Portal | Manuals | References | Downloads | Info | Programs | JCLs | Master the Mainframes
IBM Mainframe Computers Forums Index
 
Register
 
IBM Mainframe Computers Forums Index Mainframe: Search IBM Mainframe Forum: FAQ Memberlist Usergroups Profile Log in to check your private messages Log in
 

 

MIPS reduction for Batch job

 
Post new topic   This topic is locked: you cannot edit posts or make replies.    IBMMAINFRAMES.com Support Forums -> All Other Mainframe Topics
View previous topic :: :: View next topic  
Author Message
Virendra Shambharkar

New User


Joined: 26 Aug 2015
Posts: 28
Location: India

PostPosted: Mon Nov 07, 2016 4:02 pm    Post subject: MIPS reduction for Batch job
Reply with quote

Hi,

If we reduce the CPU time and run time by the same does the % of MIPS consumed by the job remain the same?

Say a Job runs for 10 hours and take 1 hour of CPU and we reduce the job run time to 5 hours and CPU time to 0.5 hours the MIPS consumed by the Job are the same?

Any information on this will be of great help.

Thanks in advance.

Virendra
Back to top
View user's profile Send private message

Robert Sample

Global Moderator


Joined: 06 Jun 2008
Posts: 7913
Location: Bellevue, IA

PostPosted: Mon Nov 07, 2016 6:04 pm    Post subject:
Reply with quote

Quote:
If we reduce the CPU time and run time by the same does the % of MIPS consumed by the job remain the same?
No, because:
1. MIPS is a measurement of MACHINE capacity and hence cannot be applied to individual jobs.
2. Since MIPS is a measurement of MACHINE capacity, the ONLY way to change the MIPS is to change the machine -- get a larger machine for more, get a smaller machine for less.
3. MIPS is a concept that really makes very little sense these days -- how do you apply million instructions per second (which is what MIPS stands for) to a z13 with 140 processors, some of which are CP, some are zIIP, some are SAP, some are IFL, some are ICF?

For your example, 10 hours is 36,000 seconds and 1 hour of CPU is 3,600 seconds while 5 hours is 18,000 seconds and 1/2 hour of CPU is 1,800 seconds. You have reduced overall CPU consumption of the job by half, but the workload on the system is, most likely, the same since the job still needs 10% of a processor for the elapsed time.
Back to top
View user's profile Send private message
Rohit Umarjikar

Senior Member


Joined: 21 Sep 2010
Posts: 1610
Location: NY,USA

PostPosted: Mon Nov 07, 2016 7:39 pm    Post subject:
Reply with quote

I couldn't wait to recall this. Please take a look
http://ibmmainframes.com/viewtopic.php?t=60895&highlight=
Back to top
View user's profile Send private message
steve-myers

Active User


Joined: 30 Nov 2013
Posts: 461
Location: The Universe

PostPosted: Mon Nov 07, 2016 10:23 pm    Post subject:
Reply with quote

As the link in Mr. Umarjikar's post explains "MIPS" for a single job is sort of meaningless. There are two, and only two, ways to alter this. Neither one is very realistic.
  • Change to a machine that can execute the instructions in your program at a different rate.
  • Change your program to use instructions that execute at a different rate. By definition that will change the program's "MIPS."
Back to top
View user's profile Send private message
Robert Sample

Global Moderator


Joined: 06 Jun 2008
Posts: 7913
Location: Bellevue, IA

PostPosted: Mon Nov 07, 2016 11:12 pm    Post subject:
Reply with quote

Note that 1 MSU = 6 MIPS is specific to a machine type; the ratio varies by machine (IIRC, the z13 is more like 1 MSU = 8 MIPS). Furthermore, Wikipedia has some good comments about the practicality of MSU:
Quote:
A million service units (MSU) is a measurement of the amount of processing work a computer can perform in one hour. The term is most commonly associated with IBM mainframes. It reflects how IBM rates the machine in terms of charging capacity. The technical measure of processing power on IBM mainframes, however, are Service Units per second (or SU/sec).

One “service unit” originally related to an actual hardware performance measurement (a specific model’s instruction performance). However, that relationship disappeared many years ago as hardware and software evolved. MSUs are now like other common (but physically imprecise) measurements, such as “cans of coffee” or “tubes of toothpaste.” (Cans and tubes can vary in physical size depending on brand, market, and other factors. Some coffee cans contain 500 grams and others 13 ounces, for example.)

