zIIP processor (ENGINE type IP)

[fish-8]

 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----

[davekreiss]

The amount of time a process [1] that can run on a zIIP or zAAP is
called eligible time.  If there are any zIIP or zAAP processors and the
dispatcher can dispatch this eligible time on those processors than any
time spent running work on those processors is not part of chargeable
CPU time.  So not only is eligibility important but also the
availability of the specialty [2] processors important.  Another key
factor is that specialty processors always run at full speed (won't
matter on a Hercules emulation).  So a CPU bound process on a regular
processor can run faster (use less CPU time) on a specialty processor.

I suspect Hercules doesn't have a way to define a processor of a zIIP or
zAAP type.

Dave

 

1 -task mode is JAVA/XML work is zAAP eligible and enclave SRB work is
zIIP eligible

2 - zIIP or zAAP processor

 

From: hercules-390@... [mailto:hercules-390@...]
On Behalf Of Fish
Sent: Monday, June 22, 2009 1:19 PM
To: Hercules-390
Subject: [hercules-390] zIIP processor (ENGINE type IP)

 







-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
<mailto:fish%40softdevlabs.com>
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----





[Non-text portions of this message have been removed]

[Dave Wade]

>-----Original Message-----
>From: hercules-390@... [mailto:hercules-390@...] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 21:34
>To: hercules-390@...
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>The amount of time a process [1] that can run on a zIIP or zAAP is
>called eligible time.  If there are any zIIP or zAAP processors and the
>dispatcher can dispatch this eligible time on those processors than any
>time spent running work on those processors is not part of chargeable
>CPU time.  

What decides what is charged? Is this an integrated part of zOS or something
recorded in the HMC ...

> So not only is eligibility important but also the
>availability of the specialty [2] processors important.  Another key
>factor is that specialty processors always run at full speed (won't
>matter on a Hercules emulation).  So a CPU bound process on a regular
>processor can run faster (use less CPU time) on a specialty processor.
>
>I suspect Hercules doesn't have a way to define a processor of a zIIP or
>zAAP type.
>

http://www.hercules-390.org/hercconf.html#ENGINES

>Dave
>
>
>1 -task mode is JAVA/XML work is zAAP eligible and enclave SRB work is
>zIIP eligible
>
>2 - zIIP or zAAP processor
>
>

From: hercules-390@... [mailto:hercules-390@...]
On Behalf Of Fish
Sent: Monday, June 22, 2009 1:19 PM
To: Hercules-390
Subject: [hercules-390] zIIP processor (ENGINE type IP)

 







-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
<mailto:fish%40softdevlabs.com>
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----





[Non-text portions of this message have been removed]



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

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

[davekreiss]

This is a WLM thing.   Time spent on a specialty processor is recorded
in SMF type 30 in special fields for zIIP and zAAP.   And the 4 hour
rolling MSU for CP (reported in SCRT) is reduced by time spent on
specialty processors since it time not running on a general processor.

 

I didn't realize that you could specify specialty processors but of
course that makes sense, the Hercules developers are on the ball...

 

Dave

 

From: hercules-390@... [mailto:hercules-390@...]
On Behalf Of Dave Wade
Sent: Monday, June 22, 2009 3:11 PM
To: hercules-390@...
Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)

 






>-----Original Message-----
>From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 21:34
>To: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>The amount of time a process [1] that can run on a zIIP or zAAP is
>called eligible time. If there are any zIIP or zAAP processors and the
>dispatcher can dispatch this eligible time on those processors than any
>time spent running work on those processors is not part of chargeable
>CPU time.

What decides what is charged? Is this an integrated part of zOS or
something
recorded in the HMC ...

> So not only is eligibility important but also the
>availability of the specialty [2] processors important. Another key
>factor is that specialty processors always run at full speed (won't
>matter on a Hercules emulation). So a CPU bound process on a regular
>processor can run faster (use less CPU time) on a specialty processor.
>
>I suspect Hercules doesn't have a way to define a processor of a zIIP
or
>zAAP type.
>

http://www.hercules-390.org/hercconf.html#ENGINES

>Dave
>
>
>1 -task mode is JAVA/XML work is zAAP eligible and enclave SRB work is
>zIIP eligible
>
>2 - zIIP or zAAP processor
>
>

