Hardware Dyslexia, or How Specifications Sheets Are Not Straight-forward

With each new device leak and spec announcement we see product-cycle hype rise. Those who frequent enthusiast circles typically want the best they can get out of their favorite phones, and in the last few years, the amount of discussions regarding the cameras and processors of flagship devices has seemingly outgrown sensible expectations. When you factor in the fact that 2015 has been a rather stagnant year in terms of flagship hardware, you get a recipe for disaster.

More Hardware, More Variables

The main issue I see is that there is a lack of holism when interpreting specifications. Admittedly, I feel like this is more of a current phenomenon than anything. Various technologies such as AMOLED have branched away and evolved at a different pace than others in the same category. This year in particular, we saw a much more diverse repertoire of SoCs in flagship devices due to the Snapdragon 810's disappointments. The Exynos 7420, the 810, the 808, the 801, the Atom Z3580 and the Helios X10 were some of the names we saw hit our most awaited phones, each with a set of strengths and weakness that gave them vastly different results.

This becomes a problem when you take into account that simple specification sheets and their values do not represent the entirety of the hardware component and its inner workings, but rather a few characteristics that are easily digestible and marketed. By now, most enthusiasts know that a megapixel count says little about image quality. Many mainstream consumers are aware of this too — otherwise, the iPhone wouldn't have the "great camera" badge in the collective subconscious. So while we now know that 20MP does not necessarily mean good pictures, a lot of specifications still entice us when in reality, we shouldn't be looking at the numbers themselves but the package as a whole, and the relationships between those numbers.

Processors, Screens, Batteries…

An example I always find myself bringing up when discussing phone performance with friends has to do with processors, frequencies and cores. If you take a look at the Snapdragon 805, the term "quad-core" and the frequency "2.7GHz" sound mighty impressive. They should, because this is a very impressive processor indeed. My Note 4 is still an immensely powerful device, but so is its Exynos counterpart. This one has an Exynos 5433, with a very mighty "octa-core" term and a not-so-impressive "1.9GHz" frequency. The average consumer can sometimes be carried away with these numbers and fill in the blanks of his knowledge with assumptions as to how they relate, and this results in something I've taken to call "hardware dyslexia".

As with most things, the devil is in the details, and the details in this case are the things that you won't find listed on your typical carrier store. The Snapdragon 805 and the Exynos 5433 have remarkably similar performance output in important categories despite those different numbers. The Snapdragon 805 uses Qualcomm's custom "Krait" cores, while Samsung opted for a more vanilla ARM Cortex solution. Moreover, the "octa-core" term is misleading, as the big.LITTLE architecture typically has 4 powerful cores and 4 energy-efficient ones, and the packs typically act in isolation unless homogeneous multi-processing takes place. But that solution came much later than octa-core phones; owners of the Exynos S4 and Note 3 should remember some heartbreaking news.

The processor takeaway: There are many factors that influence performance in and out of hardware. Benchmarks do help predict the theoretical potential of the hardware, but the potential does not always translate to the real world. Many factors, such as software, throttling and bottlenecking, can greatly alter the results. When it comes to user experience, numbers should never be blindly followed.

This also takes us to screens: with leaks, we typically only hear about the screen size and resolution. To me, learning these things is completely useless as far as predicting image quality goes. Full HD is still good for most people so I do not get disappointed when this shows up, nor classify the screen as lowly, but excluding that, one cannot even rely on OEMs' past history and promises. An example would be the M9, which surprised everyone by having a worse screen than the M8 in metrics like brightness, contrast and even saturation accuracy, when the M8 was known to be one of the finest LCD panels in these same areas. LG's promises of accurate color reproduction in the LG G4 were proven inaccurate by technical reviews, which put it behind the pack leaders in both saturation and GMB accuracy. Quoting AnandTech:

"As one might have guessed from the previous two results, the end result for the LG G4 is that color accuracy isn't quite as good as one might have hoped. Due to an excessively wide gamut and a cold color cast, the G4's display ends up slightly disappointing."

