With the second generation thread getting closed and our endless love for speculation, it's time for one that focuses on the third generation MacBook. This post will act as a rational summary of what is expected in the next generation MacBook based on what we know today. This is a WikiPost, so anyone can edit it as new information surfaces.
Introduction
Skylake's successor is called Kaby Lake and it marks as a dramatic change in Intel's product strategy. In the past, Intel's CPUs followed a "tick-tock" model, where tick refers to a die shrink and tock refers to a new architecture. Basically, one year would be a tick, the next year a tock and the year after that another tick. Kaby Lake breaks the model because similar to Skylake, it is also a tock, which means two tocks in a row. In nutshell, shrinking transistors has become increasingly more difficult and expensive, which forced Intel to push back the 10nm node by a year (codenamed Cannonlake, expected in H2'17 now).
What this means is that Kaby Lake is essentially a refined version of Skylake. It uses the same 14nm FinFET manufacturing process and is built upon the Skylake microarchitecture, but features some enhancements to differentiate it from Skylake. Intel hasn't officially said much about Kaby Lake, so right now all we have is leaked information and rumours that may or may not be accurate.
What is almost certain is that the Y-series Core M chips will remain dual core with base TDP of 4.5W. It will feature GT2 graphics without eDRAM and will come in BGA1515 package similar to Skylake. (source)
Mobile Kaby Lake chips are scheduled to be released in Q3 this year and we will likely hear more details about the schedule during Computex in early June. In any case, it's unlikely that the MacBook will receive an update until early 2017, though, given Apple's past update behaviour.
CPU Performance
As the CPU architecture and process node are mostly unchanged from Skylake, it's unlikely that Kaby Lake will bring any notable increases in CPU performance. I would expect performance gains to be in the low single digits, most likely around 3-5%. This is clock for clock (i.e. IPC) gains - individual SKUs may of course carry higher gains due to matured 14nm process, but honestly I don't expect Kaby Lake to provide any substantial improvements in the CPU department.
RAM Support
Leaked information reveals that the memory controller in Kaby Lake Core M won't support DDR4 of any kind. That means the third generation MacBook will continue to use LPDDR3. A bump from 1867MHz to 2000MHz or 2133MHz is possible if the memory controller is updated to support higher bandwidths, but personally I find this unlikely and the performance gains would be marginal anyway.
GPU Performance & Media Playback
The GPU department may show greater gains, though, as it's rumoured that Kaby Lake will feature next generation (Gen10) graphics architecture (source). How much, that we don't know. Unfortunately the SKUs with faster GT3 graphics with eDRAM are 15W and up, so the improvements will be limited to the architectural enhancements. If Broadwell to Skylake upgrade is of any guidance, then the upgrade should be 10-40% depending on the test, so a conservative estimate would be in the 10-20% range. Not much, but pretty good for a year-over-year gain.
Hardware decode engine is expected to fully support Google's VP9, which is an alternative to HEVC codec (source). Skylake only features partial VP9 support and decoding is done in a hybrid way involving both CPU and GPU, whereas Kaby Lake will do it using the fixed function hardware blocks only, meaning longer battery life (source). Both HEVC and VP9 supports are also updated to support 10-bit content (i.e. higher quality). For the end-user these aren't a big deal since Skylake already has HEVC and VP9 support, but the improved VP9 engine and support for 10-bit may improve battery life and compatibility with some services that use such formats. Not many do at the moment, but in the future all will due to the huge gains in quality/bitrate ratio.
USB 3.1 & Thunderbolt 3
Leaked information shows that the high-end mobile and desktop SKUs with a separate Platform Controller Hub (PCH) have native support for USB 3.1 (likely means Gen 2 i.e. 10Gbps - source). However, the single-chip mobile SKUs including Core M will still be limited to USB 3.0 (or USB 3.1 Gen 1 i.e. 5Gbps), unfortunately (source).
In other words, the only way to obtain USB 3.1 support would be to use an additional controller, such as Intel's Alpine Ridge chip, which also supports Thunderbolt 3. That is unlikely, though, because the logic board simply has no space for an additional chip. If Apple had plans for Thunderbolt support in the MacBook, then I would argue that the first generation would already have had Thunderbolt support. This means that the 2017 MacBook will continue to be limited to USB 3.0 speeds like the first two generations. Cannonlake may bring native USB 3.1 support to Core M in 2018 MacBook, but there isn't any info on that yet.
