-
After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.
When high specs attack - why a you should aim for that "sweet spot."
- Thread starter Medion
- Start date
SiempreTuna
Android Expert
Very interesting.
I thought you were going to be arguing that actually, CPU performance is about as relevant in phones as it is in PCs these days, i.e. completely irrelevant to the vast majority of buyers.
But this was way more informative
I thought you were going to be arguing that actually, CPU performance is about as relevant in phones as it is in PCs these days, i.e. completely irrelevant to the vast majority of buyers.
But this was way more informative
Medion
Android Expert
I wanted to further highlight my points with a follow-up. I got some good hands on time with a Nexus 5, but didn't have a third device to take comparative side by side video with. I apologize for that. I did two main benchmarks to highlight my main point, while also getting some general impressions.
General Impressions:
There were few things that either phone did noticeably better than the other. The UI was fast and fluid throughout on both. Both myself and the Nexus 5 owner felt that the display on the Moto X was superior, but that's VERY subjective. But aside from vibrancy, neither phone's display offered an apparent reason to take one over the other. Apps and 3D Games looked largely the same in terms of their sharpness. Neither of us could detect a noticeable difference. So in terms of performance and quality, both phones were largely the same despite dual-core 720p vs. quad-core 1080p. And that was my primary point,. That the higher specs don't make an appreciable difference for the average person, but they certainly murder that battery faster.
But I did state some differences, and I wanted to highlight those. We ran a compression/decompression test with ES File Explorer, and a decryption test with KeePassDroid using a database set to 3 seconds worth of transformation rounds (over 150k) on a Core 2 Quad Q6600. In both cases, the Nexus 5 finished its task nearly twice as fast as the Moto X. This is to be expected as these are truly parallelized tasks. It should be noted that 1) Almost no one does this in real world usage (and most of the forum readers here have likely done neither), and 2) the Moto X finished in a satisfactory time.
The second difference was what I deem to be poorly coded apps. Anyone who has used the Costo app knows that the Warehouse Deals section is a struggle for a device (unless Dev Options --> Force GPU Rendering is checked). We did it without dev options to get the casual usage scenario. Like my S4 GPE, the Nexus 5 was not a truly smooth experience, though it was superior to that of the S4 GPE. The Moto X was noticeably smoother, though still not iPhone smooth. Enabling the Dev Options makes both devices equally smooth, but most users won't do that. And a poorly coded app as big as Costco does represent something that the average user will be exposed to.
Benchmarks:
I chose two tests here to highlight vastly different points. Please allow me to explain.
First was 3DMark using the Unlimited benchmark. This counts the number of total frames through a battery of tests. This is a composite score (like Quadrant and AnTuTu), but focuses more on things that would matter to a game, specifically the CPU and GPU. It's rendered offscreen at a specific resolution regardless of the device's screen resolution.
In terms of raw horsepower measured in DMIPS, we're comparing a 2.2 ghz Krait 400 quad-core (3.39 DMIPS/mhz) to a 1.7ghz Krait 300 dual-core (also 3.39 DMIPS/mhz). So in raw horsepower, it's 29,832 vs. 11,526. In other words, the Nexus 5's GPU is approximately 2.6x more powerful than the CPU in the Moto X. The Nexus 5's Adreno 330 GPU is also advertised as being up to 50% faster than the previous 320 (Qualcomm doesn't disclose which 320 as there are two generations of the product). Bottom line is that the Nexus 5 should have been significantly faster than the Moto X in this benchmark. It wasn't and I suspect thermal throttling to be the culprit.
Nexus 5 won, 13,558 to 10,907, a margin of 24%.
The second benchmark, Epic Citadel was meant to show how a typical game would play on these devices. Most games use a device's native resolution rather than a user selectable one (though some games have this). Again, despite the more accurate colors and higher resolution, both phone owners in this test subjectively preferred the way that the game looked on the Moto X. But, how did they perform? Well, the Adreno 330 may be 50% faster, but it has to push more than twice as many pixels. So you can probably guess. The Moto X offered a smoother, more consistent experience. Due to Vsync limiting both games to 60fps, the dips below this were jarring and noticeable. They rarely affected the Moto X, which finished with an average framerate of 58.7. They were very noticeable on the Nexus 5, which finished with an average framerate of 49.0.
Conclusion:
The Nexus 5 is more powerful. There is no denying that. But the lower specs of the Moto X, due to thermal throttling, common usage scenarios, and higher overhead cause a device like the Moto X to be comparable or better for the average user. Or to put it plainly, Motorola took the iPhone approach of marrying specs to each other and the software for a superior user experience.
