Presumably some of FHR's data correlation relies on the speed of a user's computer. e.g. we may have different expected values for startup times for 1 GHz machines than from 3 GHz machines. Also, a modern Core i7 is surely faster than a Pentium 4 running at around the same clock. We should consider submitting more CPU info in the FHR payload to facilitate meaningful correlation. Currently, we just collect MB of memory and number of CPU cores. First stop is Daniel for whether said data would be helpful. Next, we go off to Privacy folk.
Yes, CPU type and speed was one of the originally considered factors to study. We opted to start with just # of cores, but would eventually like to determine whether these other factors are useful. Maybe by including them in some future version of nightly/aurora and collecting enough data to determine whether they provide strong correlations.
FHR is going away per bug 1209088, wontfix.