Howdy!
I use PowerBI desktop often for data shaping (for machine learning projects). My data science friends are always blown away at what Power BI can easily do.
2 questions around performance:
Question 1:
I am getting a new personal desktop. My old one has an Intel i5 6400 quad core 2.70GHz with 8GB RAM. My new desktop has an Intel i5 8600k with 6 cores 3.6GHz with 64GB RAM.
Will I notice much of a performance improvement for "resource intensive" tasks with the new desktop? ... or was the old desktop sufficient enough and the new desktop won't make much performance difference?
Question 2:
I have some data manipulation work I am doing. There are 2 tables in the mix. The primary data table has ~200 columns and 9,000 rows. You can think of the second table as a "custom calendar" type of deal. That table is there for moving average usage. This is the "lookup" table I reference when implementing Rob Collie's "Greatest Formula In The World" approach for moving averages and moving sums. The secondary table also has 9,000 rows. All those rows are needed and I am not able to skinny that "custom calendar" table down.
I did a bunch of prep work to shape the data. Then I had created 195 measures using Collie's "Greatest" formual to get rolling values. I added all 195 measures to a table visualization on a Power BI tab. I added a single filter so I can run my output in "chunks". For 1/3rd of the data, it takes 60 minutes for the table visualization to populate and then another 15 minutes for me to export the output to a csv file. I need to do this 3 times to get all of my data exported.
I have read that the Filter function can really impact performance (I *think* it's because it limits what can be processed in parallel).
Are there any recommendations on how I might approach this differently to speed up performance? As an FYI, this is not a user-facing tool. It is NOT for users. This is me using PowerBI to quickly do some kick-butt data manipulation that in other tools would be crazy difficult to pull off.
Thanks!
