Last week at their 4G/5G summit, Qualcomm introduced three new Snapdragons, the 653, 626, and 427. These mid-life kickers and updates do the usual power and performance boost but add some key features too.
On the surface they new trio replace the 652, 625, and 425 respectively making them 1, 1, and 2 better. The first two get the usual 10% performance boost mainly from a shrink to 14nm but the 425 isn’t shrunk so it just gets a little optimization. If you only look at those top line numbers however you miss the point, the phone market, especially in the mid-range, is not a performance game, it is morphing into a features arena faster than most people think.
Before we go into those features, numbers still matter which is why the 653 doubles the memory capacity to a full 8GB. If you have a clue about technology and how Android works, you will avoid 8GB phones like the plague due to power draw without real world benefit. If you are a clueless buyer who likes big numbers on the sales floor bullet point list, 8GB is a bigger number than 4 or 6GB so Qualcomm’s customers can hit that target now if they choose. For some large markets battery life is far less important than a number on the packaging so OEMs and carriers need something to sell to the clueless. This is why you see 8, 10, and even 12-core phones when 4 cores are nearly useless in the real world.
That is one low point ‘feature’, the rest are actually user beneficial and included in all three new Snapdragons. Those would be an X9 (Cat 7 download and Cat 13 upload), dual camera support, DSPs, and antenna boosting hardware. All these features have some obvious effects on the user but the real target for many of them is the carriers who are crying out for several things on the list and OEM/ODMs concerned about BoM and TCO.
This cry was answered by the X9 modem which has the unusual combination of Cat 7 300Mbps download (2x20MHz CA with 64-QAM) modulation plus Cat 13 150Mbps uploads (2x20MHz CA with 64-QAM and data compression). Why is this important, other than the top line speed number to users? Some carriers won’t allow deployment of new phones with less than LTE Cat 7 most likely due to their less efficient use of spectrum. If your chip is cheap and burns battery, is slow as dirt, and heats your thigh, that is your problem.
If your chip is cheap and uses spectrum inefficiently, that is the carrier’s problem and they tend to not like it. It will be interesting to see how players with their own modem in the low end price tier address this little conundrum. Qualcomm now has Cat 7 in all but the 200 series and Cat 13 seems to unique to them anywhere but the top end of the market. This feature is a shot across the bow to the low end of the market, but the gun isn’t aimed by Qualcomm as much as operators in areas with tight spectrum resources.
That brings us to the next feature, the antenna boosting hardware called TruSignal Antenna Tuner by Qualcomm is embodied by two chips, the WTR2965 and WTR3965. The idea is pretty simple, when you grab a phone you change it’s impedance. If your phone has a metal case, it probably uses it for an antenna on a few bands, it would be silly not to. Changing the impedance on an antenna is not a good thing, remember the iPhone 4 and antennagate?
TruSignal hardware essentially re-tunes that antenna on the fly to compensate for your hand or other indignities to the signal propagation path and makes things just magically work. It has been in the high end 800-series phones for quite a while now and it is why those devices can both have sleek metal cases and decent signal strength, but it tends to blow out the BoM on the mid-range devices and is right out for the low end. With the new Snapdragon 653/626/427 trio Qualcomm is now pushing this pretty vital tech to the volume space. Users won’t directly see the results but they phones they buy should behave a lot better especially in non-perfect signal conditions.
Dual camera support is just that and is something the users will notice because, well, it is pretty obvious. Almost all phones have at least two cameras, dual camera support means two at once processing different streams. Before you wonder about what use case this embodies, think stereo or depth camera, not front and back simultaneously. There are also a bunch of tricks you can do with monochrome sensors augmenting RGB ones for sharpness and the like but these are unlikely to hit the mid-tier. Stereo optics and 360 degree/VR cameras are ready for the mid-tier price point, both users and device makes have been crying out for this feature, it is a current ‘hot’ trend. Qualcomm effectively just enabled the mid-tier to do 3D and VR with the dual camera support effectively halving the price of entry in the end user device.
Last on the list is DSPs and again it will do a lot for the users but not in a way they will notice directly. The number crunching power they provide will mean better image processing at lower power but the real benefit is in the filters and software. All modern phones have an ISP to do the obvious conversions and compression but the DSP will allow a lot of additional cleanup like low-light filters and temporal sharpening for video that isn’t feasible in the time or energy budget on a DSP-free phone. It won’t change the world but it will make cameras better for the volume tier of the market. More importantly a common hardware base means a common code base for the software and firmware coders. This results in a ton of cost savings for the OEMs which makes a potentially more expensive ‘premium’ Qualcomm product more attractive at a net lower implementation cost.
That in a nutshell sums up the new Qualcomm Snapdragon 653, 626, and 427 trio. On the surface they are a bit faster with a few new features. If you look at why those features are there, they satisfy some pretty hard requirements from carriers, open up VR and 3D to the mid-range, make calls drop less, and consolidate software efforts. The users sees some of this directly and some because, well, things just work now. It is a fairly compelling story for customers, that would be OEMs and carriers, to pay more up front to save money on the total BoM. Not a bad play for a ‘minor’ update.S|A
Latest posts by Charlie Demerjian (see all)
- HyperX ships it’s 60 millionth enthusiast memory module - Oct 15, 2018
- Bittware/Nallatech water cools 300W of Xilinx FPGA - Oct 12, 2018
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018