From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ]
On Behalf Of Fish
Sent: Monday, June 22, 2009 1:19 PM
To: Hercules-390
Subject: [hercules-390] zIIP processor (ENGINE type IP)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
<mailto:fish%40softdevlabs.com>
<mailto:fish%40softdevlabs.com>
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----

[Non-text portions of this message have been removed]

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

Community email addresses:
Post message: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
Subscribe: hercules-390-subscribe@...
<mailto:hercules-390-subscribe%40yahoogroups.com>
Unsubscribe: hercules-390-unsubscribe@...
<mailto:hercules-390-unsubscribe%40yahoogroups.com>
List owner: hercules-390-owner@...
<mailto:hercules-390-owner%40yahoogroups.com>

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org

Yahoo! Groups Links





[Non-text portions of this message have been removed]

[Dave Wade]

>-----Original Message-----
>From: hercules-390@... [mailto:hercules-390@...] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 23:23
>To: hercules-390@...
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>This is a WLM thing.   Time spent on a specialty processor is recorded
>in SMF type 30 in special fields for zIIP and zAAP.   And the 4 hour
>rolling MSU for CP (reported in SCRT) is reduced by time spent on
>specialty processors since it time not running on a general processor.

I wonder if Hercules isn't reporting speciality processors properly if this
is wrong...

>I didn't realize that you could specify specialty processors but of
>course that makes sense, the Hercules developers are on the ball...

The differences are pretty minor....

>Dave

 

From: hercules-390@... [mailto:hercules-390@...]
On Behalf Of Dave Wade
Sent: Monday, June 22, 2009 3:11 PM
To: hercules-390@...
Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)

 






>-----Original Message-----
>From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 21:34
>To: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>The amount of time a process [1] that can run on a zIIP or zAAP is
>called eligible time. If there are any zIIP or zAAP processors and the
>dispatcher can dispatch this eligible time on those processors than any
>time spent running work on those processors is not part of chargeable
>CPU time.

What decides what is charged? Is this an integrated part of zOS or
something
recorded in the HMC ...

> So not only is eligibility important but also the
>availability of the specialty [2] processors important. Another key
>factor is that specialty processors always run at full speed (won't
>matter on a Hercules emulation). So a CPU bound process on a regular
>processor can run faster (use less CPU time) on a specialty processor.
>
>I suspect Hercules doesn't have a way to define a processor of a zIIP
or
>zAAP type.
>

http://www.hercules-390.org/hercconf.html#ENGINES

>Dave
>
>
>1 -task mode is JAVA/XML work is zAAP eligible and enclave SRB work is
>zIIP eligible
>
>2 - zIIP or zAAP processor
>
>

From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ]
On Behalf Of Fish
Sent: Monday, June 22, 2009 1:19 PM
To: Hercules-390
Subject: [hercules-390] zIIP processor (ENGINE type IP)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
<mailto:fish%40softdevlabs.com>
<mailto:fish%40softdevlabs.com>
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----

[Non-text portions of this message have been removed]

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

Community email addresses:
Post message: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
Subscribe: hercules-390-subscribe@...
<mailto:hercules-390-subscribe%40yahoogroups.com>
Unsubscribe: hercules-390-unsubscribe@...
<mailto:hercules-390-unsubscribe%40yahoogroups.com>
List owner: hercules-390-owner@...
<mailto:hercules-390-owner%40yahoogroups.com>

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org

Yahoo! Groups Links





[Non-text portions of this message have been removed]



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

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

[davekreiss]

 

A DM=CPU should tell the story.

RESPONSE=SYSA                        

 IEE174I 17.41.47 DISPLAY M 535      

 PROCESSOR STATUS                    

 ID  CPU                  SERIAL    

 00  +                     0XXXXX2098

 01  +                     0XXXXX 2098

 02  +                     0XXXXX 2098

 03  +A                    0XXXXX 2098

 04  +I                    0XXXXX 2098

Serial number Xed out : - )

 

In this  case processor 3 is zAAP and 4 is zIIP.  

 

If the D M=CPU under Hercules doesn't show these as the correct
processor types the results would be as seen.

 

Dave

 

From: hercules-390@... [mailto:hercules-390@...]
On Behalf Of Dave Wade
Sent: Monday, June 22, 2009 3:33 PM
To: hercules-390@...
Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)

 






>-----Original Message-----
>From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 23:23
>To: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>This is a WLM thing. Time spent on a specialty processor is recorded
>in SMF type 30 in special fields for zIIP and zAAP. And the 4 hour
>rolling MSU for CP (reported in SCRT) is reduced by time spent on
>specialty processors since it time not running on a general processor.

