In my own experience I have personally never had a motherboard that didnt suffer from pops, clicks, hisses or other weird noises.
Not necessarily the board's fault. Finding a case where the front panel does not introduce any ground loops does not seem to be trivial. I've had good luck with some C2D era Fujitsu-Siemens Celsius machines, so maybe those guys care for their audio at least. Other prebuilts, not so much. (And it's not just prebuilts. I've seen complaints about rather expensive cases, too, like some Lian Lis if memory serves.) This is what I just found on a ca. 2006 Asus midtower:
Audio ground from HDA cable connects to one patch of ground, which then goes to chassis ground via the LHS retainer (screw - bracket - screw).
Headphone and mic jack ground joins USB and misc switch / LED ground, going to chassis ground via the RHS retainer.
In practice, this means there is a ground loop running over the front panel audio cable, and worse yet, headphone ground return is shared with USB and whatnot, with some more ground loop action.
Boards by themselves can be messed up, too, of course - subpar grounding schemes, sharing 3.3 V between DVDD and AVDD, less than ideally clean supplies...
Implementation is key with these things.
Now I don't know about modern chips, but one notoriously weak spot of Realtek codecs has always been the mic input. More or less noisy mic bias, usually covering up that the amplifiers themselves are far from state of the art in input noise either. (I know that these chips are supposed to eliminate external components and conserve board space as far as possible, but one capacitor would go a long way in solving this problem.) Periodic ripple in the ADC digital filter also tends to be at rather uninspiring levels in all but a few chips. Basically, everything that doesn't reflect in big numbers in the spec sheet is rather neglected. Recording in multiples of 44.1 kHz does not seem to be a good idea either, it seems to result in massive amounts of jitter with substantially degraded SNR and aliasing. The playback side is far more robust.
In any case, providing a decent line-level output for a headphone amp or similar is the area where I'd see the best chances of onboard audio working well. On the recording side, make sure you stick with multiples of 48 kHz (don't forget to set up sound devices properly).
Many of them can also not output 2VRMS.
I would even say
most. Typical output ranges from 1.0-1.2 Vrms on 5 V AVDD and about 0.8 Vrms on 3.3 V AVDD.
I also passionately hate Realteks drivers in a way that I cannot overstate, getting a DAC with better drivers can save you a lot of headaches.
Oh. What sort of issues have you had with them? The most serious problem I ever found was sample rate being set up all wrong when selecting 44.1 kHz in driver versions from about 2011, which was subsequently fixed. (Generic implementation also seem to be limited to 16 bit samples on the recording side even if the ADCs themselves would definitely support 24 bit, which is silly.) Mind you, I have little experience with their latest chips and Microsoft's bug collection version 10... there have been a few changes and issues with sound drivers in general since 7.
I've seen a lot of sound driver troubles over the years, and honestly, the Realtek drivers would seem to be working pretty well to me. At least there are no weird suspend / hibernation issues, that sort of stuff is a total dealbreaker for me. I still have an ESI Juli@ floating around - a no doubt good and unique card that just stops working after hibernation.
And don't even get me started on the hoops I had to jump through to get my Asus Xonars to produce the sample rates I wanted, or theiir other driver quirks. (Hint - for Xonar D1 & D2, best use driver builds 1800 or older.)