The article has a subhead of "Total GC pause time", which is right (based on my skim of the code).
The subtitle below that says "Average total garbage-collection pause time during the measurement window", which seems to imply that 100ms is the average STW time. I don't think that's right.
My suggestion is to drop "Average" from that (even though I think I understand what you're saying, its averaging across multiple runs).
Alternatively, consider running the binary with GODEBUG=gctrace=1, and then separately computing the average STW time, and present that graph first as "Average GC Stop-the-World time", and the current graph as "Total GC Stop-the-World time".
The article has a subhead of "Total GC pause time", which is right (based on my skim of the code).
The subtitle below that says "Average total garbage-collection pause time during the measurement window", which seems to imply that 100ms is the average STW time. I don't think that's right.
My suggestion is to drop "Average" from that (even though I think I understand what you're saying, its averaging across multiple runs).
Alternatively, consider running the binary with
GODEBUG=gctrace=1, and then separately computing the average STW time, and present that graph first as "Average GC Stop-the-World time", and the current graph as "Total GC Stop-the-World time".