As requested, some impression on using Viper4Android with the S5:
First, some comments on the ease of use. Getting V4A FX working on the S5 required the following steps (I am specifically using the Sprint variant, but it sounds like the instructions are the same for all versions of the S5):
2. Install V4A as a system app
3. Install V4A driver from app (I used the Super Quality version)
4. In the Sound section of the settings menu, set the music effects processing to V4A
5. Edit build.prop to disable low-power audio tunneling (more on this in a bit)
After performing the previous steps and rebooting, V4A is now ready to use. I could get it to work in PowerAmp, but it could not get it working in the stock Samsung app. In fact, if I tried playing music with the Samsung music app, V4A would stop working altogether (including in PowerAmp), and would only start functioning again after a reboot.
I also mentioned earlier that using V4A requires you to disable low-power audio (LPA) in your build.prop file. For those not aware, LPA is a new feature that allows Android to directly offload sound processing to a hardware DSP. This has two potential advantages: (1) Audio processing on the DSP is typically more energy efficient than the CPU. Now for most people this may not be a big deal, as sound would make up a relatively small percentage of the total energy use compared to the screen and radios. However, if you are someone who spends the entire day listening to audio on your phone, it could have a noticeable impact on battery life. One website I found claimed that with LPA enabled they were able to get a 50% increase in battery life when using their phone only as a music player. (2) Having the CPU do audio processing could potentially affect the CPU's responsiveness for apps. For basic audio processing I don't think this will be very noticeable on a device as powerful as the S5, but if you are using particularly complex filters, maybe it could make a difference. Personally, I didn't feel a difference even when using the Super Quality driver. You can always use the High Quality or Low Power drivers if you feel like you need to mitigate either of these effects.
Now onto the impact on sound quality. I experimented with different settings while using my AUD-5X IEMs. For reference, they have a sensitivity of 118 dBm and an impedance of 48 ohms. Let me preface my comments by saying that I am not someone who particularly likes to mess around with effects on any of my sources. Personally I find that having too many small things to tweak causes me to get preoccupied with technical minutiae and gets in my way of enjoying music. That is to say, I am a novice when it comes to configuring the more advanced features of V4A. If you are a V4A pro, your experience may vary.
AGC: Some people have expressed concern over the output level of the S5. Enabling automatic gain control in V4A produces a significant increase in output levels, even when leaving the effect level at Slight and the maximum gain at 2X. Whereas most music got too loud for me at around 60-70% max volume, I can reach the same level at 40-50% volume with AGC enabled. AGC does introduce an audible noise/hiss to the sound output, but the magnitude of the hiss is low enough that I can really only make it out if the volume is set around 10%, so it is inaudible during normal listening, except for the moments of silence between songs. Still, since I don't have a problem with the base output level, I prefer to leave AGC turned off.
Equalizer: It's a software equalizer that works similar to any number of other equalizers.
Bass Boost: It achieves greater bass separation and texturing. If you are looking for a more fun sounds, it is worth turning on.
Clarity Boost: The idea here is to achieve greater transparency and separation in the midrange. The effect definitely does achieve more separation, but introduces audible noise. Even with 'Natural' setting and a 3.5 dB effect level, I could notice the noise during quite parts. I tried 'XHifi' setting and 6.0 dB, as suggested by someone on a forum, but this produced noise that was audible even at normal listening levels. I keep this setting disabled.
Convolver: I didn't have any IR data handy, so I didn't test this.