If you haven’t yet ordered your new MacBook Pro, AMD has provided a little help by showing what kind of performance differences you can expect if you opt for any of the upgrades on offer.

If you’re buying the 13-inch model, then you don’t get much in the way of GPU options. All the machines have Intel Iris integrated graphics, and the only difference between them is a 540 chip in the base model versus 550 in the other two – with no configuration options beyond that.

But things get more interesting with the 15-inch MacBook Pro …


All 15-inch specs have an AMD Radeon Pro GPU, built on the latest Polaris architecture. Here’s what the company has to say about them before we get into the numbers:

The MacBook Pro powered by state of the art Radeon™ Pro 400 Series Graphics is built on the cutting edge Polaris architecture with future-ready display technology. Precisely fabricated using the latest 14nm FinFET process means high quality visuals are achieved without sacrificing portability.

[The chips feature] 4th generation Graphics Core Next [which] accelerate workloads normally reserved for the main processor. It features versatile asynchronous compute, updated shader engines, enhanced memory compression and new geometry capabilities in a compact and efficient package.

Looking at the options Apple offers, the base model has the 450 with 2GB memory, with the option of a $200 upgrade to the top-spec 460 with 4GB memory.

The higher-spec machine steps things up to the Radeon Pro 455 with 2GB memory, with the same option of maxing out on the 460 – but this time as a $100 upgrade given the higher starting-point.

What do you get for your money? AMD’s microsite details the performance differences. All three GPUs offer the same memory bandwidth of 80GB/s, though the top-end 460 of course has twice the memory. The differences between the chips show up in compute units and peak performance.

  • Radeon Pro 450: 10 compute units, peak performance of 1 teraflop
  • Radeon Pro 455: 12 compute units, peak performance of 1.3 teraflops
  • Radeon Pro 460: 16 compute units, peak performance of 1.86 teraflops

We’ll of course need to await benchmarks and reviews before we get a sense of what this means in real-life use, but on paper at least you are getting a significant bump in performance if you opt for one of the higher-spec options. It’s also worth remembering that GPU chips these days don’t just handle graphics processing – they are also used for some of the more intensive computing work that used to be done by the CPU.

It’s making me feel a little better about my own choice – watch this space for my take on the new machines. Which option have you ordered or are planning to order? Let us know in the comments.