New Features and Specifications
We got to spend some time with Qualcomm engineers to talk about the 810 SoC, what it brings to the table and then do some benchmarking. Ready for your next flagship phone?
Introduction
It is increasingly obvious that in the high end smartphone and tablet market, much like we saw occur over the last several years in the PC space, consumers are becoming more concerned with features and experiences than just raw specifications. There is still plenty to drool over when looking at and talking about 4K screens in the palm of your hand, octa-core processors and mobile SoC GPUs measuring performance in hundreds of GFLOPS, but at the end of the day the vast majority of consumers want something that does something to “wow” them.
As a result, device manufacturers and SoC vendors are shifting priorities for performance, features and how those are presented both the public and to the media. Take this week’s Qualcomm event in San Diego where a team of VPs, PR personnel and engineers walked me through the new Snapdragon 810 processor. Rather than showing slide after slide of comparative performance numbers to the competition, I was shown room after room of demos. Wi-Fi, LTE, 4K capture and playback, gaming capability, thermals, antennae modifications, etc. The goal is showcase the experience of the entire platform – something that Qualcomm has been providing for longer than just about anyone in this business, while educating consumers on the need for balance too.
As a 15-year veteran of the hardware space my first reaction here couldn’t have been scripted any more precisely: a company that doesn’t show performance numbers has something to hide. But I was given time with a reference platform featuring the Snapdragon 810 processor in a tablet form-factor and the results show impressive increases over the 801 and 805 processors from the previous family. Rumors of the chips heat issues seem overblown, but that part will be hard to prove for sure until we get retail hardware in our hands to confirm.
Today’s story will outline the primary feature changes of the Snapdragon 810 SoC, though there was so much detail presented at the event with such a short window of time for writing that I definitely won’t be able to get to it all. I will follow up the gory specification details with performance results compared to a wide array of other tablets and smartphones to provide some context to where 810 stands in the market.
Let’s dive in! Continue reading our preview of the new Qualcomm Snapdragon 810 SoC!!
Snapdragon 810
The Qualcomm Snapdragon 810 processor is a SoC (system on a chip) that combines nearly every aspect of hardware a device needs to operate with exception of memory. Wi-Fi was left off of the SoC with this generation as well but I’ll discuss that a bit later. It marks the first flagship chip from Qualcomm to use TSMC's 20nm process technology, the same used on Apple's A8 and A8X. This will provide an interesting comparison to the upcoming Samsung parts that will instead use 16nm process technology.
Processor
The majority of tech enthusiasts know Qualcomm for its processor lineup, which is typically a custom design based on ARM’s CPU architecture. The Krait family of cores was introduced in late 2012 and shared architectural similarity to ARMs Cortex-A15. Before that Qualcomm built the Scorpion core, based on the Cortex-A8 and A9. But these custom designs included additional performance, power efficiency and features that differentiated them from “off-the-shelf” processors from ARM.
But for the Snapdragon 810, the company is using parts directly from ARM’s design books: the Cortex-A57 and Cortex-A53. There are four of each of these 64-bit ARMv8 processing cores which operate in a big.LITTLE configuration allowing the larger and faster (but more power hungry) cores to run during intensive computing scenarios while the smaller and more power efficient cores operate during standard use cases and in idle states. We have been covering the technology behind ARM’s big.LITTLE initiative for several years and the methodology is well understood.
This shift away from custom designed cores is very likely a temporary fix for Qualcomm and you should expect follow-up generations of Snapdragon to reintegrate custom 64-bit cores. For now though, Qualcomm had to ensure that it wasn’t being perceived as falling behind in the ecosystem with a processing capability of only 32-bits, even if the need for 64-bits is somewhat exaggerated for today’s users. It also means that Qualcomm’s SoC loses one of its key selling points over other parts like Samsung’s Exynos 5433 and 7410, both of which also integrate the same A57+A53 8-core design. This might also explain the rumored move from Snapdragon to Exynos for Samsung’s Galaxy S6 later this year.
From a performance perspective, we still expect the Snapdragon 810 to be faster than the 801 and 805 designs that used quad Krait custom cores. The 64-bit ISA buys software some “free” performance upgrades, and the continued migration of Android Lollipop will keep that upward momentum throughout 2015.
Graphics
The graphics portion of the new Snapdragon 810 is powered by a new GPU design, the Adreno 430. This mobile GPU supports OpenGL ES 3.1, OpenCL 1.2 Full and improves performance over the Adreno 420 as well. Qualcomm claims that A430 will offer 30% “faster graphics performance” and “100% faster GPGPU compute performance” when compared to A420. Truthfully we don’t know much about the exact specifications of the Adreno 430 just yet, but it looks like it will be a modular step forward – increased shaders, cache and clock speed to gain the performance we see in our benchmarks.