I wonder if Hercules isn't reporting speciality processors properly if
this
is wrong...

>I didn't realize that you could specify specialty processors but of
>course that makes sense, the Hercules developers are on the ball...

The differences are pretty minor....

>Dave

From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com> ]
On Behalf Of Dave Wade
Sent: Monday, June 22, 2009 3:11 PM
To: hercules-390@... <mailto:hercules-390%40yahoogroups.com>

Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)

>-----Original Message-----
>From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com> ] On
>Behalf Of David Kreiss
>Sent: 22 June 2009 21:34
>To: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com>
>Subject: RE: [hercules-390] zIIP processor (ENGINE type IP)
>
>The amount of time a process [1] that can run on a zIIP or zAAP is
>called eligible time. If there are any zIIP or zAAP processors and the
>dispatcher can dispatch this eligible time on those processors than any
>time spent running work on those processors is not part of chargeable
>CPU time.

What decides what is charged? Is this an integrated part of zOS or
something
recorded in the HMC ...

> So not only is eligibility important but also the
>availability of the specialty [2] processors important. Another key
>factor is that specialty processors always run at full speed (won't
>matter on a Hercules emulation). So a CPU bound process on a regular
>processor can run faster (use less CPU time) on a specialty processor.
>
>I suspect Hercules doesn't have a way to define a processor of a zIIP
or
>zAAP type.
>

http://www.hercules-390.org/hercconf.html#ENGINES

>Dave
>
>
>1 -task mode is JAVA/XML work is zAAP eligible and enclave SRB work is
>zIIP eligible
>
>2 - zIIP or zAAP processor
>
>

From: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com>
[mailto:hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com> ]
On Behalf Of Fish
Sent: Monday, June 22, 2009 1:19 PM
To: Hercules-390
Subject: [hercules-390] zIIP processor (ENGINE type IP)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a question about how cpu-types are implemented in Hercules.

I'm told that if a certain workload (or program or process or
whatever terminology you prefer) is run on a real box without any
zIIPs, the workload is "charged" the full amount (say 50 minutes).
When run on a real box with a zIIP however, it is "charged" only .2
minutes of processor time. They tell me this is because the bulk of
their workload is designed to run on the zIIP processor which is
"free" (or something like that; I'm not all that familiar with it).

When they run the same program on Hercules with a zIIP defined
however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
minutes, and they'd like to know why.

Does anyone know what they're referring to and more importantly, can
anyone explain WHY this happens under Hercules? They would like the
"chargeable time" their program uses when run on Hercules with a zIIP
defined to be the same .2 minutes as when the same program is run on
a real box. How do they do that? What are they doing wrong? Or is
Hercules doing something wrong? If so, what?

Thanks in advance for any insight anyone can provide.

- --
"Fish" (David B. Trout) - fish@...
<mailto:fish%40softdevlabs.com>
<mailto:fish%40softdevlabs.com>
<mailto:fish%40softdevlabs.com>
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
7115Bzqfb46/pdaUmNxgxvAo
=InTY
-----END PGP SIGNATURE-----

[Non-text portions of this message have been removed]

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

Community email addresses:
Post message: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
<mailto:hercules-390%40yahoogroups.com>
Subscribe: hercules-390-subscribe@...
<mailto:hercules-390-subscribe%40yahoogroups.com>
<mailto:hercules-390-subscribe%40yahoogroups.com>
Unsubscribe: hercules-390-unsubscribe@...
<mailto:hercules-390-unsubscribe%40yahoogroups.com>
<mailto:hercules-390-unsubscribe%40yahoogroups.com>
List owner: hercules-390-owner@...
<mailto:hercules-390-owner%40yahoogroups.com>
<mailto:hercules-390-owner%40yahoogroups.com>

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org

Yahoo! Groups Links

[Non-text portions of this message have been removed]

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

Community email addresses:
Post message: hercules-390@...
<mailto:hercules-390%40yahoogroups.com>
Subscribe: hercules-390-subscribe@...
<mailto:hercules-390-subscribe%40yahoogroups.com>
Unsubscribe: hercules-390-unsubscribe@...
<mailto:hercules-390-unsubscribe%40yahoogroups.com>
List owner: hercules-390-owner@...
<mailto:hercules-390-owner%40yahoogroups.com>

