Just found out about this topic. M2C...
What is the target product here? The X7? Or some other product? For now, I'll assume we're addressing the design features of the X7...
With respect to power saving, from spec sheets, the ES9018S consumes 100 mW of power during operation, while the ES9018K2M consumes 40 mW, with a 10 mW low power state. If we go with two of them, then that's 80 mW operational power, and 20 mW in low power. How will that reflect in terms of battery life? If we're talking about the X7, and it indeed is an Android-based device with a large ~4" capacitative touchscreen, then it's pretty important that we try to save power anywhere possible.
In fact, for power saving measures, I'd even go as far as to suggest a separate and independent "mid-fi" mode in the X7, where the main circuit is perhaps derived from the X1, with a PCM5122 or ES9016K2M and perhaps a lower-power chip amp (e.g. MAX97220A, ES9601, TPA6130A2, etc.) that provides decent sound quality at a fraction of the power consumption, for times when SQ is not paramount, i.e. loud environments that prevent a user from doing critical listening. Then, when desired, the user can switch to "hi-fi" mode to consume more power, and with balanced output.
With all that being said, I agree with the sentiment that designs shouldn't be confined only to the choice of ESS chips merely because they're a popular choice and that audiophiles tend to have a positive opinion of them. Personally, I find that well-designed PCM1792A topologies compete every bit as well against ESS designs.
The key phrase here is well-designed. Anything can sound extremely good with any high-quality DAC chip, high-quality I/V conversion, and high-quality amplification stage, and precise power regulation. The size of the X7 needs to be informed by the parts that are necessary for high-end performance centered around a SMD-centric board layout. With SMD components front and center, the power supply rails should be highly regulated and consequently low-impedance, with parts specified for least ripple effect. The LPF design should contain capacitor components just big and robust enough to sound good, but not unnecessarily big (like some brands like to do, by bragging that they use desktop components in portable gear, even though the benefits, if any, are near imperceptible.)
The performance specs of most top-end DAC chips aren't usually met in the majority of designs. If you can achieve >120 dB SNR, out of ground-centered 2 Vrms outputs, you're already at an elite level ---- a level where most manufacturers charge >$1000 for the product in question. If you can achieve <0.0008 THD+N out of your headphone outputs, then you're at an elite level. So, using whatever flagship level DAC chip, regardless of whether it's the ES9018S, ES9018K2M, PCM1792A, AK4399, WM8741, etc. is already superfluous. The question is layout and design. There is the added question of the choice of digital filters built-in to each DAC product --- some use traditional FIR linear-phase filters, others give the option of using IIR minimum-phase filters, and with slow or hard roll-off variants. Choosing between these filter designs is pretty important as well, unless FiiO chooses to implement their own digital filter (which can be costly for R&D).
Thus, I believe the choice between the ES9018S and (2×) ES9018K2M should be a design and layout issue, rather than a feature-related issue. I think the bigger question is whether it's easier to rid of DC offset with decoupling capacitors in the 9018S or 9018K2M --- will the decoupling path be short enough? Will the smaller form factor of the 9018K2M allow for more advantageous placement of clocks? Will the lower operating voltage of the K2M result in a greater or lesser ability to reject power noise from the AVcc? These are the questions that I'm sure FiiO engineers are looking to answer for themselves when testing different design paradigms, and I believe they should be free to make those decisions free from market pressure.
