Looking Towards 2016
ARM expects a good and interesting 2016 in the mobile SOC market
ARM invited us to a short conversation with them on the prospects of 2016. The initial answer as to how they feel the upcoming year will pan out is, “Interesting”. We covered a variety of topics ranging from VR to process technology. ARM is not announcing any new products at this time, but throughout this year they will continue to push their latest Mali graphics products as well as the Cortex A72.
Trends to Watch in 2016
The one overriding trend that we will see is that of “good phones at every price point”. ARM’s IP scales from very low to very high end mobile SOCs and their partners are taking advantage of the length and breadth of these technologies. High end phones based on custom cores (Apple, Qualcomm) will compete against those licensing the Cortex A72 and A57 parts for their phones. Lower end options that are less expensive and pull less power (which then requires less battery) will flesh out the midrange and budget parts. Unlike several years ago, the products from top to bottom are eminently usable and relatively powerful products.
Camera improvements will also take center stage for many products and continue to be a selling point and an area of differentiation for competitors. Improved sensors and software will obviously be the areas where the ARM partners will focus on, but ARM is putting some work into this area as well. Post processing requires quite a bit of power to do quickly and effectively. ARM is helping here to leverage the Neon SIMD engine and leveraging the power of the Mali GPU.
4K video is becoming more and more common as well with handhelds, and ARM is hoping to leverage that capability in shooting static pictures. A single 4K frame is around 8 megapixels in size. So instead of capturing video, the handheld can achieve a “best shot” type functionality. So the phone captures the 4K video and then users can choose the best shot available to them in that period of time. This is a simple idea that will be a nice feature for those with a product that can capture 4K video.
Qualcomm features their “Hexagon DSP” as a fairly general processing unit that deals primarily with workloads like post processing photos and real-time video effects. It is aimed to be a very efficient number cruncher that can address a variety of workloads, but those of course will not cover everything. ARM does feature some DSP functionality in their lower end Cortex M SOCs that can leverage the performance of those parts in very specific and power efficient operations. For ARM’s larger SOCs they are relying on the Neon units as well as the GPU to cover most of those functions. The software support for these units is common as compared to more specialized DSPs. Qualcomm has opened up the functionality of their Hexagon units to 3rd party software developers, but so far that push is just beginning.
The only current ARM product line (Cortex M series) that integrate DSPs to increase performance in specific workloads while keeping power consumption low.
Another aspect not often considered is that a secondary DSP is not always needed. ARM has a very large portfolio of products that can actually allow their partners to integrate a “hidden” Cortex processor that will take over some of these functions. In this case a phone producer would create their own SOC, put in a smaller Cortex core that is not detailed, and direct their propriety software to leverage that extra core and tune the software to take the most advantage out of it. I would not doubt that this is already in play, especially with the companies pushing their cameras as a big selling point and differentiator.
VR is another area that ARM is looking at and addressing. Samsung has really taken a pretty good lead on the competition with their partnership with Oculus. HTC is working with Valve to implement their PC centric VR option. ARM has some thoughts and ideas on this arena as well. ARM feels that there are three basic areas: augmented reality for the professional market, gaming and graphics heavy workloads, and VR video playback.
Samsung is at the forefront of Android VR, but ARM looks to help support this growing ecosystem.
Augmented reality does not necessarily need to push the pixels, so it can be implemented for more professional applications (complex surgery, engineering/product visualization) without requiring heavy duty graphics hardware. Gaming at 4K and 120 fps is currently impossible with current mobile SOCs and is a stretch for even top end, high powered PC hardware. More simple titles may be able to work on some SOCs, but this is an area that just is not going to move forward for a couple of years. Finally we have VR video, an area that can be enjoyed now with current and upcoming mobile SOCs. Running video at 4K is already possible on high end phones and will be an area to pursue once more VR hardware is available for consumer use.
Process technology is an interesting area for ARM. As we have seen over the past few years, ARM has weathered the process delays far better than other 3rd party silicon designers. Due to their primary thrust of “low power or bust” ARM has been able to utilize design to get around limitations in process technology. Now that we are finally getting to where we have several next generation process nodes open for production, we are seeing A72 parts being fabricated on 28 nm processes as well as down to the latest 14 and 16 nm FF lines. Of interest to me was their thoughts on the latest GLOBALFOUNDRIES 22 nm FDSOI. ARM certainly supports it, but their partners will be the ones to actually implement their products on that process if it fits their needs.
