BenchLife has some details on Intel's (NASDAQ:INTC) upcoming Apollo Lake platform, which is the follow-on to the 2015 "Braswell" platform for low-cost notebooks and desktop PCs. The platform promises to bring a number of interesting new features to the table if BenchLife's information is good.
Everything gets better
In moving from Braswell to Apollo Lake, Intel plans to introduce its next-generation Goldmont CPU cores, which I believe will bring a solid improvement to the Airmont core found in Braswell. Additionally, the site says Apollo Lake will bring Intel's Gen. 9 graphics and media technologies, which should improve both media and gaming performance.
The platform promises to bring support for eMMC 5.0 storage, which AnandTech says raises peak theoretical data transfer speeds from 200MB/s with eMMC 4.5 (found in this year's Braswell chips) to 400MB/s. Finally, BenchLife says that Apollo Lake brings USB Type-C support.
All told, Apollo Lake looks like a compelling upgrade from Braswell and could lead to very compelling, low-cost PCs.
When is it coming?
Intel announced that Braswell began shipping to customers in April and that investors should expect the platform to show up in "more than 40 designs from a variety of OEMs" later this year.
Apollo Lake, per BenchLife, is expected to go into mass production in June 2016 and is expected to be formally launched at the Intel Developer Forum in 2016, suggesting an August to September launch timeframe.
Intel might need a "Braswell refresh"
It is well-known that the back to school selling season is key for PC makers. However, as Intel said in an interview with Reuters about a year ago, a chip is only "on time" for back to school if systems based on said chip are on the shelves in the July to August timeframe.
With production beginning in June and a formal launch in August or September, it seems as though Apollo Lake will miss the back to school season.
However, to help PC vendors spice up their low-cost PC lineups for the back to school season, I suspect that Intel will launch a "Braswell refresh" platform. As you may know, neither Broadwell nor Braswell was launched in 2014 to take over from Haswell and Bay Trail-M/D, respectively, leading Intel to release "refreshed" versions of these platforms.
Such refreshes are generally don't bring architecture or design changes, but such updates typically include higher CPU and graphics clock speeds, which can lead to reasonable performance increases. For example, the best Bay Trail-M chips launched in late 2013 ran the CPU at up to 2.42GHz and the graphics processor at up to 854MHz.
By the end of 2014, the very best Bay Trail-M chip ran the CPU at up to 2.66GHz and the GPU at up to 896MHz, representing improvements of 9.9% and 5%, respectively.
There is probably more headroom with Braswell than there was with Bay Trail-M
When Intel went into production on the initial Bay Trail-M likely sometime in the middle of 2013, it had been in volume production on its 22-nanometer technology for two years at this point. The technology was quite mature and, in Intel's own words, it was the "highest yielding technology" in the company's history.
With 14-nanometer, Intel described yields as being in a "healthy range" last August, but those yields had not yet matched 22-nanometer yields. I suspect that to keep costs as low as possible, Intel was willing to leave some performance on the table and be more conservative in clock speeds to be able to ship Braswell to customers by April.
By this time next year, when Intel could potentially launch a "Braswell refresh," I believe that the 14-nanometer process will be quite mature, meaning that more of the Braswell chips that come out of the factory might be able to run at higher speeds while keeping power consumption in check.
In that case, such "Braswell refresh" chips should be able to hold the line until Intel can get out Apollo Lake, which should bring much more substantial improvements.