Geekbench, back when it was 3.0, used to give to Apple an unfair advantage since it is a collection of basically split-second benchmarks. It was ridiculous since it made the iPad Pro look as fast as the Surface Pro 4, but anyone with sense knew that to be plain silly and that it was just a crazy game of boost clock roulette. With 4.0, though, balance has been restored again to the universe and it actually shows the iPad more along the lines of where it should be performance-wise. In particular, I am curious to see if the i7 model V breaks the coveted 7000-mark that the Surface Pro 4 i5 had once broken.
Another cross-platform benchmark that I really like is BAPCo's TabletMark v3. Gordon Mah Ung at PCWorld used it to debunk the Geekbench 3 myth that the iPad Pro is as fast as Surface Pro 4 and other Core-based tablets and laptops (http://www.pcworld.com/article/3006268/tablets/tested-why-the-ipad-pro-really-isnt-as-fast-a-laptop.html). This benchmark is an oldie but a goody which I also highly recommend. TabletMark measures sustained performance (none of Geekbench 3.0's skewed fraction-of-a-second shenanigans) in accurately simulated, simple real-world tasks like emailing and quick photo editing. Interested prototype testers can download it here:
Here is a sample of its results:
Here is why Geekbench 3 was a joke: