Gigabyte Z690 Aorus Pro (rev. 1.x)

[elita]

Yes, it says Z690, not Z790.

According to Gigabyte, this board "only" supports memory up to a maximum of 6200MT, highlighted here: https://www.gigabyte.com/Motherboard/Z690-AORUS-PRO-rev-1x/sp#sp

However, since I was always very curious by nature, I had my doubts, so I thought, since so many "experts" are not able to differ between a system state called "idle" and the "whatever M$ does with your computer, at a time when you were made to believe, that your computer is supposed to be "idle", but isn't."

If they are wrong about the real idle power consumption of a computer, that has an Intel Core i7-13700K aka Raptor Lake CPU, which is around 26.1 Watts, not 5 billion, they may have undersold the capabilities.

And, I was right, again. It turns out, Gigabyte left out at least 1000MT to spare as performance increase, making the stated "6200MT" to a "7200MT" or even higher. Sadly, I cannot say with 100% certainty, that 7200MT/s is the maximum clock for that board, or my tests are once more limited by the memory I got for trying to find that out.

With the latest BIOS from Gigabyte, from Nov. 2023, F28b, my upper limit was a memory overclock to 7200MT/s, but I was able to boot with 8000MT/s. That however wasn't stable. The most recent memory kit I got, is specified to 7200MT/s, but all the Kingston RAM will allow at least some overclocking.

My first memory kit for that motherboard was an XMP 4800MT/s kit with a ludacris price tag. A year later and having had the time to read as much as I could about the new hybrid many-core platforms, I was lucky to get my hands on 64GB 6000MT/s memory. Sadly the Kingston Fury Renegade RGB didn't like to be overclocked as much, but I could reach the "border, set by Gigabyte" of 6200MT/s with the same timing as advertised, CL32-38-38-80.

Not being able to overclock to much does not mean, the memory is bad. Those 6000MT/s sticks did allow for some really steep timings, like CL28-33-33-56 @1.1V with the JEDEC clock of 4800MT/s. Just for your information, I got my Raptor Lake CPU and other components in early September 2023.

Not even two weeks ago, I wanted to find out, if Gigabyte was keeping something from us, and I got a pricey 6400MT/s memory kit, which booted up, without even having to train the memory. It felt like the motherboard did wait for this memory kit to come along and finally allow it, to perform properly.

The same memory kit, same timing and voltage, I only set the memory clock from 6400 to 6800 and without much delay, it did boot up at 6800MT/s and the timings of the 6400MT/s. Leaving most settings on "auto", memtest said no to trying to bring the memory up to 7000 or 7200.

After three days of testing, I sent the memory back and ordered the 7200MT/s memory. I am still debating, if I send those back as well and order the super expensive 8000 or 7600MT. Unfortunately for me, those memory kits are mega-parsecs above my price range and if they go below €200 for the 32GB kit, I might try those.

I am no engineer and in this case, I really cannot say, if the memory or the motherboard, or the memory controller is limiting clock speeds above 7200MT/s. Sometime, the board will allow 7200MT/s as the highest stable memory clock so far.

Yes, 'sometimes'. Having a mind of its own, it appears, as if the board has mood-swings and some of the time, I cannot set anything above 7000MT/s. Then I have to go down to 6800MT/s. And it may be as close as 10 minutes later and suddenly every memtest passes with flying colors at 7200MT/s.

In the few nights, I was able to dive a little deeper into the professional world of "you did not overclock anything, as long as cpu-z wasn't there and take some DNA samples, some fiber, some dirt particles and fingerprints for the forensic lab" - err, I meant overclockers.

At least, Gigabyte didn't over-sell the PCI-e bandwidth performance that would be possible if you have the right gear, of course. You can decide for yourself, if cpu-z has the final word or not and check this out:

https://valid.x86.fr/vrazvc or https://valid.x86.fr/kxf1kz. In my list, there are also tests and validations with my main pc, which has the previous Gen 12th Intel Core i7-12700 Alder Lake CPU on another Z690 Aorus Pro, also BIOS F28b. Though both board "should" be the same, but when I got my Raptor Lake parts, I noticed that both board were not equal, and I have revision 1.1 and 1.2 of the Z690 Aorus Pro.

Since the Alder Lake CPU wasn't meant to overclock, all I do there from time to time would be checking, what kind of gain one would get from higher clocking memory.

 

Speaking of wild timings: https://valid.x86.fr/wz2ple

Next time, I have the time to bench with 7000 or 7200MT/s memory clocks, I will validate those by cpu-z and then start benching. So next time, you prepare to hang me from the hanging tree for leaving of cpu-z, maybe check first.

[elita]

Okay, that took a little longer than I thought, but at least I got this to work on my 6-layer 4-dimm motherboard from 2021:

Turns out, the Aorus Pro can do the same things the Z690 Master can do. And the latest bios is from Early this January. (The beta-bios before that wasn't allowing that same memory kit to go anywhere beyond 7200MT stable. It is still the same Kingston Fury Renegade 7200MT/s kit I am using for this.)

I also had to make a slight modification to cool the memory, to allow for stable temperatures in a closed case and to not go hotter than 40° C.

The HWInfo screens and other benchmarks will follow. Stay tuned for finding out the maximum possible memory clock that isn't possible on a 6-layer, 4-dimm motherboard and especially not with (only) a Z690 chipset.

Or is it?