Wilco (Wilco.Dijkstra.delete@this.ntlworld.com) on August 10, 2014 5:48 am wrote:
>
> So given the ISAs are very similar it is hard to believe there will be a big performance difference between
> 32 and 64-bit

I would expect you to be right, and I think people are confused by the geekbench numbers on A7, which are fairly wildly different between the builds.

Now, some people are confused by geekbench numbers just because they can't even read them, and they look at the combined numbers which include things like hw-accelerated crypto, which exists in arm64 but not arm32. There were a number of breathless "ooh, 64-bit is much faster" by people who apparently didn't catch on.

But even if you look at the non-crypto geekbench numbers, there were real differences between 32-bit and 64-bit geekbench builds that people seem to attribute to arm64 being magical. I suspect it's more likely to be a geekbench build issue. Or possibly it's a real A7 quirk - it might handle some 32-bit construct badly just because Apple didn't care.

I *really* hope the ARM ecosystem gets good enough that we'll see better benchmarks. I am not exactly known as an ARM fanboi, but I'd _love_ to finally see an ARM laptop that is actually good enough to be worth running as a real laptop (yeah, I actually use chromebooks, but the only one so far that was good enough to actually use for real Linux and kernel development was intel-based)

Linus