The loading benchmark currently runs with 'cold' and 'warm' cache-temperature settings [1]. The way this works:

 . For 'cold' setting: the browser is started, the page is loaded once, and the benchmark measures the metrics from that.
 . For 'warm' setting: the browser is started, the page is loaded once [2], then re-loaded, and the benchmark measures the metrics for the second load.

It should be possible to measure the metrics for 'cold' setting for the first load in the 'warm' setting. Therefore, I believe if the benchmarks that run in both 'cold' and 'warm' settings are updated to measure the results for 'cold' from the first load in 'warm', that would reduce the runtime for these benchmarks quite a lot (maybe as much as ~50%?).

sullivan@, nednguyen@ for thoughts.
kouhei@, ksakamoto@ as owners for loading benchmark.
tdresser, vmiura@ fyi

[1] https://cs.chromium.org/chromium/src/tools/perf/page_sets/loading_desktop.py?l=25
[2] https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/page/cache_temperature.py?l=66