Recognizing that this post is nine years old, I still feel compelled to answer since I don't think anyone answered the exact question, how do you compare the execution time of the two queries?

There are a bunch of ways, all free, all built in to SQL Server (2012 or greater for the majority, 2016 or greater for Query Store):

• Include Client Statistics (only allows for 10 measures)

• Connection Properties

• SET STATISTICS TIME

• QueryTimeStats in an actual execution plan

• QueryStore

• Dynamic Management Views

• Trace Events (batch completed)

• Extended Events (batch completed)

The last two actually include a whole slew of events that can capture query metrics, but these are all the ways to measure actual query performance. In this blog post, I compare all of them to attempt to assess their accuracy. Nightmare is, they're all a little different, but the principal exception is using "Include Client Statistics" It's just wildly inaccurate so I would not ever suggest using it.

For as much accuracy as possible, I will use Extended Events. I'll ensure that capturing the execution plan with runtime metrics (aka Actual Plan) is off (it negatively impacts performance measurement since it adds overhead). I execute each query 50 times (GO 50). I then use the Live Data Window in Extended Events to get an average of the execution times.

For other choices, and when to choose them, look to the blog post.