 1960's people who used K when they meant 1024 unsderstood
>> that they were misusing it. In the same epoch I also learned that K
>> was 512, which was an even worse misuse.
>>
>>> am I too old school .....
>> No; were you old schjool then you would use K as 1000 exclusively.
>>
>> --
>> Shmuel (Seymour J.) Metz, SysProg and JOAT
>> Atid/2 <http://patriot.net/~shmuel>
>> We don't care. We don't have to care, we're Congress.
>> (S877: The Shut up and Eat Your spam act of 2003)
>>
>

--
Joel C. Ewing, Bentonville, AR jce...@acm.org

[John Gilmore]

The context-sensitive distinction between a kilobyte, 1000 bytes, and
a kibibyte, 1024 bytes is, finally, a straightforward one, neither
difficult nor arcane; and it is now required.

When the computing community was small and composed of people having
scientific educations it was gratuitous: context switching was easy
for them: a kilometer was 1000 meters and a kilobyte was 1024 bytes.

The IT community is now large and comprised of all sorts of people,
most of whom are ignorant of its history and much else. A
disambiguating distinction was necessary, and it has been made.

In situations of this kind the needs of the unlettered must come
first. The word 'inflammable' does not mean 'cannot be set aflame';
some of the unlettered nevertheless sometimes judged that it did; the
case for disambiguation, for using 'flammable' instead, was thus
compelling; and the change was made. As Quine put the matter,
"Semiliteracy, however offensive, is not a capital offense".

John Gilmore, Ashland, MA 01721 - USA

[Charles Mills]

OK, I'll join in and beat this to death some more. <g>

Not only is the IT community now larger and more unwashed, memory and disk
sizes are larger. When people started using Kilo- to mean 1024 they were
only off by 2.4%. But if you use mega- to mean 2**20 you are off by 4.86%,
and if you use giga- when you mean 2**30 you are off by almost 7.4%, and so
forth.

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On

Behalf Of John Gilmore
Sent: Saturday, May 04, 2013 8:57 AM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: OT - What is the proper term for "K" notation?

The context-sensitive distinction between a kilobyte, 1000 bytes, and a
kibibyte, 1024 bytes is, finally, a straightforward one, neither difficult
nor arcane; and it is now required.

When the computing community was small and composed of people having
scientific educations it was gratuitous: context switching was easy for
them: a kilometer was 1000 meters and a kilobyte was 1024 bytes.

The IT community is now large and comprised of all sorts of people, most of
whom are ignorant of its history and much else. A disambiguating
distinction was necessary, and it has been made.

In situations of this kind the needs of the unlettered must come first. The
word 'inflammable' does not mean 'cannot be set aflame'; some of the
unlettered nevertheless sometimes judged that it did; the case for
disambiguation, for using 'flammable' instead, was thus compelling; and the
change was made. As Quine put the matter, "Semiliteracy, however offensive,
is not a capital offense".

[Shmuel Metz , Seymour J.]

In
<OF2170D0C5.B3DD89D3-ON85257B...@tsys.tss.net>,
on 05/03/2013

at 04:35 PM, Kirk Talman <rkue...@TSYS.COM> said:

>But is the notation such that 1234567 = 1205.657Ki?

ITYM 1205.631Ki.

>And how would one write the Mi value to as many places?

foo = (foo/1048576)Mi; the division is a normal decimal division to as
many poaces as you need.

>I am guessing that this is a kind of unnatural blend (cross-breed)
>between decimal and "binary prefix" notation.

No; those units *are* binary prefix notation. For an unholy blend,
look at givinf capacities in units of 1024000 :-(

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Shmuel Metz , Seymour J.]

In
<CAJTOO5_ED5s0uQiEf8AYTiXs...@mail.gmail.com>,
on 05/03/2013

at 05:22 PM, Mike Schwab <mike.a...@GMAIL.COM> said:

>When you build a memory chip, the input is X number of address bits,

Hasn't anybody built ternary memory?

"Silly wabbit, trits are for kids."

