An odd thing about jitter, and a question

I'm running latency-histogram with nothing else going on. I realize lots should be going on to figure out usable software stepping latency numbers, but I found something odd and wanted to simplify.
I ran these two commands:
latency-histogram --sbins 1000
latency-histogram --nobase --sbins 1000
The second result quickly diverged from the first servo results, and has a different shape:
-14.1 to 13.0
-22.0 to 18.1
It looks to me like there is a difference in the two servo measurements. This is odd.

A question: If I'm using software stepping, which one matters, base or servo (or both)?

Yes. What's holding me up is having blown up my parallel port :<
So I'm getting lost in details while waiting for the card (well, I bought two of them) to show up.
For other people's reference, the Dell Sunix 0GP385 cards work fine with LinuxCNC and are readily available on ebay.