The Adreno 420 is a unified shader model with a VLIW-5 pipeline, 128 ALU/shaders and clock speeds ranging from 500-600 MHz. 420 is rated at 337.5 MTriangles/s and up to 4.8 GigaPixels/s along with a peak performance rating of 172.8 GFLOPS. The new Adreno 430 is rated at 388.8 GFLOPS at its top speed (600-800 MHz), which is a 2.25x increase and should result in impressive gaming and graphics gains if true.
Architecturally, I don’t think the Adreno 430 offers any features that weren’t already supported on the A420. With the Adreno 4x series we saw Qualcomm add support for DX11.2 hardware tessellation and full profile OpenCL, all of which are carried over the A430. Full support for the Android Extension Pack (AEP) exists as well to improve graphics feature support beyond just OpenGL ES standards. With an abundance of added compute capability though you can expect Qualcomm to have better GPGPU integration as well, which should be helpful as we dive into the 4K content discussion.
Both graphics and processor segments of the Snapdragon 810 will benefit from the inclusion of Qualcomm's first LPDDR4 1600 MHz memory controller. Compared to LPDDR3, used in nearly all modern SoC, the memory controller on the 810 should provide substantially more memory bandwidth, helping in all areas of compute: graphics, compute and GPGPU.
4K Support
Both content creation and content consumption are accelerated with the Snapdragon 810 processor. The multimedia processor on the SoC is Qualcomm’s first to enable both hardware encode and hardware decode support for H.265/HEVC. HEVC allows for as much as 64% better compression for video content at 4K resolutions but requires a lot of processing resources to be encoded and decoded in real time. The Snapdragon 810 will be able to record 4K video at 30 FPS with bit rates ranging from 20-60 mbps, and a demo we saw on-site featured 4K video editing in a nearly real-time fashion on the reference platform.
Obviously along with support for 4K recording, 810 will support 4K output on both internal displays and external displays via HDMI 2.0. It is also possible that with support for WiGig (802.11ad), a high-bitrate streaming solution will be produced for 4K content.
LTE Modem
One of the most important areas for Qualcomm, and admittedly an area I am still learning about, is the LTE modem space. The Snapdragon 810 includes the first commercial modem to support integrated Cat 9 LTE-Advanced carrier aggregation. Carrier aggregation is used to increase bandwidth for device communication without breaking backwards compatibility with previous LTE standard. By combining up to three carrier channels at 20 MHz each, you can achieve a total bandwidth of 450 Mbps. This integration of the Cat 9 LTE supports all cellular modes and bands while also supporting new standards like LTE Broadcast, VoLTE (voice LTE) and LTE DSDA (dual SIM).
Along with the modem integrated into the Snapdragon 810, Qualcomm has a next-generation 28nm RF chip (WTR3925) and the previously shipping RF360 that supports global LTE infrastructure at a reduced cost to the OEM. Qualcomm specific features like Envelope Tracker, Antenna Tuner and CMOS PA/switching are available to customers that integrate the RF360 and are able to improve power efficiency and coverage/call reliability.
WiFi and Connectivity
As I mentioned above, the Wi-Fi controller for the Snapdragon 810 is actually on a different physical chip. The QCA6174A offers 2×2 capability and quite a bit of improvement on supported connectivity, including 802.11ac (wave 2) and 802.11ad (60 GHz WiGig). Support for the second wave of 802.11ac will enable users to utilize multi-user MIMO (MU-MIMO) once you upgrade to a supported router in 2015. (As a side note, Qualcomm had doubts if currently shipping routers with claimed MU-MIMO support would actually be able to meet the standard.)
With WiGig, or 802.11ad, Qualcomm is setting up the Snapdragon 810 for a future of ultra-high bit-rate streaming, and during a live demo of the technology we saw 4K video streaming at upwards of 70 Mbps. Bursts of up to 7 Gbps have already been measured and I was assured that despite all the commotion about it, range in a single room doesn’t not appear to be an issue during implementation.
It’s not the first SoC to support it, but USB 3.0 is going to more important for flagship mobile processors going forward. After all, if you are recording and capturing 4K video then you are going to want a way to remove that video from your phone quickly. Migrating several Gigabyte files over USB 2.0 or even fast Wi-Fi can be arduous. I definitely hope we see Type-C connectors rather than the often criticized micro connection for USB 3.0.
Speaking of storage, the Snapdragon 810 will be the first part from Qualcomm to support UFS 2.0 (universal flash storage) and data rates as high as 1.45 GB/s. That is a dramatic boost over eMMC v5.0, which peaks at 400 MB/s but no phone had yet to integrate that version. As 4K video and other content creation tasks get moved to mobile devices, faster storage can help improve the user experience as well transfer speeds on and off of the phone/tablet.
Thermals
As important as performance is for a flagship device, a crucial part of the experience is known as skin temperature, the temperature of the phone surface where you are holding it. This is controlled by a combination of power consumption of the unit and heat dissipation properties of the enclosure. In a normal use case, with a room temperature of 25C, skin temperatures should be in the 30-35C range, while in heavy uses cases (gaming, 4K recording) that can reach as high as 45C.