>When you build a computer disk, you can to return any number of 512
>byte sectors,

Sectors? We don't need no stinking sectors. The IBM 1301, 1302 and
23xx disks did not have sectors.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Shmuel Metz , Seymour J.]

In <8D96815F-E298-46FD...@yahoo.com>, on 05/03/2013

at 06:08 PM, Scott Ford <scott_...@YAHOO.COM> said:

>I always used K, with the understanding it was 1024

I guess that you never programmed the CDC 6x00 and Cyber 70 machines,
where they referred to 512 as "octal K".

[Shmuel Metz , Seymour J.]

In
<CAE1XxDFKms1d9P+EMdPiOaPN...@mail.gmail.com>,
on 05/04/2013

at 08:56 AM, John Gilmore <jwgl...@GMAIL.COM> said:

>In situations of this kind the needs of the unlettered must come
>first. The word 'inflammable' does not mean 'cannot be set aflame';
>some of the unlettered nevertheless sometimes judged that it did;
>the case for disambiguation, for using 'flammable' instead, was
>thus compelling; and the change was made. As Quine put the
>matter, "Semiliteracy, however offensive, is not a capital
>offense".

Sometimes it is, e.g., reading "inflamable" as "nonflamable" and
lighting a match near an open container.

--
Shmuel (Seymour J.) Metz, SysProg and JOAT
Atid/2 <http://patriot.net/~shmuel>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)

[Lloyd Fuller]

Actually, Univac played with it back in the 1960s/1970s. I believe they were
working with someone from OSU. As far as I know, they never marketed a product
with it, but it was at least researched.

Lloyd

----- Original Message ----
From: Shmuel Metz (Seymour J.) <shmue...@PATRIOT.NET>
To: IBM-...@LISTSERV.UA.EDU
Sent: Mon, May 6, 2013 8:00:34 AM
Subject: Re: OT - What is the proper term for "K" notation?

[Shmuel Metz , Seymour J.]

In <1367845369.97...@web181401.mail.ne1.yahoo.com>, on
05/06/2013

at 06:02 AM, Lloyd Fuller <lefu...@SBCGLOBAL.NET> said:

>Actually, Univac played with it back in the 1960s/1970s.

Any ternary logic or memory in the 1960's was probably implemented
with discrete transistors rather than with IC's.

[Lloyd Fuller]

Not sure. I was just talking one time to one of the military people that were
involved with Univac and the university (Oregon State, I think). He mentioned
that they had experimented with it. And from the time frame you are probably
correct that it was transistors rather than IC's.

Lloyd



----- Original Message ----
From: Shmuel Metz (Seymour J.) <shmue...@PATRIOT.NET>
To: IBM-...@LISTSERV.UA.EDU
Sent: Wed, May 8, 2013 4:28:58 PM
Subject: Re: OT - What is the proper term for "K" notation?

[Joel C. Ewing]

On 05/08/2013 08:02 PM, Lloyd Fuller wrote:
> Not sure. I was just talking one time to one of the military people that were
> involved with Univac and the university (Oregon State, I think). He mentioned
> that they had experimented with it. And from the time frame you are probably
> correct that it was transistors rather than IC's.
>
> Lloyd
>
>
>
> ----- Original Message ----
> From: Shmuel Metz (Seymour J.) <shmue...@PATRIOT.NET>
> To: IBM-...@LISTSERV.UA.EDU
> Sent: Wed, May 8, 2013 4:28:58 PM
> Subject: Re: OT - What is the proper term for "K" notation?
>
> In <1367845369.97...@web181401.mail.ne1.yahoo.com>, on
> 05/06/2013
> at 06:02 AM, Lloyd Fuller <lefu...@SBCGLOBAL.NET> said:
>
>> Actually, Univac played with it back in the 1960s/1970s.
> Any ternary logic or memory in the 1960's was probably implemented
> with discrete transistors rather than with IC's.
>

The Soviets actually built a computer model (Setun) with ternary logic
in 1958, fifty built before production was finally halted in 1965.
See http://en.wikipedia.org/wiki/Ternary_computer
and
http://en.wikipedia.org/wiki/Setun
Those articles don't go into much detail, but other sources describe
memory as being ternary as well, using two magnetic cores for a single
ternary digit or two tracks on drum memory. Their ternary approach was
apparently an ingenious solution to minimize costs within the
constraints of existing technology.

