Google’s Pixel 2 smartphone does not simply have top-of-the-line smartphone cameras ever; it additionally has customized silicon devoted to the digicam that is not even energetic but. Besides the Snapdragon 835, the Pixel 2 has a complete different SoC for picture processing known as the “Pixel Visual Core.” The chip represents Google’s first-ever client SoC, however Google did not construct the chip by itself. CNBC came upon the chip was a collaboration between Intel and Google.
CNBC made the connection after seeing that the serial variety of the chip begins with “SR3,” which can be used on some Intel chips. The outlet ran its scoop by Google, which confirmed Intel was concerned.
Knowing that Intel helped with the event of the chip was sufficient info to start out digging with, since something touched by Intel might be associated to the digicam chip, proper? This led me to the codeword “Easel,” which, certain sufficient, appears to be Google’s codename for the Pixel Visual Core. You can poke round platform//google/easel/ within the Android supply, the place you will discover the few bits of associated code which are presently public. Opening up the device-tree blob binary current on the Pixel 2 additionally prominently reveals the phrase “Monette Hill,” which appears like some sort of Intel codename.
A smartphone design win for Intel is a uncommon prevalence, for the reason that trade’s reliance on ARM processors means Intel is often absent from the world’s hottest computing type issue. The firm has made inroads on smartphone modems, which present up in sure iPhones. For Android OEMs, a separate Intel modem is a tricky promote when Qualcomm can provide modems built-in with its SoCs.
Google’s Pixel Visual Core is not energetic but, however the firm says it will likely be turned on with the launch of Android Eight.1. The Eight-core Image Processing Unit (IPU) will supposedly permit Google’s HDR+ picture processing to run “5x faster and at less than 1/10th the energy” of the present CPU-driven implementation. It may even be a programmable platform for different Google imaging and machine-learning capabilities.