If you think power isn’t important then you must have been living under a rock for the last decade. But you could try asking Tim Cook, the CEO of Apple, too. You have probably heard by now, since you are presumably something to do with the semiconductor industry, that for the first time ever Apple decided to dual source the application processor A9 that is used in both the iPhone 6s and 6s Plus (the big-screen one). They never announced they were using both foundries, but ChipWorks in Canada discovered it when they bought some phones and then ripped up the chips under electron microscopes as they always do. The two chips are in the TSMC 16FF+ process and the Samsung 14LPP. None of us mortals know how close the processes are and at what level the design was finished for each foundry: identical RTL at a minimum, identical netlist maybe. My guess is that the design was taped out in TSMC and then the layout run in Samsung with some tweaks (SRAM cell, I/Os). The metal 1 pitch of the two processes is the same (64nm) and the minimum gate pitch for Samsung is tighter (78nm vs. 90nm), so it is not out of the question (as it would be to attempt to tape out for Samsung and then try and run in TSMC). But for sure the functionality of the two application processors is the same, the clock rates are the same, and so the behavior to the user is the same. Except for one thing. The power dissipation. Once you have the netlist and the clock, the power dissipation, and therefore the battery life, is what you get. It turns out that the design dissipates more power in the Samsung process than the TSMC process. How much is a little unclear since it depends so much on what you are doing. After all, even within a given foundry, natural process variation can result in power differences that might be significant. Apple is not trying to produce parts to different performance points, but a company like Intel often “bins” the parts depending on which corner of the process they hit, which is why, if you pay a premium, you can buy a PC (or Mac) with the same processor but higher clock speed. In principle, Apple could bin parts and sell the ones with the longest battery life at a premium, but that would not fit their image, plus it would mess up their supply chain completely. In case you were wondering, the chip that is in the phone does not appear on the outside of the box. You cannot go into your local Apple Store and choose between TSMC and Samsung. If you have an iPhone 6s or 6s Plus you can tell which chip you have easily enough, by running a free app (I was going to provide a link but they just pulled the app I was going to link to; Google to find the latest). The basic idea is to find the name of the application processor. If It has an "M" in the middle it is TSMC. If not then it is Samsung. The difference between chips from the two different foundries (indeed, any two chips) depends on what you are doing on the phone. The biggest difference I have seen is on Reddit where a poster raydizzle found that when running the Geekbench battery test, the difference could be as large as six hours of battery life for Samsung versus eight hours for TSMC. But this is not in any way a realistic test, it does its best to run the CPU flat out continuously. In normal use, the screen backlight usually will consume more than the application processor. Interestingly, Samsung reportedly switched from the Qualcomm Snapdragon to their own internal chips for the high-end Galaxy phones due to power and thermal issues. However, I believe that was a 20nm TSMC chip, which is a very different process (not FinFET but planar, with high leakage). Talking of Qualcomm, if you haven't seen it, watch their video of the butter benchmark. Yes, butter. Really. (Please visit the site to view this video) Apple never talks much about these supply details. But to show you how important power is, they made a statement to TechCrunch: With the Apple-designed A9 chip in your iPhone 6s or iPhone 6s Plus, you are getting the most advanced smartphone chip in the world. Every chip we ship meets Apple’s highest standards for providing incredible performance and deliver great battery life, regardless of iPhone 6s capacity, color, or model. Certain manufactured lab tests which run the processors with a continuous heavy workload until the battery depletes are not representative of real-world usage, since they spend an unrealistic amount of time at the highest CPU performance state. It’s a misleading way to measure real-world battery life. Our testing and customer data show the actual battery life of the iPhone 6s and iPhone 6s Plus, even taking into account variable component differences, vary within just 2-3% of each other. Unless you are someone who spends all their time running games…and in a dark room so the backlight for the display is really low…and by all your time, I mean “all”, you never go on Facebook…in that case you might notice a difference. If you are normal, then you do all sorts of things on your phone: email, text, twitter, play a game, watch YouTube, take photos, listen to music. Even make a phone call on rare occasions. I really don’t think the difference is going to be significant. My opinion. I agree with Wired Magazine. It doesn’t matter which chip your iPhone has, get over it . By the way, the habit of putting "gate" on the end of anything to name a scandal has been around for over forty years. Those of you who are not old enough might wonder what President Nixon did with water to create such a scandal that it got called "watergate". In fact, Watergate is the original scandal with the -gate suffix; it is the name of a building in Washington DC. I suppose now we would have to call it Watergate-gate.![]()
![]()
↧