If the Soviets were trying that route, that alone would probably have
caused our military to explore it. I would speculate that once
large-scale integrated circuits tailored for binary logic became
universally available, that effectively guaranteed that computers with
binary logic and memory were much cheaper to build than ternary
computers and probably explains why binary computers eventually replaced
ternary development in the Soviet Union as well.

--
Joel C. Ewing, Bentonville, AR jce...@acm.org

[Mike Schwab]

Ternary, with values of 0,1,2, could be useful in a binary addition logic.
You are only going to store 0 or 1, and the 2 represents a delayed carry of 1+1.

--
Mike A Schwab, Springfield IL USA
Where do Forest Rangers go to get away from it all?

[Ed Finnell]

Yeah, they even came up with flip-flap-flop nomenclature. Died pretty
quick, but it sounded cool.


In a message dated 5/8/2013 10:46:50 P.M. Central Daylight Time,

mike.a...@GMAIL.COM writes:

Ternary, with values of 0,1,2, could be useful in a binary addition logic.
You are only going to store 0 or 1, and the 2 represents a delayed carry
of 1+1

[zMan]

There's a joke in there somewhere about "flop" and Soviet technology, I
suspect...

--
zMan -- "I've got a mainframe and I'm not afraid to use it"

[John Gilmore]

As Ogden Nash wrote, "Candy is dandy but liquor is quicker".

There are situations in which a radix other than 2 is episodically
useful; and when these situations arise a binary machine can be
programmed very readily to exploit, say, ternary or octal logic.
Moreover, binary machines can and almost always do exhibit some
crucial ternary, signum-like behavior.

John Gilmore, Ashland, MA 01721 - USA

[Elardus Engelbrecht]

zMan wrote:

>There's a joke in there somewhere about "flop" and Soviet technology, I suspect...

One version is 'AeroFlop' aka 'AeroSplat'. Google 'AeroFlop' about complaints about AeroFlot (Russian Airline).

Groete / Greetings
Elardus Engelbrecht (Not a Flop! ;-D )

[John Gilmore]

Jokes about Soviet technology could backfire.

When an 'important technological announcement' was bruited by Tass,
one American wag's response was, "They're going to announce that
they've invented the Waring Blender"; but what they announced was
Sputnik;, and Tass then felt free to describe a new American 'Sputnik
cocktail', made with two parts vodka and one part sour grapes.

John Gilmore, Ashland, MA 01721 - USA

[Joel C. Ewing]

On 05/09/2013 06:53 AM, John Gilmore wrote:
> As Ogden Nash wrote, "Candy is dandy but liquor is quicker".
>
> There are situations in which a radix other than 2 is episodically
> useful; and when these situations arise a binary machine can be
> programmed very readily to exploit, say, ternary or octal logic.
> Moreover, binary machines can and almost always do exhibit some
> crucial ternary, signum-like behavior.
>
> John Gilmore, Ashland, MA 01721 - USA
>
>

According to the various articles, the advantages of the Setun ternary
computer were "lower electricity consumption and lower production cost",
and that when the Setun at Moscow State University was eventually
replaced by a binary machine of roughly equivalent capability the binary
machine "was 2.5 times the cost of the Setun". The first Setun built
was also reported to have run 17 years without a failure. Simulating
ternary logic and arithmetic on a binary machine would obviously have
lost those advantages of a radically different hardware design.

The articles imply the decision to phase out Soviet ternary machines
must have been based more on the economic costs and inconvenience of
being different from the rest of the world than the costs of the
hardware itself.

--
Joel C. Ewing, Bentonville, AR jce...@acm.org

[John Gilmore]

Joel C. Ewing wrote:

| The first Setun built was also reported to
| have run 17 years without a failure.

Yes, indeed; and it did so by tapping the Hegelian absolute, without
consuming electric power.

John Gilmore, Ashland, MA 01721 - USA

[Tony's new office PC via Mozilla]

This is a first! A quotation and an author was cited that I have
actually heard of in my past. The following paragraph did send me to
the dictionary for four definitions though......