By moving to a 20nm process technology with the Snapdragon 810, Qualcomm is able to produce better performance at lower thermals, resulting in devices that are more comfortable to hold than could be built with Snapdragon 801.
This example presented by Qualcomm at the tech day, shows 20-30 minutes of gaming time between devices. In the result from the Snapdragon 801, skin temperatures reach as high as 45C within a 20 minute window. The Snapdragon 810 maintains a 40-41C temperature while gaming for as long as 30 minutes. The end result is the ability for the GPU and CPU to remain at higher clocks for longer periods of time, producing better frame rates for users.
Using 4K recording as another skin temperature example, the SD 810 has an even larger advantage thanks to the native HEVC/H.265 encode acceleration. On the older SoC, temperatures reached 42-43C within 6 minutes but the Snapdragon 810 only hit 35C in the same time span.
The cool thing about
The cool thing about UFS–which I just learned–is that it uses the SCSI command stack instead of the SD style of command passing that eMMC uses. So, this should show some small block read/write performance improvements much like UAS did for USB. That could make a big difference. Be sure to test it if a device ever uses such a storage memory, please!
It will definitely be
It will definitely be something we pay attention to going forward.
WOW those SunSpider JS
WOW those SunSpider JS benchmarks, among others for the A8, and A8x really jump out, the A8 having the Cyclone 2 microarchitecture, that we have never really had an indepth review owing to Anand’s departure from anandtech. I’m no great fan of Apple, but the custom wider order superscaluar of the A8/A8X, as well as the Denver microarchitecture from Nvidia, sure perform with their 6, and 6+ IPC respectively, wide execution pipline designs. It’s no wonder the mobile phone makers are going to the 8 core ARM reference design A53/A57 Big/Little designs with their 3 IPC per core narrower superscalar resources. Arm holdings’ newest reference design(A72), if it retains its narrower superscalar design, with only 3 IPC per core, is going to have problems competing on the top tier, but for sure those Custom offerings from Apple, and maybe Nvidia(what’s up with Denver?), as well as AMD’s K12 custom ARM microarchitecture can/will continue to take the high end ARMv8A ISA based market. It’s the high IPC per core custom ARM designs that will continue to give Intel headaches in the mobile arena.
I like your CPU tables, and charts, but could you include the IPC per core figures for all these CPU microarchitectures in the future, and try to get more information on metrics such as reorder buffer size, numbers of floating point/integer pipelines, and other microarchitecture details specific to each makers CPU core.
The custom microarchitectures have much more in the way of execution resources, beyond the standard ARM Holdings’ reference designs.
Those A8/A8X designs(mostly from Apples’ acquisition of P.A. semiconductor, and other IP/company acquisitions) are really top notch, and hopefully AMD’s Jim Keller will be designing such a custom ARMv8a ISA based design(K12) to compete with the Cyclone microarchitecture. The newest ARM holdings reference design A72 if it retains the 3 IPC rate, will have to make up for the deficiency with more cores, and hopefully the cores can be individually power gated, or power gated in groups of 2, for finer power scaling, which would be great for octo-core or higher Tablet/phone SKUs. Apple did a great job of acquiring brainpower, and AMD is on the right track, with Jim Keller, and Lisa Su, and getting AMD a custom ARM ISA based microarchitecture.
Even when the rumors were a
Even when the rumors were a consideration, I am more concerned with the heat generated by displays than the CPUs at this point.
Very interesting article
Very interesting article Ryan. My only concern is that the reference tablet and phone are very large and could be dissipating heat better than, let’s say, a phone with an HTC One chassis?
I know that’s an obvious point and that this is a preview, but it still bugs me. Again, great article 😀
So with ufs will we get more
So with ufs will we get more then 100kb to 1.1mb transfers? I have a old Asus Tf700 and I tell ya transferring anything to or off is super slow. Or transferring from Pc to tablets you can only transfer 1 file at a time so annoying.
Anything and everything “old”ASUS T1700.
The CPU is a NVIDIA® Tegra® 3 Quad-core, 1GB memory, Storage 32GB / 64GB *1 EMMC + 8G life time ASUS Webstorage space, USB 2.0, Wi-Fi a/b/g/n 2.4 Mhz, Storage 32GB or 64GB. Compare those with what this will have…
Anything and everything “old” is going to be slow. Doesn’t matter what it is. USB, Wi-FI, and storage.
Just out of curiosity, I looked up your
I think we will be able to safely consider the Asus T1700 to be a dinosaur one this Qualcomm SoC hits the markets.
So this means the Asus zen 2
So this means the Asus zen 2 is aimed squarely to counter an top of the line 20nm?Intel pretty much set their timetable to counter next gen top of the line with 14 mm?(apple iOS,Samsung)men this will be a interesting 2015