Procedure Process Speed

I have a process in a Procedure that puts a 1 into a cube where the outlet has data. This procedure is running very slow and I am trying to make it faster. I was previously using an if statement to say if that data in the first cube for the required date range then 1 else 0. This was very slow. I have created a cube that contains only the data that needs the 1. Therefore the if statement is no longer required. When I use the algorythm a=1, I would expect to see a 1 next to all detail from the cube, but I don't. How can I achieve this without the If statement. See attached is the process with the if statement. Cube A contains only the required data and I want to put a 1 into cube B, thanks.

Find more posts tagged with

Accepted answers

All comments

Alexander Kappes

Dear Louise Just,

thanks for your request.

According to your provided information it is very hard to give some suggestions.

Does your new Cube (which consists the needed Dimensions) has any sparsity? Did you check, if the combinations are acitve? Because DataFlows from scratch didn´t create any new combinations.

Using the origin status (with the if). did you think about pre selecting all dimensions according your source cube "ACases ...". For that case it could be perhaps a solution when you have selected to change your if statement through b = a / a.

In that case you got also the value 1 as result and for the case that a can´t be zero the dataflow will perform in a quick manner.

Hope it helps

regards

Alexander Kappes

Louise Just

Hi Alexander,

Thanks for your feedback, let me try b=a/a, I feel that will work. Sometimes we are just too close and can't see the obvious.

Regards

Louise

Alexander Kappes

Hi Louise Just,

but please be aware that a isn´t allowed to be zero for that case.

Division by zero isn´t defined in mathematics and also not in BOARD :-)

regards

Alexander Kappes

Daniele Di Lorenzo

I commonly use the approach b = a/a in data flow to obtain a flag cube (b) with 1 based on existing values in cube (a)

the 2 cubes should have same structure so data flow runs in HBMP mode and is fast.

then, narrowing the selection as much as possible increases performances further

Louise Just

Thanks everyone for your assistance, the a/a worked perfectly and very fast too.