Then you have the fact that many reviewers and consumers still believe Samsung screens and AMOLED panels as a whole are very saturated, when in reality these panels have made tremendous leaps forward, and reputable screen analysis site DisplayMate has called the screen of both the Note 4 and Note 5 the best screen they have tested.

The screen takeaway: look for empirical luminosity, contrast, and color accuracy measurements rather than buzzword panel names.

Battery size is also a specification people wildly rave about, when in today's mobile world, optimization can trump high capacity. We've seen many devices with ridiculously large batteries underperform (like the Droid Turbo), and devices with smaller batteries perform well. This is because every other component matters – screens, processors, modems, audio chips, and the software that optimizes each of these components. When the battery size of the Note5 was revealed, many were disappointed as it was not only lower than the 4,100mAh we had seen rumored, but 220mAh lower than the previous Note. I too had my doubts, though Samsung's screens have become more power-efficient since the Note 4, as has the Exynos 7420 (in part due to its class-leading 14nm FinFET manufacturing process). Now that I have the phone in my hand, I find myself surprised that I am able to squeeze over 5 hours of screen usage on a full day out of a 3,000mAh li-on trooper.

The battery takeaway: real-world battery tests that match the way you yourself will use the phone, or that indicate the device's potential for improvement with future ROM and governor advancements, are far superior to flat capacity specs.

This year, we even saw the disconnect between specs and real-world performance extend to power cords. The OnePlus 2 was touted to have a USB Type C plug, something which was considered a universal step forward. Reality came crashing down once more when we learned that the plug is new, but the USB standard is not (that is, it does not come with USB 3.1). To make things worse, no quick charging is found in the device despite it having a Quick Charge 2.0 ready chipset… Qualcomm's flagship, at that. The end result is a marketing campaign that touts progress while the phone falls as flat as its new cable.

All The Pieces Matter

What I am getting to is that specifications by themselves mean little, particularly the popular numbers, terms and names we see released and leaked every other day. Without context, these numbers are little more than marketing gimmicks in leak form. Especially when we know that:

• The best processor and RAM do not make for a smooth user experience by themselves, as shown by the Note5 (more on this in an upcoming review)
• More cores and higher frequencies are not necessarily superior, nor the only way to improve performance (something the Exynos line has shown time and again).
• A Quantum Display is not going to be better-looking through promises, and traditions do get broken as technology evolves.
• Wishful thinking won't make a new port perform better than an old one, not even if details are hidden away until people get a hands-on.
• A big battery can be crippled by all of the above, and a small battery can last a long time with the right set-up. Examples of all of this are everywhere.

Yet we still get excited at learning about a battery size, or a screen resolution. I believe that numbers by themselves mean little to nothing, and that they must not only be contrasted, but also considered within the whole. That is, there are many variables to consider, and optimizations in one place can trickle down to another. There is a lot that specification sheets don't mention, and manufacturers can design a phone's internals in such a way that they intelligently synergize to bring out an unexpected package. Software also plays a major role, and we all know a bad ROM can kill your battery and a good one can save your day.

We all commit some hardware dyslexia at some point, though. Especially for devices we cannot wait for — we grab every new detail and hold it dear, expecting the best out of it. Sometimes we are disappointed, but it can go both ways. That small battery might end up lasting you all day, and that non-flagship processor might give you a hell of a fast user interface. Numbers are numbers, and while higher ones are usually better, they don't imply that the end result – a holistic one – is ultimately superior. In many ways, I've learned to ground my expectations and look at the connections rather than the numbers themselves, and not get carried away by the hype. It's hard, and with Nexus phones in the horizon, I personally have to try to remain objective and not go on the hype rollercoaster that comes with every anticipated release. Ultimately, we need to take off the fanboy goggles and realize that there's always more than meets the eye. And that's what I personally love about technology, so I hope you will find some of that in my future pieces.

 

source: xdadevelopers