GPU Performance and Closing Thoughts
GFXBench Offscreen – T-Rex
The T-Rex test is based on OpenGL ES 2.0 and includes textures, material, geometry and particle effective that were highly detailed at the time of release. The graphics rendering engine features planar reflection, specular highlights and soft shadows, providing a good workout even for flagship smartphones and tablets. Offscreen tests are run at 1080p, regardless of the device’s native resolution, and are best used to compare the performance between competing silicon, not competing devices.
The Adreno 430 performance is definitely an improvement over the A420, bring in a score that is 22% faster than the Snapdragon 805. It outpaces the Apple A8 graphics core as well, with the Mali T760 in the Exynos 5433 bringing up the rear.
GFXBench Offscreen – Manhattan
Manhattan was the first benchmark to utilize OpenGL ES 3.0 features and uses a nighttime setting with a lot of external illumination to stress the GPU. It uses a deferred rending engine with multiple render targets for the geometry pass, includes both diffuse and specular lighting, uses depth shadow maps, bloom, depth of field and quite a bit more.
Again the Adreno 430 proves to be a compelling option for mobile gaming, only losing out to the Apple A8X and the two ARM devices that utilize NVIDIA’s Kepler GPU cores. The advantage for Snapdragon 810 is only about 15% over the SD 805.
GFXBench Offscreen – ALU Performance
This test measures the pure shader compute performance using a fragment shader and rending a single full-screen quad.
Interestingly, the Snapdragon 810 is actually producing a faster overall result than Apple’s A8X but still falls behind NVIDIA’s GPU option. The Adreno 430 looks to be 56% faster than the A420 GPU from previous designs.
GFXBench Offscreen – Alpha Blending Performance
Rendering semi-transparent quads with uncompressed textures allows this test to measure the alpha performance of the GPU directly.
Both the Adreno 430 and the Adreno 420 perform identically, telling us that this portion of the GPU wasn’t upgraded along with the shaders. The results are still very good, beating the Apple A8, the Tegra devices and only losing the extremely impressive result of the Apple A8X.
GFXBench Offscreen – Driver Overhead
This test renders a large number of very simple objects one-by-one, changing state with each item in a pattern consistent with real-world applications. This allows the benchmark to measure the CPU overhead of the OpenGL driver.
Snapdragon 810 and Adreno 430 are well head of the 801/805 processors (2.18x) and matches the result of the NVIDIA SHIELD Tablet. The Denver-based iteration of Tegra K1 in the Nexus 9, as well as both Apple devices, show lower driver overhead in the software stack for OpenGL.
GFXBench Offscreen – Fill Rate Performance
The portion of the test measures texture fill rate performance by rendering four layers of compressed textures, a very common scenario in gaming.
Wow, the fill rate for the new Snapdragon 810 is incredibly fast and allows the reference platform to beat Apple’s A8X for the first time in GFXBench. Likely we are seeing the benefits of the LPDDR4 memory interface running at 1600 MHz here!
Use 3DMark Ice Storm Unlimited for chip-to-chip comparisons of the hardware inside your device without vertical sync, display resolution scaling and other operating system factors affecting the result. In Unlimited mode the rendering engine uses a fixed time step between frames and renders exactly the same frames in every run on every device. The frames are rendered in 720p resolution "offscreen" while the display is updated with small frame thumbnails every 100 frames to show progress.
Ice Storm Graphics test 1 stresses the hardware’s ability to process lots of vertices while keeping the pixel load relatively light. Hardware on this level may have dedicated capacity for separate vertex and pixel processing. Stressing both capacities individually reveals the hardware’s limitations in both aspects. Pixel load is kept low by excluding expensive post processing steps, and by not rendering particle effects.
Graphics test 2 stresses the hardware’s ability to process lots of pixels. It tests the ability to read textures, do per pixel computations and write to render targets. The additional pixel processing compared to Graphics test 1 comes from including particles and post processing effects such as bloom, streaks and motion blur. The numbers of vertices and triangles are considerably lower than in Graphics test 1 because shadows are not drawn and the processed geometry has a lower number of polygons.
The purpose of the Physics test is to benchmark the hardware’s ability to do gameplay physics simulations on CPU. The GPU load is kept as low as possible to ensure that only the CPU’s capabilities are stressed. The test has four simulated worlds. Each world has two soft bodies and two rigid bodies colliding with each other. One thread per available CPU core is used to run simulations. All physics are computed on the CPU with soft body vertex data updated to the GPU each frame. The background is drawn as a static image for the least possible GPU load. The Ice Storm Physics test uses the Bullet Open Source Physics Library.
