Thank you. I was able to modify the polling period and get the change I desired. Below is some investigative/entertaining information if it helps anyone else.
With the default polling period of 2000 ms (2 sec), I placed my cursor at each side of a single grid box and measured the width to be 16 sec. I zoomed in on a particularly transient section of the CPU package power, and counted the ridges noticing that there were indeed 8 sample points spanning the single grid box. I changed the polling period to 8 sec, expecting to see only 2 data points per box assuming that the x-axis is not scalable in any way. But interestingly, the software seems to want to ensure 8 sample points to a single grid box, and this is what allows the x-axis scale to change. I'm assuming it has something to do with 8 bits per byte. With the 8 sec polling period, each box is roughly 64 sec wide.
Anyway I changed the polling period to a sub-second amount to get the smaller scale I desired. Interestingly, I noticed some pronounced sawtooth patterning in the CPU temperature that I didn't notice originally, an effect which was probably masked before by the slower polling frequency. Thanks again.