How is Your Parking? Foglight can Help.

Have you ever looked at Resource Monitor in Windows and wondered why some of your cpu cores were "parked"? That sure doesn't sound good!

This post does a good job at explaining the details. Needless to say I was intrigued, and I wanted to find out if Windows does a good job at managing core parking, or if manual intervention might be better.

I downloaded the Passmark Performance Test utility (free 30-day trial) and set it up to run 3 iterations of a "very long" test. The Foglight Windows collector was also set to gather data every 2 minutes during the test. 

At 12:50, I started 3 iterations of a "very long" Passmark cpu test with a 50% cpu core parking index.

At 13:05, I repeated the tests above. My Passmark cpu benchmark result was 6958. We'll call this "Test 1".

I let the system rest while I grabbed a bite to eat, then at 13:40, I turned cpu core parking off (manual intervention).

I then started 3 iterations of a "very long" Passmark cpu test and repeated. My Passmark cpu benchmark climbed a bit to 7022, for an almost 1% improvement. We'll call this "Test 2".

Looking at the graphs in Foglight, the cpu utilizations appear to be similar, maybe even a bit lower for "Test 2". So it appears we got a small improvement in the benchmark with approximately the same cpu utilization as measured from the OS. It would be interesting to go in-depth and measure the cpu GHz, power-states, voltages and temperatures while repeating these tests. I was expecting a much wider gap in performance between the automatic and manual modes of core parking but I can conclude that Windows 7 does a pretty good job at parking!

Anonymous