How to calculate top 10 without running out of memory

[martingaleh]

I have 1 table with only 116k rows (Not 116M) and its broken down by customer |sales| invoice number
I want a table that aggregates sales by customer and I want it to just show the top 10 customers. Here's my dax:
sAmt = sum(sales)
calculate([sAmt],filter(table[customer],RANKX(all(table[customer]),abs([sAmt]))<=10))
which failed miserably because it filtered nothing and gave me every amount.

Then I tried filter(all(table[customer]),rankx(all(table[customer]),abs]),abs([sAmt]))<=10))

which filtered nothing

but when I tried addcolumns(all(table[customer]),"rank",rankx(all(table[customer]),abs([sAmt])))

The rank is right. Maybe the filter context inside filter is different from the one with addcolumns

A suggested syntax here:
https://powerpivotpro.com/2014/11/displaying-top-n-bottom-n-and-all-others/

would be
filter(table,rankx(all(table[customer]),abs]),abs([sAmt]))<=10))

but filter would iterate on every single row in the table which is 116k. Plus I thought calculate works in that filter provides a table of 1 column of just the customers I want and then calculate calculates the value of just those customers

 

[martingaleh]

Never mind, I think I figured it out. filter(all(kimurasan[company]),calculate(rankx(all(kimurasan[company]),abs([sAmtKimurasanInvoicesMinuslatest])))<10)
Holy god, this is a hard language. At least in javascript, you can look through the scope chain to determine the scope. It's almost impossible to figure out the filter context and which fucntions change what filter context

oh and, of course, if you're interested in a table where company is the row filter, you have to keep it:
calculate([sAmt],keepfilters(filter(table[customer],RANKX(all(table[customer]),abs([sAmt]))<=10)))

can you image tableau with subquery language this complicated?