Apple (NASDAQ:AAPL) recently refreshed its 15-inch Retina MacBook Pro lineup of systems and elected not to wait for Intel's (NASDAQ:INTC) latest 14-nanometer Broadwell chips. Instead, Apple wound up using the same Haswell parts it had deployed in the 2014 editions of the systems.
Interestingly enough, Intel just formally announced the 14-nanometer Broadwell chips that would have been suitable for the 15-inch Retina MacBook Pro refresh, and I think I have a reasonable hypothesis as to why Apple may have chosen to skip Broadwell for those systems.
Not a particularly huge boost
The 15-inch Retina MacBook Pro can be configured with one of two parts: a Core i7-4870HQ or a Core i7-4980HQ. The 4870HQ runs the CPU at 2.5GHz base frequency and can turbo up to 3.7GHz; the GPU runs at 200MHz base and can turbo to 1.2GHz.
The higher-end 4980HQ runs the CPU at 2.8GHz base and can turbo to 4.0GHz. The graphics also run at 200MHz base but can turbo to 1.3GHz in graphically intensive workloads.
Now, if you look at the crop of Broadwell-H chips that Intel just announced, you'll note that the direct successor to the 4980HQ is the i7-5950HQ. The CPU runs at 2.9GHz base, and can turbo to 3.7GHz, and the GPU runs at 300MHz base and can turbo to 1.150GHz. The GPU in this case is based on an updated graphics architecture with more execution units (48 versus 40 in the Haswell parts).
All told, if we assume that Intel's claim that Broadwell delivers about 5.5% improved CPU performance per clock over Haswell, then the 5950HQ delivers better CPU performance at base speeds, but it might actually be slightly slower at max turbo.
Now, thanks to the 14-nanometer process, I suspect the Broadwell-H can sustain higher CPU performance for longer than the comparable Haswell parts. Additionally, Intel says that graphics performance moves up by 20% relative to the Haswell Iris Pro -- respectable, but not game-changing.
Lots of risk, little reward
According to the Intel press release for the Broadwell Iris Pro chips, systems based on these chips should be available within the next 30 to 60 days, or between early July and early August. That's about the correct timeframe for the back-to-school selling season, as Intel CEO Brian Krzanich said in an interview with Reuters that systems need to be on shelves in July/August to capture back-to-school demand.
So, Apple could either refresh the systems in May or attract customers sooner by refreshing the systems in other ways besides the processor. Given that the Haswell chip inside of the 2014 15-inch Retina MacBook Pro is already quite fast, and given that the improvements with Broadwell-H aren't game-changing, Apple's course of action makes sense.
The Skylake factor
One more thing to keep in mind is that while Broadwell with Iris Pro doesn't seem to be a huge upgrade from Haswell with Iris Pro, Skylake with Iris Pro should bring substantial improvements. According to a slide that leaked to the Web, Skylake-H with Iris Pro will offer "up to" 50% graphics improvement over Broadwell-H with Iris Pro.
Given that Broadwell-H with Iris Pro offers around 20% greater graphics performance than Haswell with Iris Pro, Skylake with Iris Pro should deliver up to an 80% increase from today's Haswell chips with Iris Pro.
Additionally, while Broadwell was a slight tweak of Haswell CPU, Skylake is a full-blown "tock" on the CPU front, which should help make Skylake an even more compelling offering.
If the recent leak from BenchLife is correct that Intel will launch Skylake with Iris Pro sometime between September and November, then it may make more sense for Apple to wait for Skylake (and do a full system redesign) than to refresh with Broadwell -- and then bring out a brand-new system design with Skylake shortly thereafter.