Back in 2012 I picked up an HDLink for half-price at a consignment sale. 3 years later, I finally got around to trying it out.
I was able to confirm everything Steve said.
I ran SMPTE bars out of Resolve (w/ a Decklink), to a Tektronix scope. Then I did the same through the HDLink, and overlaid the results:
You basically don't see anything special, because the signal was identical.
Then I created a LUT. First I installed it in Resolve and measured. Then ran it through pipeline #2 (LUT in HDLink), and as you can see from the overlay, it's a mess. The results are decidedly different.
Here's a PSD file with the 4 frame grabs:http://drewlahat.com/misc/Color/HDLink_Tests.psd
Now in my case, the LUT is for calibrating a consumer client display, which was at >dE11 beforehand. So it's still worth using vs. not using anything, but at least be aware of what you're doing. I also measured the display chain with the LUTs running in Resolve vs. the HDLink, and magically, the skew that the HDLink added made it even better. Oh well.