I said this before and I'll say it again. Any benchmark that doesn't state why one candidate is faster than the other is not a benchmark. It's bullshit marketing material.
A benchmark that doesn't state what one candidate bottlenecks on and how the faster candidate avoids the bottleneck (and at what cost) is really worth nothing. It means whoever ran the results doesn't understand what they are testing.
A benchmark that doesn't state what one candidate bottlenecks on and how the faster candidate avoids the bottleneck (and at what cost) is really worth nothing. It means whoever ran the results doesn't understand what they are testing.