CPU And System Performance
In this section, we evaluate system-level performance by running a series of synthetic and real-world workloads, along with some browser-based Web tests. There are several facets to overall device performance, including single- and multi-threaded CPU performance, memory and storage speed, and GPU rendering, all of which will be probed by our suite of benchmarks. If you're interested in learning more about how these benchmarks work, what versions we use, or our testing methodology, please read our article about how we test mobile device system performance.
The Nexus 6 and Galaxy Note 4 both use Qualcomm's Snapdragon 805 SoC, with four Krait 450 CPU cores running at up to 2.65GHz. There's also an Adreno 420 GPU on-board paired with two 64-bit channels of LPDDR3-800 RAM for a total of 25.6 GB/s of throughput (the same peak bandwidth offered by the newer Snapdragon 810 and Exynos 7420 with LPDDR4 RAM). Based on hardware specs alone, the two phones should perform similarly. Differences in software or elsewhere in the design could upset this equilibrium, however.
For the overall score in Basemark OS II (the number shown in black on the far left), we do see nearly equal performance between the Nexus 6 and Note 4. The Web test is basically the same, but the Nexus 6 holds a 24% advantage in the single- and multi-threaded CPU System test. Positions flip in the Graphics GPU test, where the Note 4 outscores the Nexus 6 by 18%.
The Memory test, which is meant to measure the speed of the internal NAND storage, has morphed into a RAM test. In order to ensure the benchmark would work across a wide range of devices, the dataset had to be kept small. However, in higher-end devices the operating system uses a RAM cache as a buffer to speed up reads and writes to internal storage. For modern flagships like the Nexus 6, this memory cache is large enough to hold the benchmark dataset and it never actually hits the NAND. Therefore, the Memory test represents the speed of the memory cache rather than NAND, and it's why the results do not agree with those from the AndEBench Storage test below.
In the CPU-centric CoreMark-HPC test, we see the Nexus 6 pull ahead of the Note 4 again, albeit by a slimmer 12% margin. We also see the Note 4 post a higher 3D graphics score, with a similar 16% margin.
Internal storage performance is where the Nexus 6 really falters. Overall, it's 2.4x slower than the internal NAND in the Note 4. The table below shows the breakdown of the storage test.
AndEBench Pro Storage Test
Header Cell - Column 0 | 512B SW | 512B RW | 4KB SR | 4KB SW | 4KB RR | 4KB RW | 256KB SR | 256KB RR |
---|---|---|---|---|---|---|---|---|
Galaxy Note 4 | 508 | 1010 | 20781 | 3072 | 20722 | 8078 | 97075 | 96908 |
Nexus 6 (5.0) | 249 | 609 | 5905 | 1708 | 5968 | 6195 | 21194 | 26086 |
Nexus 6 (5.1) | 163 | 702 | 8211 | 1041 | 6779 | 6587 | 30966 | 33187 |
*First letter: S=sequential, R=random
**Second letter: R=read, W=write
***Values in KB/s - Higher is better
For the numbers in the table, the Note 4 was running Android 5.0.1 (update arrived after these charts were made), although the scores did not change significantly compared to KitKat. Write performance for the Nexus 6 running the original Android 5.0 build hovers around one-half that for the Note 4, while read performance takes an even larger 3.5-4.5x hit. After we completed all of our charts (of course), the Android 5.1 update for the Nexus 6 appeared, which is supposed to improve storage performance. Overall, we do see some modest gains for reads (now only 2.5-3.1x slower than the Note 4) and a couple of regressions for writes. The read values did fluctuate a bit more than usual, with the 256KB reads ranging from 18761-38985 KB/s. The relatively small differences in CPU and GPU speed shown above should not be noticeable in everyday use, but the Nexus 6's reduced storage performance definitely impacts app load times.
The cause for the low storage performance is Android's Full Disk Encryption (FDE), which has been an option since Honeycomb. This is no longer an option for the Nexus 6 however, since it's enabled by default. The snag is that the Nexus 6 performs the file encryption/decryption on the CPU rather than dedicated hardware, which hurts performance and battery life. At least your files remain secure if the device is lost or stolen.
The Nexus 6's CPU performance advantage over the Note 4 disappears in Geekbench 3. Whatever small difference there is, its origin is elusive. There does not seem to be a clear pattern between benchmarks, whether they are single- or multi-core, integer or floating-point, or stress cache or memory bandwidth.
It's also interesting to see the older Snapdragon 800, running four Krait 400 CPU cores at 2.26GHz, perform nearly as well as the 2.65GHz Krait 450 cores. In the Single-Core test, the Snapdragon 805 is only 9% faster for integer operations and 4% faster for floating-point. Krait 450's 17% clock rate advantage doesn't translate into much higher performance.
PCMark is a new addition to our benchmark suite, which is why we have limited data for comparison. Unlike our other tests, PCMark is based on real-world use cases rather than being purely synthetic or algorithmic.
Here we see little variation in terms of the total score; the Nexus 6 only scores 14% higher than the LG G3. Looking at the individual workloads shows more variation. The Nexus 6 performs well on the Writing test, but falls behind on the Video Playback test.
We see similar performance between the Nexus 6 and Note 4 across our browser-based tests. The Nexus 6 manages a meager 15% improvement over the Nexus 5, basically reflecting the clock rate difference between the two.
In terms of CPU and system performance, the Nexus 6 does not offer much more than the Nexus 5. Its small gains will be hard to notice in everyday use. What will be noticeable, is the lower internal storage performance due to FDE. Informal testing relative to the Note 4 showed apps launching anywhere from under a second to several seconds slower on the Nexus 6.