TM1 Forum

Hello,

I have a small problem with data spread. When I want to do a proportional spread. The time of spread is very long. Sometimes I must wait a 3-5 minutes. Do you know , how solve this problem ?

I looked for information about it on IBM site but I don't find a particular answer.

Thank you

Best regards
Kamil

Other than the speed of the CPU on the server running the TM1 instance the thing affecting the time to perform a data spread is the number of leaf cells which need to be updated.

Although to do a proportional spread you may be keying a value over just one cell. If the elements at the coordinates of that cell are highly consolidated on several dimensions then there may potentially be not just hundreds or thousands of leaf cells to update but millions of cells. (Remember that in TM1 all numeric values are held in leaf cells.) Spreading is just a shortcut to leaf data entry.

So the (only) way to speed this up is to perform the spread at a less highly consolidated level. (and be careful nut to spread over an intersection that contains billions of cells).

Added to this, look at feeders too. The leaf level cells receiving the data input might very well send feeders to potentially many other cells. If that’s the case you can notice a slower performance.

To add on to Lotsarams comments.

I have found some clients to have thousands and thousands of very small values. The clients complain about performance with exports, calculations and inputs. After reviewing I often find tiny values, for example 0.00000008. These are invisible to the user due to formating. But the system uses them for proportional spreads, feeding etc.

So you could also check your model for tiny values that should really be zero.