And one final note - while I didn't conduct such specific measurements, one of my links in the first post points out that they were able to take thermal measurements and measure the clock rate of the CPU during tests. While the MotoX would remain at 1.7ghz on both cores throughout the duration of the tests, the Snapdragon 600 device test quickly went down to 1.0ghz for the duration of the test due to heat. Yes, a 1.7/1.9ghz Snapdragon 600 is faster than a 1.7ghz Snapdragon S4 on a test bench, but inside of a phone's chassis, that theoretical speed is useless. It becomes 1.7ghz vs. 1.0ghz.
General Impressions:
There were few things that either phone did noticeably better than the other. The UI was fast and fluid throughout on both. Both myself and the Nexus 5 owner felt that the display on the Moto X was superior, but that's VERY subjective. But aside from vibrancy, neither phone's display offered an apparent reason to take one over the other. Apps and 3D Games looked largely the same in terms of their sharpness. Neither of us could detect a noticeable difference. So in terms of performance and quality, both phones were largely the same despite dual-core 720p vs. quad-core 1080p. And that was my primary point,. That the higher specs don't make an appreciable difference for the average person, but they certainly murder that battery faster.
But I did state some differences, and I wanted to highlight those. We ran a compression/decompression test with ES File Explorer, and a decryption test with KeePassDroid using a database set to 3 seconds worth of transformation rounds (over 150k) on a Core 2 Quad Q6600. In both cases, the Nexus 5 finished its task nearly twice as fast as the Moto X. This is to be expected as these are truly parallelized tasks. It should be noted that 1) Almost no one does this in real world usage (and most of the forum readers here have likely done neither), and 2) the Moto X finished in a satisfactory time.
The second difference was what I deem to be poorly coded apps. Anyone who has used the Costo app knows that the Warehouse Deals section is a struggle for a device (unless Dev Options --> Force GPU Rendering is checked). We did it without dev options to get the casual usage scenario. Like my S4 GPE, the Nexus 5 was not a truly smooth experience, though it was superior to that of the S4 GPE. The Moto X was noticeably smoother, though still not iPhone smooth. Enabling the Dev Options makes both devices equally smooth, but most users won't do that. And a poorly coded app as big as Costco does represent something that the average user will be exposed to.
Benchmarks:
I chose two tests here to highlight vastly different points. Please allow me to explain.
First was 3DMark using the Unlimited benchmark. This counts the number of total frames through a battery of tests. This is a composite score (like Quadrant and AnTuTu), but focuses more on things that would matter to a game, specifically the CPU and GPU. It's rendered offscreen at a specific resolution regardless of the device's screen resolution.
In terms of raw horsepower measured in DMIPS, we're comparing a 2.2 ghz Krait 400 quad-core (3.39 DMIPS/mhz) to a 1.7ghz Krait 300 dual-core (also 3.39 DMIPS/mhz). So in raw horsepower, it's 29,832 vs. 11,526. In other words, the Nexus 5's GPU is approximately 2.6x more powerful than the CPU in the Moto X. The Nexus 5's Adreno 330 GPU is also advertised as being up to 50% faster than the previous 320 (Qualcomm doesn't disclose which 320 as there are two generations of the product). Bottom line is that the Nexus 5 should have been significantly faster than the Moto X in this benchmark. It wasn't and I suspect thermal throttling to be the culprit.
Nexus 5 won, 13,558 to 10,907, a margin of 24%.
The second benchmark, Epic Citadel was meant to show how a typical game would play on these devices. Most games use a device's native resolution rather than a user selectable one (though some games have this). Again, despite the more accurate colors and higher resolution, both phone owners in this test subjectively preferred the way that the game looked on the Moto X. But, how did they perform? Well, the Adreno 330 may be 50% faster, but it has to push more than twice as many pixels. So you can probably guess. The Moto X offered a smoother, more consistent experience. Due to Vsync limiting both games to 60fps, the dips below this were jarring and noticeable. They rarely affected the Moto X, which finished with an average framerate of 58.7. They were very noticeable on the Nexus 5, which finished with an average framerate of 49.0.
Conclusion:
The Nexus 5 is more powerful. There is no denying that. But the lower specs of the Moto X, due to thermal throttling, common usage scenarios, and higher overhead cause a device like the Moto X to be comparable or better for the average user. Or to put it plainly, Motorola took the iPhone approach of marrying specs to each other and the software for a superior user experience.
And one final note - while I didn't conduct such specific measurements, one of my links in the first post points out that they were able to take thermal measurements and measure the clock rate of the CPU during tests. While the MotoX would remain at 1.7ghz on both cores throughout the duration of the tests, the Snapdragon 600 device test quickly went down to 1.0ghz for the duration of the test due to heat. Yes, a 1.7/1.9ghz Snapdragon 600 is faster than a 1.7ghz Snapdragon S4 on a test bench, but inside of a phone's chassis, that theoretical speed is useless. It becomes 1.7ghz vs. 1.0ghz.
