We start out from stock settings and try to see how far we can push the CPU and memory systems before we encounter system instability or crashing. Once we find a benchmark stable overclock, we run a few tests and note the increase of score. For fun, we often push the clock up as far as it will go and still boot to windows. While not in any way stable, its fun to see how far you can actually push it.
As this board’s prowess is overclocking, we are going to deviate from the normal routine here a bit. After testing our usual I7-7820X CPU, we’ll also throw all the hardware we have in it and see how it fares with other CPU’s as well.
Starting out from a base 3.6Ghz, shown here with Turbo Boost 3.0 pushing us up to 4Ghz at this particular moment, and the ram set at XMP speeds of 3200Mhz @ C14 timings, we see how far we can go.
Using ASUS’s automatic system optimization built into the Dual Intelligent Processors 5 suite required a few mouse clicks, and 2 reboots.
After a reboot, the system steps through frequency multipliers and runs some small stability tests before moving on. At one point, the system stopped responding and rebooted. After it finished the last reboot after its run of about 3 minutes, we were given up to 4.50 GHz with 4 cores active, and 4.4GHz with 8 cores active. This proved stable through several benchmarks, apparently ASUS’s overclocking system knows what it’s doing.
We’ll move on and see what we can do by hand from the BIOS.
We were able to push the I7-7820X up to just shy of 5 GHz, giving us a new Cinebench score of 2098 over the stock 1747.
Next we set out to push the memory. Our 3200Mhz Corsair Dominator Kit has been able to give us around 3600Mhz before on good platforms, lets see what it can do here.
ASUS has put great effort into memory trace optimization on the APEX, which has set a few world records, and it does not disappoint here. We managed a staggering 4057MHz at stock 1.35V and slightly relaxed timings.
This speed gave us a huge increase in bandwidth, going from ~85GB/s to nearly breaking the 100GB/s mark on DRAM writes. We also saw a nice decrease in overall latency from 71.6 ns to 67.0 ns and when coupled with the huge increase in bandwidth, will result in a large performance jump in memory intensive applications.
we would normally stop here, But this board demands to be pushed, and while we don’t have the abilities currently to venture into sub-ambient extreme cooling, we can at least feed it a few more CPU’s as a snack and see how it does with them.
First up on the menu is our quad core Core i7-7740x with a stock speed of 4.3Ghz.
We managed a fully stable 5.3Ghz, and bootable at nearly 5.5Ghz.
The next CPU we have on hand is the ten core i9-7900X, with a base clock speed of 3.3Ghz.
After some tinkering and testing, we managed to push this 10 core behemoth up to 4.9Ghz! The heat output of this CPU at this speed is impressive, and thermals appear to be our limiting factor here with cores hitting the 100C mark during testing.
Despite stress testing our custom cooling loop, we still managed s large increase in Cinebench scores, going from 2177 to 2576 cb.
If there was an easy way to keep this much power running cooler under load, our non-stable overclock of 5.2Ghz would likely be stable. Can you imagine 10 cores and twenty threads at 5.2Ghz+?
Overall, every piece of hardware we threw at the Rampage VI Apex just made it laugh. Our already fairly high strung 7740X managed to get nearly a 25% increase, and despite the cable melting power draw and related heat output, the ten core i9-7900X managed nearly a 50% increase in stock speeds, while being 100% stable! Our ram kit had a similar experience, nearly doubling its previous best overclock increase to break the 4Ghz barrier, nearly 500Mhz better than out previous best.
It's pretty obvious anything we do while above ambient, even with extreme watercooling isn’t even putting a dent in this boards abilities. I do hope we have the opportunity to revisit this with more extreme cooling solutions in the future.