We would like to do some comparison between different processors on different servers.

Would you have any sample SQL scripts to run a stress test and provide results for comparision?

Thanks,

Paurav

Thank-you for bringing to light such a nice, yet simple means of studying the “CPU pressure”.

Many-a-times, I have come across quite a few friends and colleagues who always say – “Server performance issue? Bump-up the RAM to double the present amount”. I must add here that we use a lot of virtualization for our development and RND servers, but ultimately, there is a limit and difference in the amount of juice a server can have and what is really necessary.

Often the case is that a bad query is the root cause of your CPU spiking up for long durations of time. – which ultimately ends up frying the board. That should be the real focus of attention rather than spending thousands on hardware upgrades.

After all, if the space shuttle can still run on 8086 processors – how much computing power do we really need?

Another point where most are caught unaware is the difference between “Multi-core CPU” and “Multi-CPU” systems. While the former acts like multiple CPUs, it ultimately is probably one single physical CPU – that would reduce the number of schedulers, and lead to a bottle neck in a BI-like scenario where CPU loads are often high.

Your article is a step forward in bringing awareness about these subtle differences, and the fact that not everything has the same solution.

Looking forward to many such articles in the future.

Trivial point, but the phrase "expected results" has always had a pretty specific meaning for SQL Server and it is not expected performance (as in the quote) but whether the results of a query are correct.