You can't. Using it won't allow the EK block to touch the GPU die. The stock AIO has a slight circular protrusion on the cold plate that allow it to work; the EK block does not. 
 
However, the temps on the memory are comparable to the temps I saw when using the AIO. I may add heatsinks, but I doubt it would be worth the hassle. 

bp7178
Lapped the EK block. Did another run with Time Spy Extreme Stress Test using the Thermalright TFX paste. Max temp observed was 35c on the GPU, 69c on MEM1. As you can see in the chart, 33c was MUCH more typical. The GPU-Z log only shows it hitting 35c very briefly. 
 
This is very big improvement from my first initial run using the TSEST when using the HC block. 
 
The temps I see now are much more consistent with every watercooled GPU I've owned. mid-30c while gaming, maybe touching 40c during benchmarks. Considering this is a $60 block as opposed to $300 for the KPE Hydro Copper, I can't see recommending the HC. 
 

 

What's your room temps? Mine gets to around 32-34c at room temps around 66-68f.

@ AHowes, I keep the thermostat set at 70f, so about 21c. When I move everything into a case, I'm going to set up thermal sensors on a Commander Pro so I can get radiator intake and exhaust temps in addition to water temps. This is the one spot I think the Z390 Dark is lacking, availability of headers for thermal sensors. 
 
@bcavnaugh, TFX does seem to work well, its certainly has a high thermal transfer rating. But, I found it to be thick and hard to spread, it isn't a "workable" as Kryonaut in this regard. I tried spreading the TFX at first, but ended up just going with the star pattern as in EK's instructions. The only reason I'd pick Krynoaut over TFX is that it spread easier and comes with that flat nozzle which makes applying it super easy. TFX does seem to benefit from a few thermal cycles to let it break-in a bit. 

I ran TSEST again using a Corsair ML120 fan (at 100%) directed over the GPU block to add airflow for the memory. Granted, there's no heatsinks on the memory, but the temp for MEM1 still hit 60-61c at its max. This is still over 30c under the max temp not not really much of a difference as compared to no fan. 
 
I ran TSEST again with +1200 on the memory (8200Mhz), no change in temps and finished at a 99.1% stability. 
 
https://www.3dmark.com/tsst/570300

Yeah your temps are off, this is with using distilled water, a very very low reservoir and Kryonaut.

https://i.imgur.com/g403KuS.png

https://i.imgur.com/inEPOAw.png

https://i.imgur.com/bgp5IJB.jpg

https://www.3dmark.com/tsst/577029

https://www.3dmark.com/pr/121371

bp7178
Your ambient temp is much lower if you're at 21c at idle. 
 
Even though my ambient temp is 21c (~70F) the card would be a few degrees higher than this. With the HC block, I was over 40 C and with the EK block I was in the mid 30s with the same ambient temp...well, within my level of control anway. 
 
That +130 offset is nuts though. I can't get above +80 (2115-2130Mhz) in a gaming load without random crashes unless I start applying more voltage. 

I can do a constant 2190MHz in most games without touching 36c, Time Spy is abit funny with frequencies. I usually run +135/+1375 in-game, I can do +1650 memory in some benchmarks however.

That time spy was with +130 at 2175MHz without a single dip.