Storage
The MacBook has always used a custom PCIe NVMe SSD controller designed by Apple. The first generation used a PCIe 2.0 x4 design, whereas the second generation migrated to a 3.0 x2 design. Given that PCIe 4.0 is still at least a couple of years away, I expect Apple to keep the same PCIe 3.0 x2 controller in the third generation MacBook.
We may, however, see a performance and capacity boost from 3D NAND. Samsung has been selling 3D NAND equipped SSDs since 2014 and Intel-Micron and SK Hynix are just starting to sell their first 3D NAND drives this quarter. With 3D NAND being more widely available and priced closer to planar NAND, I would expect Apple to adopt it in the next generation MacBook. The benefit of 3D NAND is improved performance due to larger cell geometry (i.e. better performance even with the same controller) and larger die capacities due to vertical scaling. Samsung and Intel-Micron are both already shipping 256Gbit dies, which would allow Apple to offer 1TB of storage in the MacBook.
Design
It is very unlikely that we will see a design change next year. Apple typically keeps the same design in Macs for at least three generations (e.g. original MacBook Air), so the third generation will merely be a spec bump, similar to this year's update. That means no second USB port or other fundamental changes to the product.
Miscellaneous
The 480p FaceTime camera creates a lot of discussion and I believe it's entirely possible that Apple updates it to 720p next year. I don't think there are any technical reasons as to why Apple chose 480p as 720p sensors are very small nowadays, so I think it was an economic decision.
Summary
All in all, the 2017 update is likely going to be a marginal one. Kaby Lake will bring some enhancements to performance, but compared to the 2016 update the improvements won't be as big. Battery life may be marginally improved, but not enough to warrant a change to the 10-hour rating. Anyone who is waiting for a bigger update will have to wait for Cannonlake in 2018.
The addition of 1TB of storage, likely as a BTO, is likely going to be the only "real" update in the 2017 model, but will only be useful to certain people.
Price drops are possible, perhaps even likely, as the MacBook will no longer be as innovative as it was in 2015. Price cuts are also required if Apple's strategy is to position the MacBook as a MacBook Air replacement in the long run, which to me makes sense given how much the two overlap.
Introduction
Skylake's successor is called Kaby Lake and it marks as a dramatic change in Intel's product strategy. In the past, Intel's CPUs followed a "tick-tock" model, where tick refers to a die shrink and tock refers to a new architecture. Basically, one year would be a tick, the next year a tock and the year after that another tick. Kaby Lake breaks the model because similar to Skylake, it is also a tock, which means two tocks in a row. In nutshell, shrinking transistors has become increasingly more difficult and expensive, which forced Intel to push back the 10nm node by a year (codenamed Cannonlake, expected in H2'17 now).
What this means is that Kaby Lake is essentially a refined version of Skylake. It uses the same 14nm FinFET manufacturing process and is built upon the Skylake microarchitecture, but features some enhancements to differentiate it from Skylake. Intel hasn't officially said much about Kaby Lake, so right now all we have is leaked information and rumours that may or may not be accurate.
What is almost certain is that the Y-series Core M chips will remain dual core with base TDP of 4.5W. It will feature GT2 graphics without eDRAM and will come in BGA1515 package similar to Skylake. (source)
Mobile Kaby Lake chips are scheduled to be released in Q3 this year and we will likely hear more details about the schedule during Computex in early June. In any case, it's unlikely that the MacBook will receive an update until early 2017, though, given Apple's past update behaviour.
CPU Performance
As the CPU architecture and process node are mostly unchanged from Skylake, it's unlikely that Kaby Lake will bring any notable increases in CPU performance. I would expect performance gains to be in the low single digits, most likely around 3-5%. This is clock for clock (i.e. IPC) gains - individual SKUs may of course carry higher gains due to matured 14nm process, but honestly I don't expect Kaby Lake to provide any substantial improvements in the CPU department.
RAM Support
Leaked information reveals that the memory controller in Kaby Lake Core M won't support DDR4 of any kind. That means the third generation MacBook will continue to use LPDDR3. A bump from 1867MHz to 2000MHz or 2133MHz is possible if the memory controller is updated to support higher bandwidths, but personally I find this unlikely and the performance gains would be marginal anyway.
