TM1 Forum

Hi, All

We have been performance monitoring issue from PAX users. Back end we monitor several time that User view retrieval consumes highest CPU and more memory allocations (This is happening since a week). Environment - 11.02.05 planning analytics

MTQ:

There are two instance on the TM1 server. Finance (having MTQ=6) and Accounting (MTQ=8) and Total Cores on the server (16).

IBM articulate following on MTQ Configuration:

1) Sum of the cores assigned to each TM1 instance should not be more then Total Cores on the server
2) MTQ = -1 means T=16-1+1=16=All cores will be utilized


MTQ =6 ; Does it mean T=16+6+1=23? and MTQ=8; Does it mean T= 16+8+1=25? (If this is correct, what would happen when total assigned MTQ cores more than server cores?) - Any performance impact?

In order to assign 6 cores to be utilized by Finance instance, MTQ=-11? and In order to assign 8 cores to be utilized by Accounting instance, MTQ=-9?

Moreover, I would like to have detailed knowledge on CacheFriendlyMalloc configuration parameter and its impact on Performance where User group is large.


Please feel free if further information is required.

Thank You

Dharav

dharav9 wrote: ↑Thu Jan 10, 2019 6:24 pm MTQ = -1 means T=16-1+1=16=All cores will be utilized

MTQ =6 ; Does it mean T=16+6+1=23? and MTQ=8; Does it mean T= 16+8+1=25? (If this is correct, what would happen when total assigned MTQ cores more than server cores?) - Any performance impact?

No!
Although the concept is confusing I do think the documentation on this is reasonably clear.
MTQ=0 disables MTQ
therefore ...
MTQ=-1 (or MTQ=ALL) is used to set using all available CPU cores
MTQ=-2 is therefore use all available cores less 1, MTQ=-3 all cores less 2, etc.

Otherwise it's quite straightforward
MTQ=2 use 2 cores
MTQ=3 use 3 cores
etc.

No you can't configure more cores than what is available.

Thank You Lotsaram.

Have you ever com across with following configuration parameter?

CacheFriendlyMalloc configuration parameter ? : its impacts on Performance where User group is large (Standard IBM Note)