Yeah, that jumped out at me, too.

If they can’t re-run the benchmarks on the same hardware, it’s hard to compare the numbers.