Most mainframe software vendors use a licensing and pricing model in which the customers are charged per MSU consumed (i.e. the number of operations the software has performed) in addition to hardware and software installation costs.[1] Others charge by total MSU system capacity. Thus, while MSU is an artificial construction, it does have a direct financial implication. In fact, software charges are why the MSU measurement exists at all.
Service units are maintained for CPU, I/O, SRB, and memory (storage) -- so 1 MSU may not relate very closely to CPU usage.

Yes, the term "MIPS" gets used a lot -- and most of the time, the usage is incorrect. The number is easy to understand, even though it has almost no meaning these days.
Back to top
View user's profile Send private message
vasanthz

Global Moderator


Joined: 28 Aug 2007
Posts: 1445
Location: Azeroth

PostPosted: Tue Nov 08, 2016 12:25 am    Post subject:
Reply with quote

Code:
One MSU = 6 Mips (here MSU and Mips represent a consumption speed, not an accumulated consumption). So a job that constantly uses 10 MSU (as displayed by monitors) uses 60 Mips.

I think this information is too generic and wrong. Every machine has its own soft MSU to MIPS conversion. Cheryl Watson CPU charts are reliable for this information.

And for the original question,
If you are trying to justify your resource saving effort to management. Then just go by saying your effort reduced CPU time & elapsed time and the service units the job consumed.
Back to top
View user's profile Send private message
Rohit Umarjikar

Senior Member


Joined: 21 Sep 2010
Posts: 1610
Location: NY,USA

PostPosted: Tue Nov 08, 2016 3:37 am    Post subject:
Reply with quote

Quote:
I think this information is too generic and wrong.
On what basis you say it wrong? It concur Robert's statement, This is the stats given by Site support which I cut down to what is needed but has much more, so TS may have one at site, He has to get talking to people. These machines we have for few systems
Code:
2964-708   z13   8.3
 2964-709   z13   8.3

in general.
Code:
Model   Common Name   MIPS/SW MSU
2964-701   z13   8.1
2964-702   z13   8.1
2964-703   z13   8.1
2964-704   z13   8.2
2964-705   z13   8.2
2964-706   z13   8.2
2964-707   z13   8.2
2964-708   z13   8.3
2964-709   z13   8.3
2964-710   z13   8.3
2964-711   z13   8.3
2964-712   z13   8.3
2964-713   z13   8.3
2964-714   z13   8.3
2964-715   z13   8.3
2964-716   z13   8.3
2964-717   z13   8.3
2964-718   z13   8.4
2964-719   z13   8.4
2964-720   z13   8.4
2964-721   z13   8.4
2964-722   z13   8.4
2964-723   z13   8.4
2964-724   z13   8.5
2964-725   z13   8.5
2964-726   z13   8.5
2964-727   z13   8.5
2964-728   z13   8.5
2964-729   z13   8.5
2964-730   z13   8.5
2964-731   z13   8.5
2964-732   z13   8.5
2964-733   z13   8.5
2964-734   z13   8.5
2964-735   z13   8.5
2964-736   z13   8.5
2964-737   z13   8.5
2964-738   z13   8.5
2964-739   z13   8.5
2964-740   z13   8.5
2964-741   z13   8.5
2964-742   z13   8.5
2964-743   z13   8.5
2964-744   z13   8.5
2964-745   z13   8.5
2964-746   z13   8.5
2964-747   z13   8.5
2964-748   z13   8.5
2964-749   z13   8.5
2964-750   z13   8.5
2964-751   z13   8.5
2964-752   z13   8.5
2964-753   z13   8.5
2964-754   z13   8.5
2964-755   z13   8.5
2964-756   z13   8.5
2964-757   z13   8.5
2964-758   z13   8.5
2964-759   z13   8.5
2964-760   z13   8.5
2964-761   z13   8.5
2964-762   z13   8.5
2964-763   z13   8.5
2964-764   z13   8.5
2964-765   z13   8.5
2964-766   z13   8.5
2964-767   z13   8.5
2964-768   z13   8.5
2964-769   z13   8.5
2964-770   z13   8.5
2964-771   z13   8.5
2964-772   z13   8.5
2964-773   z13   8.5
2964-774   z13   8.5
2964-775   z13   8.5
2964-776   z13   8.5
2964-777   z13   8.5
2964-778   z13   8.5
2964-779   z13   8.5
2964-780   z13   8.5
2964-781   z13   8.5
2964-782   z13   8.5
2964-783   z13   8.5
2964-784   z13   8.5
2964-785   z13   8.5
2964-786   z13   8.5
2964-787   z13   8.5
2964-788   z13   8.5
2964-789   z13   8.5
2964-790   z13   8.5
2964-791   z13   8.5
2964-792   z13   8.5
2964-793   z13   8.5
2964-794   z13   8.5
2964-795   z13   8.5
2964-796   z13   8.5
2964-797   z13   8.5
2964-798   z13   8.5
2964-799   z13   8.5
2964-7A0   z13   8.5
2964-7A1   z13   8.5
2964-7A2   z13   8.5
2964-7A3   z13   8.5
2964-7A4   z13   8.5
2964-7A5   z13   8.5
2964-7A6   z13   8.5
2964-7A7   z13   8.5
2964-7A8   z13   8.5
2964-7A9   z13   8.5
2964-7B0   z13   8.5
2964-7B1   z13   8.5
2964-7B2   z13   8.5
2964-7B3   z13   8.5
2964-7B4   z13   8.5
2964-7B5   z13   8.5
2964-7B6   z13   8.5
2964-7B7   z13   8.5
2964-7B8   z13   8.5
2964-7B9   z13   8.5
2964-7C0   z13   8.5
2964-7C1   z13   8.5
2964-7C2   z13   8.5
2964-7C3   z13   8.5
2964-7C4   z13   8.5
2964-7C5   z13   8.5
2964-7C6   z13   8.5
2964-7C7   z13   8.5
2964-7C8   z13   8.5
2964-7C9   z13   8.5
2964-7D0   z13   8.5
2964-7D1   z13   8.5
2964-7D2   z13   8.5
2964-7D3   z13   8.5
2964-7D4   z13   8.5
2964-7D5   z13   8.5
2964-7D6   z13   8.5
2964-7D7   z13   8.5
2964-7D8   z13   8.5
2964-7D9   z13   8.5
2964-7E0   z13   8.5
2964-7E1   z13   8.5


