Hello joshuaronis and welcome to Sound Design SE.
Your problem is related to the drivers of your sound card as well as to the capabilities your computer has. Reading the thread from the link you attached I have to say that those people there explained quite well what is going on and what the possible solutions would be.
Generally speaking, and without going into too much detail, the rule of thumb on latency values is that the greater the buffer used, the greater the latency from input to output will be (and most probably any other processing related to that).
It seems that the software you are using is not capable of changing the sampling rate or the buffer size used by the drivers. You could possibly do that from the Windows audio control panel as is show in the thread of the link. Additionally, as stated there (in the link), the supported sampling rates depend on the hardware used (and quite rarely on the drivers - most probably this is not an issue nowadays). This means that, if your hardware does not support sampling rates other than 44.1 KHz and 48 KHz, then you are stuck with those.
Now, regarding the buffer size and latency, it seems that you can tweak those from the panel you present in the second picture. It seems that the slider refers to the latency (in ms) and not the buffer size, which may prove to be easier to use for you (since you seek a short delay).
The first value of buffer size refers to Pulse Code Modulation (PCM) coding while the second value refers to Delta Sigma Demodulation (DSD) coding. Without trying to explain those here (you can create a different question for that, or even better ask at DSP SE), in the former (PCM) samples are taken every 1/sampling-rate and they are quantised based on the bit-depth used (16-bit for CD quality, while 24-bit is quite often seen nowadays) and in the latter (DSD) the sampling rate used is very high (2.8224 MHz in your case as it seems) with one bit of coding. This one bit is used as the result of a comparison of the current analogue amplitude compared to the one of the previous sample (one if current amplitude is higher, zero if lower) and consequently, the values of "packets" of samples are summed (this is what the Sigma means in the DSD acronym) and from those a digitised PCM-like value is calculated.
Please keep in mind that this is a very vague explanation of each technique and a whole lot of details are omitted (on purpose) here. DSD has certain advantages against PCM like noise-shaping and ease of hardware manufacturing, while it has some drawbacks related to the unbounded error due to the summation operation (I do not provide references here but a quick search on IEEE or AES would provide some papers on all of those topics).
To conclude about your problem, you should try to minimise the value shown in the second picture until you get the lowest latency values without getting dropouts (clicks and glitches). I am not sure you should bother too much about the two different sampling rates as they are most probably calculated internally from the given values (and the sampling rate chosen from the Windows control panel, or any other place where you can control your hardware from).
Finally, getting an external audio interface would most probably improve audio quality, latency and possibly processing capabilities (as some of the processing will be moved to the interface hardware and the drivers will be tailored to your interface). Whether this is needed or not is up to you to decide. I would strongly suggest trying your luck with your current on-board audio interface and if you cannot work with that move to an external one. Alternatively, you could possibly borrow one and test it before buying one, to avoid finding out that what you bought is does not provide a solution to your problem.
