You can win any performance comparison if you're willing to fail on most of the benchmark's components and argue that only the components you complete should be compared.
You can make any claim true if you tailor the set of benchmarks used to your desired winner.