In the graphics sub-test, the Snapdragon 810 does a very good job of showing its capability with Adreno 430, matching the Tegra K1 implementation of the SHIELD Tablet, and losing only to the Nexus 9 implementation. The Apple A8X scores 10% lower, and the Snapdragon 805/Adreno 420 is 35% lower.
The CPU-bound physics tests paint a different picture – the A57+A53 result of the SD 810 dramatically lower than that of the Exynos 5433 found in the Samsung Galaxy Note 4, and is slower than our scores from the Nexus 6 and OnePlus One using previous generation Snapdragon products.
Basemark X is the world’s most popular benchmarking tool for evaluation and cross-platform comparison of gaming and graphics performance between Android, iOS and Windows Phone 8 smartphone and tablets.
Basemark X is the only vendor-independent benchmark that utilizes the real-world game engine Unity which is very popular among game developers. This means that it scores correlate exceptionally well with real-life gaming performance.
Basemark X includes two game-like graphics tests: Dunes and Hangar. Both tests contain heavy graphics content rendered with detail and complexity, thus pushing the measured device to the limit. The polygon counts in test sequences are up to 911,000.
Here is a result that confuses for sure – at both medium and high quality settings in Basemark X, the new Snapdragon 810 and Adreno 430 actually lose to the Snapdragon 805 and Adreno 420. How that’s possible is escaping me right now, but both the phone and tablet scores show similar results, so I can assume it is not a bug of a single test run.
Basemark OS II is a system-level All-In-One benchmarking tool designed for measuring overall performance of smartphones and tablets from all platforms, including Android, iOS and Windows phone 8.
The benchmark features a comprehensive suite of tests including system, internal and external memory, graphics, web browsing, camera, battery and CPU consumption.
After the result in Basemark X, this result in Basemark OS II is completely the opposite: the Adreno 430 tops the charts beating out even the high powered Apple A8X and Tegra K1 Denver.
Closing Thoughts
For now, we only have cursory glance at the new Qualcomm Snapdragon 810 platform in a reference system design. Because of the limitations of testing we didn’t get to all of the benchmarks we wanted, and we did not get to really USE the device for an extended period of time to really test the real-world user experience it provides. Based on the performance metrics we saw on the previous pages, the 8-core Coretex-A57 / A53 design seems to provide more than enough CPU horsepower for the tasks at hand, besting the Exynos 5433 in many cases and providing a gain over previous Snapdragon parts ranging from 5-55%. Qualcomm may have difficulty differentiating without the custom core designs that we are used to seeing in its flagship SoCs, but it doesn’t seem to have hurt the leader in mobile computing very hard yet.
From a graphics perspective, the Adreno 430 is definitely a step above the Adreno 420 in our testing and is able to overpower the Apple A8 found in the iPhone 6 and iPhone 6 Plus. The A8X is a much more powerful part though, and the A430 was never on par with it during offscreen gaming tests. The lone victory there was in the GFXBench fill rate test where the LPDDR4 interface brought a large memory bandwidth advantage. NVIDIA’s Tegra K1 parts are still very strong in the GPU department and scored better than A430 in our testing, but not by as much you might have expected. NVIDIA did announce Tegra X1 using Maxwell cores at CES in January but I have yet to get hands-on time with any hardware using it.
The other very important factor when looking at mobile processors is power efficiency. Even though Qualcomm has claimed that efficiency of the Snapdragon 810 will match or be better than that of the 805 and 801, without more time with the reference devices or retail smartphones in our hands it is impossible to tell for sure. A device that is powerful and fast but with limited battery life in consumer devices just isn’t going to cut it; that has been NVIDIA’s trouble going anywhere but into tablets with Tegra K1.
The features that Qualcomm has integrated into the SoC are impressive as well – Cat 9 LTE modem, LPDDR4 interface, 4K HEVC/H.265 encode capability, 14-bit dual ISP for depth enabled photography, UFS 2.0 support and even 802.11ac/ad support through an external chip. Despite not having a custom architecture core for this part, Qualcomm has clearly put a lot of engineering to making this a flagship product.
Call me cautiously optimistic on Snapdragon 810. We still need to see retail-ready hardware to really put the claims of Qualcomm and our own benchmarking to the test. The company is eager to prove that rumors about the chip and its issues are false. If what we see here today can make it into some devices by spring of 2015, I think Snapdragon will have another winning product in the family tree.
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