Files and archives at:
http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
http://www.hercules-390.org

Yahoo! Groups Links





[Non-text portions of this message have been removed]

[Harold Grovesteen-2]

zIIP's for DB2 and XML workloads and zAAP's fo Java workloads are really
marketing tools of IBM. By offloading work to these CPU's the potential
exists to reduce usage of standard CPU's such that an organization does
not have to buy additional standard CPU's for this work. The real target
of this is reduced software costs for mainframe software. By
constraining or reducing the MSU rating of the system, software costs
are also constrained or reduced. This strategy is geared to address
those with z/OS that might be reluctant to increase database work or
introduce or expand Java workloads in z/OS. As these engines do not
increase MSU's, they can run full throttle without impacting those
charges. The concept is similar to IFL's which create a standard per CPU
cost model for software on Linux without raising the MSU rating of the
system. There are in fact ways to estimate the impact of these engines
on CPU resource requirements without actually having them.

At a recent zSeries roadshow I heard that software vendors are starting
to take notice of these approaches, not for just the mainframe but more
particularly for virtualized systems on Intel with products such as
VMWare. Because the per CPU charging model basically takes money out of
the pockets of the software vendors where virtualization is used vs.
revenue created for stand-alone physical servers, some are starting or
thinking about moving to a per instance charging model. My suspicions
are that this industry will go where Microsoft goes. If Microsoft starts
to use this per instance model, the rest of the Wintel world will too.
This of course will have the effect of causing companies to look more
seriously at application and middleware open source solutions.

On Hercules, where a CPU is really a host thread, the benefit is to
allow a z/OS system to model the physical mainframe environment for a
single LPAR.

Fish wrote:

>
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>I have a question about how cpu-types are implemented in Hercules.
>
>I'm told that if a certain workload (or program or process or
>whatever terminology you prefer) is run on a real box without any
>zIIPs, the workload is "charged" the full amount (say 50 minutes).
>When run on a real box with a zIIP however, it is "charged" only .2
>minutes of processor time. They tell me this is because the bulk of
>their workload is designed to run on the zIIP processor which is
>"free" (or something like that; I'm not all that familiar with it).
>
>When they run the same program on Hercules with a zIIP defined
>however (e.g. ENGINES 7*CP,IP), they're being "charged" the full 50
>minutes, and they'd like to know why.
>
>Does anyone know what they're referring to and more importantly, can
>anyone explain WHY this happens under Hercules? They would like the
>"chargeable time" their program uses when run on Hercules with a zIIP
>defined to be the same .2 minutes as when the same program is run on
>a real box. How do they do that? What are they doing wrong? Or is
>Hercules doing something wrong? If so, what?
>
>Thanks in advance for any insight anyone can provide.
>

This whole area was the source of the divide by zero problems we were
having with recent releases of z/OS. This is a fairly undocumented area
and exactly how z/OS uses the values retrieved and what they should be
has undergone a fair amount of discussion on the developers list.

Whether we will ever get this to match is not clear to me. What is
ambiguous from your post is whether all of the work got charged to a
standard CPU or all of the work got charged to the zIIP. Either way,
this is different than the 12 seconds identified in your message. A
guess would be that whatever values have been presented by Hercules to
z/OS has caused z/OS to dispatch the work differently than on a real
mainframe. Getting these numbers to match what the real mainframe sees
would require the same numbers. Maybe someone else knows whether these
are static or change. If they are static we have a shot at getting them
to be close.

Harold

>
>- --
>"Fish" (David B. Trout) - fish@...
>Fight Spam! Join CAUCE! <http://www.cauce.org/>
>7 reasons why HTML email is a bad thing
>http://www.georgedillon.com/web/html_email_is_evil.shtml
>PGP key fingerprints:
>RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
>DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
>
>-----BEGIN PGP SIGNATURE-----
>Version: PGP 8.1
>
>iQA/AwUBSj/nRUj11/TE7j4qEQL06wCfQlV9xiU+W9XikP6+GVOUxQJUUV0AoM++
>7115Bzqfb46/pdaUmNxgxvAo
>=InTY
>-----END PGP SIGNATURE-----
>
>
>
>------------------------------------
>
>Community email addresses:
>  Post message: hercules-390@...
>  Subscribe:    hercules-390-subscribe@...
>  Unsubscribe:  hercules-390-unsubscribe@...
>  List owner:   hercules-390-owner@...
>
>Files and archives at:
>  http://groups.yahoo.com/group/hercules-390
>
>Get the latest version of Hercules from:
>  http://www.hercules-390.org
>
>Yahoo! Groups Links
>
>
>
>
>  
>

[fish-8]

 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David Kreiss wrote:

> A DM=CPU should tell the story.
[...]
> If the D M=CPU under Hercules doesn't show these as the
> correct processor types the results would be as seen.

And therein is the problem: the D M=CPU display does show the "+I"
for processor 7, but yet their total chargeable time isn't any less.
Their total time is the same whether a zIIP is defined in Hercules or
not.

On a real machine they're different. On a real machine with a zIIP
the chargeable time ("SRB time" I think they called it?) is very
small, leading them to confidently conclude their program/workload is
indeed running on the zIIP, but on Hercules, whether a zIIP is define
or not makes no difference: the chargeable time is always the full
amount! (even though as I explained it appears as if the o/s does
know the zIIP exists since it's appearing properly in the D M=CPU
display).

The big question is WHY this behavior occurs? What is Hercules doing
wrong?

Or are they doing something wrong? Do they need to configure
something in the o/s that they possibly forgot to do?

Where should I start looking?

TIA

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSkD8Zkj11/TE7j4qEQIfoQCeLunlZ9JXC8gjmZlMFSvvSA+cPWcAoOwb
9tLpeEEOEPeMHOnRIpO2WMY5
=YCHk
-----END PGP SIGNATURE-----

[Mike Schwab]

http://publib.boulder.ibm.com/infocenter/zos/v1r9/index.jsp?topic=/com.ibm.zos.r9.e0zh300/smfrmf.htm

Type 70–79; z/OS V1R8; Bit 5 of byte 49 of the RMF product section
topic is added to indicate whether zIIP processors are available on
the system.; Support of zIIP processors

Type 70 (RMF™ Processor Activity); z/OS V1R8; New fields: In subtype
1, the following sections are extended to support new information for
group capacity limit:
CPU Control Section
PR/SM Partition Data Section
Also in subtype 1, the following sections are extended to support new
information for zIIP processors:

CPU Control Section
CPU Data Section
CPU Identification Section;

 Release update and support of zIIP processors ;

http://publib.boulder.ibm.com/infocenter/zos/v1r10/index.jsp?topic=/com.ibm.zos.r10.e0za100/chnewfu.htm

Description: By exploiting the CF architected level 15 (CFLEVEL15),
RMF provides additional data in the Monitor III Coupling Facility
reports and the Postprocessor Coupling Facility Activity report. For
example, the reports display coupling facility utilization per
structure and whether dynamic coupling facility dispatching is turned
on. In addition, RMF provides new overview conditions for the
Postprocessor based on SMF record 74-4.

When change was introduced: Integrated in z/OS V1R9 and available in
z/OS V1R8 with APAR OA17070.

http://www-01.ibm.com/support/docview.wss?uid=isg1OA17070

Repeats the info in the first post.

---

Also, my question, zXXX hardware would use Assist processors before CP
to drive the cost down.  Is this prioritization implemented in
Hercules, or does it JUST RUN without considering the workload?

On Tue, Jun 23, 2009 at 11:01 AM, Fish <fish@...> wrote:

>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> David Kreiss wrote:
>
> > A DM=CPU should tell the story.
> [...]
> > If the D M=CPU under Hercules doesn't show these as the
> > correct processor types the results would be as seen.
>
> And therein is the problem: the D M=CPU display does show the "+I"
> for processor 7, but yet their total chargeable time isn't any less.
> Their total time is the same whether a zIIP is defined in Hercules or
> not.
>
> On a real machine they're different. On a real machine with a zIIP
> the chargeable time ("SRB time" I think they called it?) is very
> small, leading them to confidently conclude their program/workload is
> indeed running on the zIIP, but on Hercules, whether a zIIP is define
> or not makes no difference: the chargeable time is always the full
> amount! (even though as I explained it appears as if the o/s does
> know the zIIP exists since it's appearing properly in the D M=CPU
> display).
>
> The big question is WHY this behavior occurs? What is Hercules doing
> wrong?
>
> Or are they doing something wrong? Do they need to configure
> something in the o/s that they possibly forgot to do?
>
> Where should I start looking?
>
> TIA
>
> - --
> "Fish" (David B. Trout) - fish@...

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


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

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[fish-8]

 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Harold Grovesteen wrote:

[...]
> [...] What is ambiguous from your post is whether all of
> the work got charged to a standard CPU or all of the work
> got charged to the zIIP.

Um... How can you tell?  (z/OS ain't my thang)


> Either way, this is different than the 12 seconds identified
> in your message. A guess would be that whatever values have
> been presented by Hercules to z/OS

And what values would those be?? That's one of the things I'd like to
know! I believe knowing what these values were (STSI? Which function
code? Some DIAG information? Which one? Etc..) would be a BIG CLUE in
locating where Hercules is going wrong.


> has caused z/OS to dispatch the work differently than on a
> real mainframe. Getting these numbers to match what the real
> mainframe sees would require the same numbers.

Depending on WHAT these numbers where (i.e. not their values but
WHERE (i.e. which instruction(s)) they're coming from) it's a little
premature I think to presume they have to match exactly. I think we
need to determine WHAT these numbers are first (i.e. WHERE they're
coming from). Only then can we decide whether they need to be the
same. Does that make sense? Did I say that right?


> Maybe someone else knows whether these are static or change.
> If they are static we have a shot at getting them to be close.

(Oops. Should have kept reading..) Yes, I believe you're on the right
track: we need to determine WHAT these values are (i.e. WHERE they're
coming from) so we can then determine whether or not they're static
or dynamic.

If they're static then it's easy: we just have to match the same
values a real machine produces.

If they're dynamic however, then we need to determine the proper way
they should be calculated and then double check our calculations to
make sure they're being calculated correctly.

But first things first: how and the heck does one determine which of
the numbers Hercules is providing are the ones involved in this
issue??

TIA

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSkEEUkj11/TE7j4qEQJ+7gCgkk7biFZw1ZDZ+Y31hwysOOYp9KAAoKTq
v+I02Nspv+39pBDe+mu1Ys4g
=R9il
-----END PGP SIGNATURE-----

[davekreiss]

> Also, my question, zXXX hardware would use Assist processors before CP
> to drive the cost down.  Is this prioritization implemented in
> Hercules, or does it JUST RUN without considering the workload?

Work gets dispatched on a zIIP or zAAP because IBM has implemented code
to cause the dispatcher to signify that the work is eligible so this is
a purely z/OS implementation.  z/OS handles all the dispatching to these
specialty processors or may dispatch to regular processors if the
specialty processor(s) are over utilized.

Dave

[davekreiss]

> (Oops. Should have kept reading..) Yes, I believe you're on the right
> track: we need to determine WHAT these values are (i.e. WHERE they're
> coming from) so we can then determine whether or not they're static
> or dynamic.
>
> If they're static then it's easy: we just have to match the same
> values a real machine produces.
>
> If they're dynamic however, then we need to determine the proper way
> they should be calculated and then double check our calculations to
> make sure they're being calculated correctly.
>
> But first things first: how and the heck does one determine which of
> the numbers Hercules is providing are the ones involved in this
> issue??

 

I suspect the numbers that are being used are from the step termination
messages here's an example:

IEF374I STEP/xxxxxxxx/STOP  2009174.1152 CPU    4MIN 29.98SEC SRB
0MIN 09.50SEC

The best way is to look at SMF data.  That is an exercise you probably
won't under take since you aren't a  z/OS expert.

Here are some z/OS parameters which also may be important.  You can have
them looked up in the z/OS Init and Tuning Guide.

IIPHONORPRIORITY= YES|NO

PROJECTCPU=YES|NO  (Should be YES but defaults to YES if a zIIP is
present at IPL)

ZIIPAWT=xxxxx

 

If you wish you can have the user extract type 30 records to an AWS tape
and I can examine the SMF data for the specific job in question though I
suspect this isn't feasible.  There are other utilities which look at
SMF data but I haven't used them - DAF comes to mind on the CB Tape web
site.

 

All of this is assuming that the problem is with z/OS.  If there is no
zIIP time in SMF then my guess is this is a Hercules problem.