Quote:
If you are trying to justify your resource saving effort to management. Then just go by saying your effort reduced CPU time & elapsed time and the service units the job consumed
What's the harm if TS wants to know MIPS?
Back to top
View user's profile Send private message
vasanthz

Global Moderator


Joined: 28 Aug 2007
Posts: 1445
Location: Azeroth

PostPosted: Tue Nov 08, 2016 6:39 am    Post subject:
Reply with quote

Quote:
What's the harm if TS wants to know MIPS?

Job cannot be measured in MIPS. MIPS is the speed of the machine.
I am tired.. I feel like keyboard warrior battling a Trump supporter
36_2_18.gif
Back to top
View user's profile Send private message
Rohit Umarjikar

Senior Member


Joined: 21 Sep 2010
Posts: 1610
Location: NY,USA

PostPosted: Wed Nov 09, 2016 12:31 am    Post subject:
Reply with quote

Quote:
Job cannot be measured in MIPS. MIPS is the speed of the machine.
You don't seems to understand the question asked. MIPS changes per model, that everyone knows by now but TS wants to know the relation between CPU and MIPS how they are related to each other.
This is the calculation at my site and this is directly coming from the same set of people ( Capacity Planner/ Performance Management Group) that you claim to be part of.
The calculation is ( this is just an example):
Code:
Cpu seconds/ (#cpus * 86400)   * totalmips
For system XXX(masked)
Cpu seconds / (9*86400) * 12371
If your job ran 60 minutes of cpu (3600 seconds) it is:
(3600 / 777600) * 12371  = 57.27 mips


I don't think any further discussion is required as TS has what is required.
Back to top
View user's profile Send private message
View previous topic :: :: View next topic  
Post new topic   This topic is locked: you cannot edit posts or make replies.    IBMMAINFRAMES.com Support Forums -> All Other Mainframe Topics All times are GMT + 6 Hours
Page 1 of 1

 

Search our Forum:

Similar Topics
Topic Author Forum Replies Posted
No new posts How does a called pgm know if its cal... Graeme Westerman COBOL Programming 4 Tue Nov 29, 2016 9:25 pm
No new posts DB2 Streaming Batch Processing Problem Manshadi DB2 4 Sat Sep 24, 2016 12:14 pm
No new posts How to show message using batch jobs madprasy TSO/ISPF 7 Fri Sep 09, 2016 6:55 pm
No new posts Don't allow CICS to submit batch jobs prino JCL & VSAM 5 Mon Jul 25, 2016 3:26 pm
No new posts Using Field names in FILEAID Batch CO... Atul Banke Compuware & Other Tools 2 Wed Jun 29, 2016 9:54 pm


Facebook
Back to Top
 
Mainframe Wiki | Forum Rules | Bookmarks | Subscriptions | FAQ | Tutorials | Contact Us