.....concluding my Vulcan Mind Meld for today.

:-)

[Sheldon Davis]

Hi

We are using BMCto take an instant snapshot of about nine hundred tables.
The BMC XBM takes a lot of CPU and the Copy Services started task ANTMAIN also takes a lot of CPU
Both these stc's have a higher priority than our CICS.
Has anyone else experienced this problem?
we use Z/OS version 1.13

Thanks in advance

Sheldon Davis

[Lizette Koehler]

I would like a little more detail

1) What is the comparison for the phrase "a lot of CPU"? Do you have
previous metrics to show this?
a) Is the workload in the system being impacted?
b) Does this spike or stay steady?
2) Have you opened a case to BMC?
3) When did this start? Did you make any recent changes to the
environment? OS upgrade/maint, hardware change? Etc...

Thanks

Lizette

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On

[John Eells]

The DFSMS copy services folk who own ANTMAIN would be interested in
knowing more if you'd like to open a PMR...

Sheldon Davis wrote:
> We are using BMCto take an instant snapshot of about nine hundred tables.
> The BMC XBM takes a lot of CPU and the Copy Services started task ANTMAIN also takes a lot of CPU
> Both these stc's have a higher priority than our CICS.
> Has anyone else experienced this problem?
> we use Z/OS version 1.13

--
John Eells
z/OS Technical Marketing
IBM Poughkeepsie
ee...@us.ibm.com

[Graham Harris]

We too have had issues in the past with "high CPU" in ANTMAIN, although
generated by DFDSS COPY rather than BMC utilities, but the root issue could
be similar, or the same.
The problem for us was eventually narrowed down to specifically occurring
when allocations were targetting (in our case) SMS storage groups with
'considerable' numbers of volumes (best part of a thousand). We opened a a
PMR for this, and we got the info back that for every dataset allocation,
SMS passes each eligible volume to SDM so it can check for flashcopy
eligibility (which in our case was not appropriate because we didnt have
any flashcopy capable devices - but the DFDSS default is to allow flashcopy
[aka FASTREP in DFDSS terms]).
So the ANTMAIN CPU burn is largely to do with SMS passing the volumes one
by one to SDM for every single allocation, and SDM checking flashcopy
eligibilty/parings for each volume.

We get round this in our case, by specifying FASTREP(NONE) in DFDSS COPY
sysin, to prevent this situation from occurring.

So if you
a) have lots of target volumes, and
b) the BMC utility facilitates some sort of flashcopy support
then you could be experiencing a similar issue.
If you can totally turn off flashcopy being invoked within BMC (even i you
dont have flashcopy capability, like us), that may help the problem go away.

There was a marketing requirement raised on the back of our PMR
(MR0719125224) suggesting some semblance of caching being done of the
flashcopy eligibilty/pairing info for each volume as it is encountered
(instead of having to look it up afresh for every allocation!), but the PMR
was closed before any feedback was given on that. Perhaps some helpful
IBMer could look up the status of that?

HTH

On 9 May 2013 17:22, John Eells <ee...@us.ibm.com> wrote:

> The DFSMS copy services folk who own ANTMAIN would be interested in
> knowing more if you'd like to open a PMR...
>
>
> Sheldon Davis wrote:
>
>> We are using BMCto take an instant snapshot of about nine hundred tables.
>> The BMC XBM takes a lot of CPU and the Copy Services started task ANTMAIN
>> also takes a lot of CPU
>> Both these stc's have a higher priority than our CICS.
>> Has anyone else experienced this problem?
>> we use Z/OS version 1.13
>>
>
>
> --
> John Eells
> z/OS Technical Marketing
> IBM Poughkeepsie
> ee...@us.ibm.com
>
>

> ------------------------------**------------------------------**----------

[Sheldon Davis]

Thanks

I tried the same copy without BMC and ANTMAIN still uses an excessive amount of CPU
I have opened a PMR with IBM

[Peter Relson]

<open can of worms>

When someone writes about "excessive" amount of CPU time, I'm always left
wondering.
"A lot" of CPU time, I understand. "More than it used to", I understand.
But "excessive" -- compared to what?