 

Dave





[Non-text portions of this message have been removed]

[Roger Bowler]

--- In hercules-390@..., Fish wrote:

> And therein is the problem: the D M=CPU display does show the "+I"
> for processor 7, but yet their total chargeable time isn't any less.
> Their total time is the same whether a zIIP is defined in Hercules
> or not.
>
> On a real machine they're different. On a real machine with a zIIP
> the chargeable time ("SRB time" I think they called it?) is very
> small, leading them to confidently conclude their program/workload
> is indeed running on the zIIP, but on Hercules, whether a zIIP is
> define or not makes no difference: the chargeable time is always
> the full amount! (even though as I explained it appears as if the
> o/s does know the zIIP exists since it's appearing properly in the
> D M=CPU display).
>
> The big question is WHY this behavior occurs? What is Hercules
> doing wrong?
>
> Or are they doing something wrong? Do they need to configure
> something in the o/s that they possibly forgot to do?
>
> Where should I start looking?

One theory that springs to mind is based on the fact that real zIIPs always run at maximum processor speed, as opposed to CPs which are restricted to run at a lower speed depending on the z9 or z10 model you chose (and hence the price you paid).

Whereas of course Hercules runs all zIIPs and CPs at maximum speed.

My theory (and it's just a guess) is that when z/OS sees that the CPs run at the same speed as the zIIPs, it may figure that it isn't worth the task-switching overhead of moving the SRB's over to the zIIP while there is still spare capacity on the CPs.

There are two ways you could test this theory:
1. Run a looper job which will soak up the capacity of the CPs, thus making it worthwhile for z/OS to schedule your SRB work on the zIIP
2. Modify hercules stsi instruction so that it fictitiously reports that the zIIPs run at a higher speed than the CPs

Regards,
Roger Bowler
Hercules "Now I *can* believe it's not a mainframe"

[Norman Hollander on DesertWiz]

From a z/OS perspective, zIIP time (eligible or actual) is included in the
CPU time.
If you are running a subcapacity model (z9 or z10), the zIIP portion of the
CPU time
is normalized to account for the difference in the processor speed.
 
From: hercules-390@... [mailto:hercules-390@...] On
Behalf Of Roger Bowler
Sent: Thursday, June 25, 2009 SYSN 04:10 AM
To: hercules-390@...
Subject: [hercules-390] Re: zIIP processor (ENGINE type IP)
 




--- In hercules-390@... <mailto:hercules-390%40yahoogroups.com>
, Fish wrote:

> And therein is the problem: the D M=CPU display does show the "+I"
> for processor 7, but yet their total chargeable time isn't any less.
> Their total time is the same whether a zIIP is defined in Hercules
> or not.
>
> On a real machine they're different. On a real machine with a zIIP
> the chargeable time ("SRB time" I think they called it?) is very
> small, leading them to confidently conclude their program/workload
> is indeed running on the zIIP, but on Hercules, whether a zIIP is
> define or not makes no difference: the chargeable time is always
> the full amount! (even though as I explained it appears as if the
> o/s does know the zIIP exists since it's appearing properly in the
> D M=CPU display).
>
> The big question is WHY this behavior occurs? What is Hercules
> doing wrong?
>
> Or are they doing something wrong? Do they need to configure
> something in the o/s that they possibly forgot to do?
>
> Where should I start looking?

One theory that springs to mind is based on the fact that real zIIPs always
run at maximum processor speed, as opposed to CPs which are restricted to
run at a lower speed depending on the z9 or z10 model you chose (and hence
the price you paid).

Whereas of course Hercules runs all zIIPs and CPs at maximum speed.

My theory (and it's just a guess) is that when z/OS sees that the CPs run at
the same speed as the zIIPs, it may figure that it isn't worth the
task-switching overhead of moving the SRB's over to the zIIP while there is
still spare capacity on the CPs.

There are two ways you could test this theory:
1. Run a looper job which will soak up the capacity of the CPs, thus making
it worthwhile for z/OS to schedule your SRB work on the zIIP
2. Modify hercules stsi instruction so that it fictitiously reports that the
zIIPs run at a higher speed than the CPs

Regards,
Roger Bowler
Hercules "Now I *can* believe it's not a mainframe"



[Non-text portions of this message have been removed]