GPU Performance & Media Playback
The GPU department may show greater gains, though, as it's rumoured that Kaby Lake will feature next generation (Gen10) graphics architecture (source). How much, that we don't know. Unfortunately the SKUs with faster GT3 graphics with eDRAM are 15W and up, so the improvements will be limited to the architectural enhancements. If Broadwell to Skylake upgrade is of any guidance, then the upgrade should be 10-40% depending on the test, so a conservative estimate would be in the 10-20% range. Not much, but pretty good for a year-over-year gain.
Hardware decode engine is expected to fully support Google's VP9, which is an alternative to HEVC codec (source). Skylake only features partial VP9 support and decoding is done in a hybrid way involving both CPU and GPU, whereas Kaby Lake will do it using the fixed function hardware blocks only, meaning longer battery life (source). Both HEVC and VP9 supports are also updated to support 10-bit content (i.e. higher quality). For the end-user these aren't a big deal since Skylake already has HEVC and VP9 support, but the improved VP9 engine and support for 10-bit may improve battery life and compatibility with some services that use such formats. Not many do at the moment, but in the future all will due to the huge gains in quality/bitrate ratio.
USB 3.1 & Thunderbolt 3
Leaked information shows that the high-end mobile and desktop SKUs with a separate Platform Controller Hub (PCH) have native support for USB 3.1 (likely means Gen 2 i.e. 10Gbps - source). However, the single-chip mobile SKUs including Core M will still be limited to USB 3.0 (or USB 3.1 Gen 1 i.e. 5Gbps), unfortunately (source).
In other words, the only way to obtain USB 3.1 support would be to use an additional controller, such as Intel's Alpine Ridge chip, which also supports Thunderbolt 3. That is unlikely, though, because the logic board simply has no space for an additional chip. If Apple had plans for Thunderbolt support in the MacBook, then I would argue that the first generation would already have had Thunderbolt support. This means that the 2017 MacBook will continue to be limited to USB 3.0 speeds like the first two generations. Cannonlake may bring native USB 3.1 support to Core M in 2018 MacBook, but there isn't any info on that yet.
Storage
The MacBook has always used a custom PCIe NVMe SSD controller designed by Apple. The first generation used a PCIe 2.0 x4 design, whereas the second generation migrated to a 3.0 x2 design. Given that PCIe 4.0 is still at least a couple of years away, I expect Apple to keep the same PCIe 3.0 x2 controller in the third generation MacBook.
We may, however, see a performance and capacity boost from 3D NAND. Samsung has been selling 3D NAND equipped SSDs since 2014 and Intel-Micron and SK Hynix are just starting to sell their first 3D NAND drives this quarter. With 3D NAND being more widely available and priced closer to planar NAND, I would expect Apple to adopt it in the next generation MacBook. The benefit of 3D NAND is improved performance due to larger cell geometry (i.e. better performance even with the same controller) and larger die capacities due to vertical scaling. Samsung and Intel-Micron are both already shipping 256Gbit dies, which would allow Apple to offer 1TB of storage in the MacBook.
Design
It is very unlikely that we will see a design change next year. Apple typically keeps the same design in Macs for at least three generations (e.g. original MacBook Air), so the third generation will merely be a spec bump, similar to this year's update. That means no second USB port or other fundamental changes to the product.
Miscellaneous
The 480p FaceTime camera creates a lot of discussion and I believe it's entirely possible that Apple updates it to 720p next year. I don't think there are any technical reasons as to why Apple chose 480p as 720p sensors are very small nowadays, so I think it was an economic decision.
Summary
All in all, the 2017 update is likely going to be a marginal one. Kaby Lake will bring some enhancements to performance, but compared to the 2016 update the improvements won't be as big. Battery life may be marginally improved, but not enough to warrant a change to the 10-hour rating. Anyone who is waiting for a bigger update will have to wait for Cannonlake in 2018.
The addition of 1TB of storage, likely as a BTO, is likely going to be the only "real" update in the 2017 model, but will only be useful to certain people.
Price drops are possible, perhaps even likely, as the MacBook will no longer be as innovative as it was in 2015. Price cuts are also required if Apple's strategy is to position the MacBook as a MacBook Air replacement in the long run, which to me makes sense given how much the two overlap.
Last edited:
