Originally Posted by Falkentyne
I mean you need a MUCH higher vcore to pass LinX 0.9.6 35000 sample size with all matching residuals than you do to pass anything you would normally use in real world.
You need an even higher vcore than what FMA3 small FFT prime95 requires.
LinX isn't just about RAM stability.
Originally Posted by reachthesky
That's the one issue with the idea of investing into a 4 dimm board with the idea of using 2 slots for an unknown unspecific amount of time until the need for more than 16gb kicks in. It's only good to do that if its a daisy chain board with 4 dimm slots because if you start out with 2 dimms on a t-topology, you are voluntarily gimping your performance for an unspecific amount of time(how ever long that is until the person needed to add sticks, we don't know). That sort of thing doesn't really flow with the purpose of enthusiast platforms right or geared towards the target market right? See what I mean? Better off selling the master and getting aorus pro, then the user will have an extra 100ish dollars to put into a good ram kit. Then if the user needs to get the full 32gb they can buy the dual ranked kit and sell the old kit. my 2 cents
. I guess I just feel like it is more cost efficient and more performance effective to either start with a daisy chain 4 dimm board or start with a 2 dimm board or get a 4 dimm t-topology board if you are going to populate the slots all at once from the start.
As per request: (readable scores
) I'm sure the dark will be able to blow these numbers away at the same cas/frequency, if you do end up doing some benching, would you mind trying out some cas 15 frequencies if you've got the time?
latency: 36.0 (can only get this lower if I clock to about 5.4 cores 5.1 cache but volts are silly).
I'm actually going to just run it daily no ****s given. I want this performance daily. I've heard many people running b-die between 1.5v-1.6v daily without degradation yet(or so others claim on reddit LOL). I don't mind joining them, the performance is SO good in games. They really ought to just make a ram stick that performs this well without having to tweak it but i imagine it would be hard to ensure compatibility with all processors. DDR5 will be exciting.
I didn't know it was ok or not frowned upon to run knowingly non-memtest stable ram configurations, but If you are going to lead by example, I will follow!
Can the z390 dark get 4800mhz on ram like the z390 apex or is it a crapshoot/luck? Been wanting an apex ever since those apex users have been pumping out stable configs between 4500-4800 in the ddr4 thread, its nuts! 35ns latency with like 70k gb/s or more performance lol. Just nuts. I gotta get these megahertz rofl.
Fair point. I wonder though if one is on a budget to begin with if the cost savings of going with a cheaper board would justify the difference in RAM cost? I guess it's just all relative to the usage case, how much RAM they really need and what type of RAM they opt for. 4GB single sticks are cheap, but although 16GB total is still good enough today with games etc... for some users 16GB is quickly becoming the new 8GB rock bottom standard.
Regarding the Dark boards, the Z390 Dark is rated at 4600+ and my 4600 kits in there posted effortlessly like it was nothing. My X299 Dark posts 4600 like it's a feather as well ... It's next level amazing how good these boards are. I'm debating on even just putting in the Z390 Dark into my build, which I was actually going to do in the first place, but since I had the Z390 Master (and because it's such a solid board) I just kept it as my primary gaming board. Honestly the Z390 Dark is way over kill for what I need in the gaming rig and the Dark fits the bill better being on my test bench.
I'm now looking forward to testing RAM on the Dark boards for bench sake and also for realistic-real-world sake as to see what it can do if I were to run my Dark as my daily. About those voltages above 1.5v daily. You know... I honestly think they'll be okay and I base this off of the fact that I've ran RAM sticks and beat them to the ground over many bench sessions, let alone the data that can easily be seen from all the other fellow overclockers/benchers out there that put their RAM through the ringer. It's not to say to run RAM at 1.5v - 1.6v for the heck of it, but like my view towards CPU's I think they are pretty resilient, because they are. The Samsung B-Die's are rated and tested at the 1.5v mark which is why I just choose to keep that mark as my max. Just personal preference and it's not because I worry that it'll fry my RAM if I run it any higher (voltages) ...
If your after MHz, give the Dark board a serious consideration. The Z390 only has two mem slots, but dang does it scale well. Extremely strong. You won't be disappointed. I'll just say one thing though, be aware though that once you experience the awesomeness of that board, everything else will seem like weak sauce lol. It'll be really hard to justify using anything else even above its price range. As for me I'm completely content with the GB Z390 Master at the moment as my gaming rig, because it's the best in class for the price range. I got this thing on sale for ~200 so I'm not complaining one bit.
All fun stuff... really enjoy this stuff a lot.
For giggles here's a 50x Cinebench R20 run on the Z390 Master that topped its class...pretty impressive. Pic attached.
I'd be curious to see what you can top out at with CB R20...would you mind running some benches at 50x to compare? I think your RAM might actually have better chips than mine... should be able to score well.
Also attached are the top recorded scores of other 9900K/KF CBR20's at 50x .... as you can see this Z390 Master / regular gaming desktop topped them all... most of the "others" are on test benches lol...