First, Windows connection properties should report the adapter link rate. Also make sure you have the latest ASUS driver.
AC1900 produces the same 1300 Mbps 5 GHz maximum link rate as AC1750 class. 600 Mbps maximum 2.4 GHz link rate requires 40 MHz bandwidth. With 20 MHz b/w, max link rate is 289 Mbps, only a 33% gain from the 216 Mbps max link rate from standard 3x3 11n.
My experience is that AC1900 routers, in many cases do not achieve higher 2.4 GHz throughput than AC1750. Check the charts and see for yourself (use the Maximum selector)
http://www.smallnetbuilder.com/lanwan/router-charts/bar/111-2_4-ghz-dn-c?see=MAX
It makes sense. The cases where 256QAM for 2.4GHz or 5GHz seems uncommon from my limited experience. Well, I can't speak to 2.4GHz as I don't have 256QAM 2.4GHz client/base station to test with.
At any rate, testing with 5GHz with my AC1750 router and AC867 (I refuse to call a NON-concurrent dual band client AC1200) Intel 7260ac I am seeing numbers that are only a tiny bit above what you'd expect if you carried over 5GHz 40MHz 11n (IE 64QAM) to 80MHz width.
5GHz 40MHz, about 28MB/sec, absolute max average transfer I have seen with 5GHz 80MHz (obviously 11ac/256QAM has to be enabled for that), has been 62MB/sec, the extra channel width could possibly account for that as 80MHz channels should support about 110% of the bandwidth of 40MHz channels (or about 62MB/sec). Yes, the link rate claims 867Mbps, but I question if it is actively using the 256QAM encoding (or if it is, if there is some extra packet loss or something, taking throughput back down).
So I wonder what might be "up" with this. I don't think my situation is entirely unique. Based on AC1900 product testing, on 2.4GHz, the extra character encoding doesn't really seem to be doing much of anything, and in 5GHz 80MHz testing, the results generally don't seem to be much better than the ~110% or so throughput increase that going from 40MHz to 80MHz channel widths with the same 64QAM encoding should result in.
I mean, I won't look a gift horse in the mouth, more than doubled bandwidth with having twice the channel width is nice, but SOMETHING seems to be holding 256QAM back from showing real benefits for both 2.4GHz and 5GHz. I wonder if it is an issue of not having good enough amplifiers or something like that...which I wouldn't be suprised if that were the case. I wonder if once 11ac becomes more mature if both client and router manufacturers actually start focusing on differentiating and improving their existing products, if we might see products start really being able to maximize 11ac.
I at least THINK that happened with 11n (and even transitioning in to 11ac continuing).
Not too many years ago my Netgear 3500L (about 6-7 years old) could hit around 21MB/sec absolute max on 2.4GHz 40MHz (to an Intel 7260ac). Same client, same location, same tests and my TP-Link WDR3600 N600 2-3 year old router can hit 24MB/sec 2.4GHz 40MHz. With my AC1750 router (still 300Mbps connection speed) on 2.4GHz 40MHz it can hit 28MB/sec 2.4GHz 40MHz. Exact same client, exact same tests and locations, etc.
I wonder if we'll see a similar evolution among 11ac products now, were every new generation we see the products manage to eek a bit more out of what is theoretically possible. Right now, I see around 76% yield for 2.4GHz 40MHz 11n (300Mbps), which is about the max possible once FECC is taken in to account. For 5GHz 80MHz 11ac however, the max I see is around 57% yield, and the absolute max should actually be higher, as 11ac (last I checked) uses low density parity checking for FECC, which I think actually results in 11ac having an absolute max, by theory, PAYLOAD rate of about 83% or so (I think standard 11n FECC is ~24% overhead, where as LDPC FECC of 11ac, and optional for 11n...but not implemented AFAIK except maybe with 11ac compatible 2.4GHz client/routers, is around 15-17% overhead).
Still pretty far from those figures, but hopefully as chipsets and radios iterate, they'll manage to improve upon those figures.