Comments on: SQL SERVER – Stream Aggregate Showplan Operator – Reason of Compute Scalar before Stream Aggregate

By: SQL SERVER – Weekly Series – Memory Lane – #015 « SQL Server Journey with SQL Authority

Sat, 09 Feb 2013 01:30:50 +0000

[...] Stream Aggregate Showplan Operator – Reason of Compute Scalar before Stream Aggregate Stream Aggregate is a quite commonly encountered showplan operator. I have often found it in very simple COUNT(*) operation’s execution plan. The Stream Aggregate operator group rows by one or more columns and then calculate one or more aggregate expression returned by the query. [...]

By: clicking here

clicking here — Thu, 07 Feb 2013 15:01:40 +0000

Greetings! Very useful advice within this post! It’s the little changes that make the most significant changes. Many thanks for sharing!

By: Johnny

Johnny — Thu, 18 Oct 2012 21:05:03 +0000

Wonderful blog! I found it while surfing around on Yahoo News.
Do you have any suggestions on how to get listed in Yahoo News?

I’ve been trying for a while but I never seem to get there! Thank you

By: Syed

Syed — Thu, 23 Feb 2012 02:45:11 +0000

Great Article. Complex topics like this are understand easily by your explanation.Thanks a lot.

By: Ashish Kadam

Ashish Kadam — Mon, 07 Nov 2011 10:54:59 +0000

hi pinal,
If both the operations cost is same then which types operation we used for query writing? if both are same why there are 2 different operations given?

By: Marko Parkkola

Marko Parkkola — Sun, 07 Feb 2010 13:15:29 +0000

Well, conversion from larger datatype (BIGINT – 64 bit) to smaller type (INT – 32 bit) does not take time at all. Other way around it could take couple of CPU cycles, depending how it is done.

I think what happens is that COUNT() and COUNT_BIG() are calculated into 64 bit variable and then type casted into 32 bit variable in the case of COUNT(). There is no memory allocation, copying or anything happening, just type casting. You just read the last 32 bit out of 64 bit wide memory space and leave the first 32 bit unnoticed. This takes zero CPU cycles.

(Well, not “last” 32 bit since memory isn’t ordered from left to right in the way one would think)

By: Brian Tkatch

Brian Tkatch — Fri, 05 Feb 2010 14:34:11 +0000

Very nice Pinal, thanx!