If you ask for something to be done, whether explicitly or implicitly, it
will take whatever amount of CPU time it takes.
Maybe in the ANTMAIN case (I have no idea) you're lucky it's not done
under your job and charged to you (as it perhaps could be).

If you view the amount as excessive, you might choose not to do that
something. The amount of processing done might be intentional and
necessary. Or it might be a problem. You're of course welcome to inquire
and it will be looked at. But there is no necessary correlation between
inquiry and change.

</end-open>

Peter Relson
z/OS Core Technology Design

[Sheldon Davis]

Hi

Thanks the problem was a storage group with many volumes.
ANTMAIN now takes 1.5 percent CPU of our machine as opposed to 90 percent.

Best Regards

Sheldon Davis

[Sheldon Davis]

My apologies if I was not clear or if I offended you.
Exessive for me was when ANTMAIN with a service class of SYSTEM took ninety percent CPU for about two minutes while the flash copies were running.
Anyway the problem is solved thanks to Graham Harris.

Best Regards

Sheldon Davis

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU] On Behalf Of Peter Relson
Sent: Monday, May 13, 2013 2:42 PM
To: IBM-...@LISTSERV.UA.EDU
Subject: Re: ANTMAIN takes a lot of CPU

[John Gilmore]

The word "excessive" is sometimes, I think, legitimate shorthand for
"unexpectedly large in comparison with what has been the case".

It is important is to remember that "excess" is symptomatic. It
identifies something unfortunate that has happened. The question who
is be blamed for that thing, when it is meaningful at all, is a
different one that may, often does, have an answer that is not the
obvious one.

John Gilmore, Ashland, MA 01721 - USA

[Graham Harris]

I understand the root cause of this issue (which is essentially repetitive
parallel execution of large numbers of SRBs within ANTMAIN when
flashcopy eligibility/pairing checking is applied against very large
numbers of volumes) is "recognised", and is intended to be addressed in a
future version of z/OS. The way to avoid ANTMAIN CPU spikes for the time
being, appears to be to make the eligible volume target list as small as
possible, or turn off the flashcopy 'trigger' in whatever software you are
using (assuming you don't actually need flashcopy of course!)

[Gibney, Dave]

Since no one has asked yet, I will.
Is SYSTEM the appropriate level of priority for this task?
I don't have any flashcopy here, so I have no experience or even knowledge of all that ANTMAIN does.


Dave Gibney
Information Technology Services
Washington State University

> -----Original Message-----
> From: IBM Mainframe Discussion List [mailto:IBM-...@LISTSERV.UA.EDU]
> On Behalf Of Graham Harris
> Sent: Monday, May 13, 2013 11:38 AM
> To: IBM-...@LISTSERV.UA.EDU
> Subject: Re: ANTMAIN takes a lot of CPU
>

[Ted MacNEIL]

ANTMAIN is one of the tasks IBM automatically assigns to SYSTEM, iirc.
So, if this is (still) true, the answer is YES.
-
Ted MacNEIL
eama...@yahoo.ca
Twitter: @TedMacNEIL

[nitz...@gmx.net]

> Since no one has asked yet, I will.
> Is SYSTEM the appropriate level of priority for this task?
> I don't have any flashcopy here, so I have no experience or even knowledge of all that ANTMAIN does.

Regardless if you use flashcopy or not, you'll have an ANTMAIN address space automatically started at IPL (and an ANTAS000). And SYSTEM is where the SPM rule in WLM puts it, so it must be on the list IBM maintains that puts things into SYSTEM. Not sure if that can be overwritten, though. (I do know that you cannot force CONSOLE out of SYSTEM anymore, and a good thing, too!)

Barbara

[John Sullivan]

I am glad that the problem was alleviated by reducing the number of volumes in the Storage Group. I would like to add that XBM and the associated BMC Utilities do have options available to turn off the use of Flashcopy.
BMC Support would be happy to work with you if you would like assistance to do so.

John Sullivan, Sr. Product Developer on XBM with BMC Software, Inc.