1. AIDA 64 will present a load profile that your PC will never see again, it's value for stability testing is therefore questionable at best. It's kinda like testing a Florida soccer mom's van by testing it by testing off road in snow, mud or hill climbing. If you choose to limit ya OC because of a temperature ceiling of say 75C under AIDA, I'd guess that it will never see 65C doing anything you actually use ya PC for. So limiting your OC to any specific multiplier, unnecessary curtails your OC because it will never see that load / temps again. While still well more than ya need, RoG Real Bench I find to be a more useful stress test as a) it presents a less intensive load than the synthetic programs, b) it's still way more load than your PC will ever say and c) it's multitasking nature exposes stability issues that synthetics do not uncover. I have has 24 hour stable P95 OCs fail in 45 minutes under RB.
2. A decent air cooler is expected to do as well as a 2 x 120mm water cooler.... any CLC type cooler can not compete with the better air coolers at comparable noise levels. And by better, I'm not just talking about the $90 offerings from Noctua and Cryroig. The $45 Scythe Fuma and $37 Scythe Mugen Max will equal or outperform both the Noctua and Cryorig offerings.
3. At 1.168 volts, you shouldn't be in need of much cooling. Here's the Kaby Lake results from overclock.net's KL Overclocking thread:
Average OC 5.03 / Average Vcore 1.36
Median OC 5.00/ Median Vcore 1.36
So what you have done basically is test a 500 pound hoist and a 750 pound hoist by lifting 375 pounds ... you will need to apply a bigger load to separate the two.
4. You should do various tests with the pump at various fixed speeds. You will find that you will see significant increase up to a point and then increased pump speed / coolant flow has less and less effect. Set up ya pump speed curve accordingly.
5. We use 1200-1250 rpm fans and I find they are inaudible at speeds up to 850 rpm. So I aim for enough radiator that I will never hear them under normal usage. Usually this means that I'm setting the fan curve to 850 rpm @ 75C with a steep curve up to 100% at 80C.
6. With regard to your plans for including the GFX card in the loop, if you consider selling the Strix, I'd suggest two option if available in your neck of the woods.s:
a) Asus Poseidon ... OK it's not as good as a real full cover water block, but it's nice compromise allowing you run on either air or water cooling.
b) MSI Seahawk EK X gives you a full cover water block (covering GPU and VRAM / memory) and quality AIB GFX card and a quality EK water block for far less than you could buy them individually. Ofcourse pricing in your area may be vastly different
https://www.newegg.com/Product/Product.aspx?Item=N82E16814127952
7. We use a spreadsheet tool based upon radiator performance data performed by MartinsLiquidLab . Using fan speeds and tested radiator performance you can reasonably calculate coolant temps. It's been accurate to within 5% so far. It is based upon a target "Delta T" (between ambient air and coolant temp) of 10C. so if air is 25C, you can expect coolant temps of 35C. I would not expect you to maintain that delta T. But doing a rough calculation
OC CPU @ 1.35 volts ~ 130 watts
OC'd MSI 1080 ~ 255 watts
Pump ~ 5 watts
Total = 390 watts
You then need to estimate the amount of heat that is handled by the radiators, versus the amount that is radiated inside the case by rad shrouds component surfaces, tubing, etc This works out to be about 60% on SLI systems where the large surface area of the water blocks handling a significant portion of the total load (with two GFX cards, they produce 80% of the load ... 65% on single systems). So 75% is a more appropriate number for single card systems
75% of 390 ~ 293 watts
The EK XT360 line isn't what you are uysing but it's the closest we have in the test results and it has the following characteristics:
1000 rpm = 145 watts
1250 rpm = 178 watts
1400 rpm = 197 watts
1800 rpm = 246 watts
2200 rpm = 291 watts
So I would expect that your "Delta T" at 1800 rpm would be very close to 10C at 2200 rpm. However, that's a design goal not a strict mandate. At 1250 rpm, you are looking at about 16.3 delta T and that's a lot better than what you'd see with a CLC and aluminum radiator at 2700 rpm.