Page 1:Redefining The Android Experience With Google's Nexus 5
Page 2:Product 360: Look And Feel
Page 3:GEL: A Better Experience
Page 4:GEL Gets Personal
Page 5:Benchmark Variance: Not Every SoC Is Created Equal
Page 6:Test Setup And Methodology
Page 7:Results: CPU Benchmarks
Page 8:Results: GPU Benchmarks
Page 9:Results: GPU Benchmarks, Continued
Page 10:Results: Web Browsing Benchmarks
Page 11:Results: Display Measurements
Page 12:Results: Battery Testing
Page 13:Does The Nexus 5 Raise Expectations?
Results: GPU Benchmarks
Futuremark is a name synonymous with benchmarking GPUs, and 3DMark for mobile has quickly become a very popular tool for doing just that. 3DMark offers three main graphical benchmarks which simulate the demands of OpenGL ES 2.0 games using shaders, particles, and physics via the company’s in-house engine. The first test, Ice Storm, runs at a fixed 1280x720 resolution, while the second, Ice Storm Extreme, increases the resolution to 1920x1080. Finally, Ice Storm Unlimited renders the scene off-screen at 720p.
While it was just released in late May of this year, and is updated quite regularly, 3DMark is already being outpaced by more recent chipsets, with Nvidia’s Tegra 4 and Qualcomm’s Snapdragon 800/Adreno 330 both easily maxing out the Extreme (1080p) benchmark.
One more thing: although we have scores for HTC's One in this test, Futuremark delisted the device for deliberate application detection. In short, this is one test where cheating is confirmed. We are deliberately running an extensive suite of tests to help mitigate some of the complex mechanisms by which device vendors are optimizing, but in this case, the infraction is called out beforehand. The other devices check out in Futuremark's results list.
Now this is more like it. It's a test that Adreno 330 and the GeForce ULP inside the Tegra 4 consistently max out. Apple's PowerVR-based graphics engine should be doing better, but surprisingly, it seems to be held back by the A7's CPU core, and only specifically in this test.
With that said, why is the LG G2 not maxing out this test as well? It uses the same Snapdragon 800 as the Nexus 5, so what's holding it back? Android 4.2.2? A slower GPU core than it should have? Perhaps we need to think of hitting this test's ceiling relatively. If maxed out means anything over 11,000 points, the LG G2 and Apple iPhone 5s really only fall slightly below that line.
Again, our Nexus 5 takes the lead, beating the Tegra 4 on everything but Physics. The G2 is up with the big dogs as well.
Interestingly, Apple's iPhone 5s is slow here too. Futuremark went to some lengths to explain why, but essentially the A7 CPU Core in the iPhone 5s doesn’t seem to handle the Bullet Physics Library and its use of “complex data structures” well.
Rightware is another benchmarking stalwart with a well known toolset, and Basemark GUI is the company’s Android UI performance test. It’s designed to simulate the demands of intensive UI use in 3D (think applications like Google Maps). As with 3DMark, though, it is starting to be overshadowed by more modern SoCs and implementations of Android (since version 4.1, more and more of Android’s UI is rendered in hardware).
At native resolution, each device seems to perform as it should, though it does seem a little strange that the Nexus 4 can't seem to beat the others given its lower screen size, with far fewer pixels for its GPU to push.
Okay, now we're seeing a shake-up. Nvidia's Tegra Note flattens the competition (as it should given its form factor) and our Nexus 5 finishes in second place, only marginally ahead of the LG G2. HTC One should not be that slow in terms of hardware, so clearly it's just the native Android 4.4 (and in particular Project Svelte) of the Nexus 4 which is making the HTC look poor.
Basemark X is a multi-platform benchmark based on a real game engine, Unity 4.0. It uses many of Unity’s modern features via the OpenGL ES 2.0 render path, just as a modern game would. Features like high poly count models, shaders with normal maps, complex LoD algorithms, extensive per-pixel lighting (including directional and point light), along with a comprehensive set of post process, particle systems, and physics effects test how a modern game might look and run. It’s an aggressive test that still hasn’t been maxed out by the latest mobile SoCs.
Yes folks, this is a tough test indeed. The A7's Rogue-based GPU comfortably leads, but it’s still below 30 FPS, even with its substantially lower target resolution.
The Nexus 4's advantage comes from its smaller screen size in this native resolution benchmark. LG's G2 should be right up there with it, so this is no doubt due to the difference between the Nexus 5's Android 4.4.2 working in its favor.
Now we're getting somewhere. When the playing field (resolution) is level, Apple's Rogue-based GPU can't just walk away with the crown. Our Nexus 5 puts in a strong showing with the equivalent LG G2 not that far behind.
It's disappointing that the two 720p-level devices perform so poorly, but HTC's One gets the worst of it. We’re pretty sure that the Sense UI isn’t doing it any favors here.
- Redefining The Android Experience With Google's Nexus 5
- Product 360: Look And Feel
- GEL: A Better Experience
- GEL Gets Personal
- Benchmark Variance: Not Every SoC Is Created Equal
- Test Setup And Methodology
- Results: CPU Benchmarks
- Results: GPU Benchmarks
- Results: GPU Benchmarks, Continued
- Results: Web Browsing Benchmarks
- Results: Display Measurements
- Results: Battery Testing
- Does The Nexus 5 Raise Expectations?