Expect more "interesting" products this next year that encompass new visual technologies while still being powered by ARM IP.
While ARM may not release a major new part this year, they are staying very active in supporting and accentuating their shipping IP with improved software and design tools. 2016 does look to be an “interesting” year in that we will see a wide variety of products powered by ARM that could encompass many next generation technologies that leverage the compute potential and power saving features of the IP.
PC Perspective's CES 2016 coverage is sponsored by Logitech.
Follow all of our coverage of the show at https://pcper.com/ces!
Interesting to me was the
Interesting to me was the mention of GloFo’s 22nm FD-SOI, it’d been quiet for a while and i thought they might have mothballed it, btu its good to know that htey’re looking into the A72 on 22nmFDSOI, gives a nice comparisson point to the FinFETs and the planar 28nm we’re still (sadly) using
The ARM Holdings’ reference
The ARM Holdings’ reference design A72 core still decodes/executes instructions at half the IPC of the Apple A7 Cyclone core, so an A72 core will execute 3 IPC per core to the Apple A7’s 6 IPC per core. Now if we could get more information on Apple’s A8 and A9 cores as well as Qualcomm’s, Samsung’s, and even AMD’s K12, Custom ARMv8A ISA running cores’ IPC per core metric, we could have some basis to compare cores without having to use the Benchmarking metrics that can be so easily gamed by the manufacturers.
Hopefully by the time that AMD’s K12 custom ARM core information is released, we will at least know the IPC per core metric from AMD on their Custom K12 SKUs, as AMD will be more forthcoming with the proper CPU core execution resources/specifications. I think that Jim Keller/his K12 design team has done up something nice with AMDs first entry into the custom design ARMv8A ISA running custom micro-architecture market that is making custom designs that only use/license the ARMv8A ISA, and create their own custom micro-architectures that run that ARMv8A ISA. When AMD’s K12 hits the market, the other makers will have to be more forthcoming with the proper CPU core specifications also, compared to the piss poor information they currently provide, and they will not be able to rely on solely the gamed benchmarking results.
Don’t get me wrong that ARM holdings’ A72 reference core is an improvement over ARM holdings’ reference core design A57 both on the power usage and Back End execution resources feeding from the instruction decoders on the front end metrics, but I guess that that is good news for the third party ARM A72 reference core licensees who do not have the resources of Qualcomm, Samsung, Apple, or AMD to create their own custom micro-architectures designed to run the ARMv8A ISA. I’m hoping that AMD’s K12 will bring SMT capabilities to its custom ARMv8A running offerings, as that would put AMD in front of even Apple, if the rarely available proper/partial information on Apple’s A8/A9 cores is correct.
it’s really weirding me out
it’s really weirding me out you keep saying ARM Holdings instead of jsut saying ARM.
I understand the proper name etc, it is just very odd
Sounds like someone needs a
Sounds like someone needs a hug 😉
Seriously – basing your
Seriously – basing your estimate of real world performance on stats like “so an A72 core will execute 3 IPC per core” is likely to be highly misleading. There are way too many different factors and design tweaks etc. to take into account before you can make meaningful judgements from that kind of design information…
The interesting part is
The interesting part is A72.
A15 wasn’t great, A57 even less so and everybody was stuck on 28nm (more or less)
Now A72 creates new opportunities in high and mid range.
SD652/650 , X20 and X30, Kirin 950, Spreadtrum, maybe LG and some others. With options, come pricing pressure and we might see some really nice phones at reasonable prices – not from the old guard ofc, it’s not like Samsung will half it’s prices any time soon.
If at least some of these A72 based SoCs will have reasonable pricing, it could/should create a huge refresh in the mid range.
The high end will be tragic though. Seems that phone makers will stop pushing other specs up (to get better margins) and once that train stops, it never restarts. Plus what’s behind piles up, the mid range will be just slightly less but at much much better prices.
So , great year for mid range but the high end is a bit suicidal.