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.
We’re including the HTC Desire Eye in our performance charts for the same reasons we included it in our discussion of camera quality. Plus, it will be interesting to see the difference in performance between last year’s Qualcomm Snapdragon 801 (4x Krait 400 @ 2.26 GHz, Adreno 330, 2GB RAM) and the new low-end Snapdragon 410.
In our first test, the Snapdragon 801 outscores the 410 by a margin of two in the single- and multi-threaded CPU System test and overall, which is pretty close to the difference in CPU frequency. The performance delta between the Adreno 330 (578MHz) and the Adreno 306 (400MHz) in the Graphics test is a more substantial 6.5x.
The overall score for the Moto E (2nd gen) is very close to the Moto G (2nd gen). However, the newer Cortex-A53 CPUs in the Moto E manage to outperform the Cortex-A7s in the Moto G by 19% in the System test. The Moto G does perform 28% better in the Memory test, which tests the speed of the internal storage.
Performance in AndEBench looks similar to what we saw in Basemark OS II, with the Desire Eye out front by a margin of ~2x and the Moto E very close to the Moto G.
Narrowing in on CPU performance, the out-of-order architecture of Krait 400, with nearly twice the clock speed, gives it about twice the performance of the narrower, in-order A53 in the Moto E. We also see the Moto E outperform the Moto G in CPU performance again, albeit by a slimmer 13% margin in CoreMark-HPC. Since both have the same number of cores running at the same frequency (1.2GHz), we can see the advantage of the A53’s architectural tweaks. Based on the A7 design, the A53 is still dual-issue, but the limitations on the second issue slot have been removed (the A7 cannot issue load/store or FP/NEON operations from its second issue slot). ARM made several other improvements to the A53’s front-end design, including improved conditional and indirect branch prediction and more instruction and data prefetching.
Despite having the same 4.2 GB/s of theoretical memory bandwidth, the Moto E gets 50% better throughput than the Moto G in the Memory Bandwidth test. Some of this advantage may be from the memory modules themselves, but a lot of it is probably due to the improved front-end design of the A53.
|AndEBench Pro Storage Test|
|512B SW||512B RW||4KB SR||4KB SW||4KB RR||4KB RW||256KB SR||256KB RR|
|Moto G (2nd gen)||213||514||10511||1625||8671||4065||42917||42979|
|Moto E (2nd gen)||184||353||6367||1176||6163||2756||44695||44572|
|First letter: S=sequential, R=randomSecond letter: R=read, W=write||Values in KB/s - Higher is betterFile Size: 5, #Folders: 3, #Files/Folder: 1|
The Moto G outperforms the Moto E once again in storage performance, this time by 22%. Looking at the test breakdown in the table above shows that the NAND in the Moto E performs better with larger blocks of data, actually performing about 4% better than the Moto G at 256KB. Still, these numbers are less than half of what we see in flagship phones, translating to slower app launch times.
The ~2x delta between the Snapdragon 801 and 410 shows up again in both the single- and multi-core Geekbench results. We also continue to see the Moto E outpace the Moto G: about 30% on average in single-core and about 20% on average in multi-core.
Performing well in synthetic tests is all well and good, but the results are not directly transferable to everyday use cases. This is certainly evident in PCMark where the Moto E outperforms the Moto G by only 8% and is nearly equal to the Desire Eye.
The reason for the discrepancy in scores between the synthetic tests and the more real-world PCMark has to do with the behavior of the CPU governor. Some OEMs will optimize for performance, routinely pushing the CPU cores to max frequency. Others will opt for a more balanced mixture of performance and battery life, limiting the CPU’s max frequency for less demanding workloads like those we see in PCMark.
The HTC Desire Eye uses a fairly aggressive CPU governor optimized for battery life. In the Writing and Web Browsing tests it more or less uses a single core clocked at a conservative 960MHz, with very short spikes up to 1958MHz. In the Photo Editing test, where we see a much wider gap between it and the Moto E, it uses a single core clocked at 1728MHz with frequent spikes to 2265MHz, its max clock frequency.
Performance scores fluctuate a bit more in the web browsing tests although the order remains the same. The Moto E is between 15-43% slower than the Desire Eye, but between 13-24% faster than the Moto G.
Our tests show the Moto E (2nd gen) pull ahead of the Moto G (2nd gen) in both CPU performance and memory bandwidth thanks to the improvements in its newer A53 CPUs. Its internal storage performance generally lags behind the Moto G by 20-25%, however.
Using it in the real world, the Moto E (2nd gen) definitely feels faster than the Moto G (2nd gen). This is quite a strange turn of events as now Motorola’s low-end phone is faster than its latest mid-range device. When the Moto G (2nd gen) was announced, our biggest criticism of it was that it did not include the A53 CPU, when other competitors’ devices did at the time.