There isn’t much about testing methodology besides that Honeywell got around 5c improvement after 1h of continuous testing

Most of them are either demanding/good looking despite being old or just easier and faster to test because of built in benchmarks like F1 series to provide additional data. I agree that some choices are questionable especially when HUB benches RT performance, Spiderman remastered instead of newer Miles Morales or Far Cry 6 RT