My 1700x is surprisingly terrible at overclocking, I can barely get it to be stable at 3.8ghz with 1.35vcore. If I put it on auto it goes up to 1.370 just to have 3.6ghz (not sure if judging by auto voltage means anything, but I’ve seen people mention deciding their overclock according to what auto voltage takes them during stress tests).
However, my ram is CMD16GX4M2B3200C16 which is not the optimal b-die samsung, in fact I have no idea what die it is because thaiphoon burner shows it as “? die?”. (https://imgur.com/a/aEDkC)
I have never been able to boot at the rated 3200mhz, tried every single DOCP profile as well as disabling GDM, enabling SOC LLC (not recommended), 1.2 SOC voltage, etc. It seems to be stable at 2900mhz right now, but who knows.
My question is, could suboptimal RAM for Ryzen require the CPU itself to have much higher voltage than it would otherwise need to maintain the same overclock?
I’ve seen countless posts with people running 3.7mhz at 1.25v and whatnot, or reaching almost 4ghz with 1.36. And some of them are 1700s and not x’s. I can’t even boot at those voltages and frequencies, not to mention be stable.
Run at stock with 3000 gskills V at 1.35v. Less than optimal but works fine.
Overclocking the cpu to only get up to 4.0 or 4.1 seems a waste with a turbo 3.8.
Maybe consider purchasing some ram that is known to clock higher than 2900 and focus on ram overclocking.
Completely unimpressed with Ryzen over its cpu overclocking for 1700x. Ram overclocking on the other hand may prove to be more useful for this plat. The 1700 is a bit of a different beast when comparing it to the 1700x.
Plenty of overclocking guides in the google verse. Start with your motherboard. You cannot expect best results for mish mashed parts. Most of the best overclockers results are a product of very careful choices in the parts they use.
The things is I see countless people showing results like 3.7 at 1.25, or even 1.15, and whatnot. This is a really nice undervolt I would like to have, but haven’t been able to. But even with people reaching 3.9 and 4.0 I see many results with much lower voltage than I would need for that. I was wondering how unstable my ram is making this whole thing.
I just managed to boot at 3200mhz for the first time (at least 100 previous attempts) after putting ProcODT on 68.6 ohms, which I never tried before. The CPU is on total default values though, voltage and multiplier. I will check stability.
Next check what your actual voltage is under load. I need to set my 1700X to 1.375v and Load Line Calibration at level 4 (Asus Prime X370 Pro board) and still get only 1.362v when all eight cores are under load running at 3.95GHz. Vdroop is real and needs to be checked and accounted for. I refuse to go over 1.4v and really 1.375v is my personal limit for everyday use.
I have not been able to push it to 4GHz, but I don’t really care about that last 50MHz push.
That said, some folks get real lucky, and that’s fine. Remember that the lucky ones are the ones who are going to be really proud of their luck and are going to be the loudest about it.
I have HWiNFO64, it looks like my voltage during load is 1.26 right now after putting it on auto and cpu multiplier on default, with ram at 3200mhz ( I managed to boot at this for the first time after putting ProcODT on 68.6 as I said).
However I get random spikes to 1.4 and 1.5 even while idling, the temp doesn’t rise but this is what the program reports. So does cpu-z and hwmonitor. Are these actual voltage spikes or misreporting? I’ll probably have to change voltage from auto though.
Going to keep checking for stability at 3200mhz ram + default cpu, and then slowly raise the cpu multiplier.
My main question is still unanswered though. Does sub-optimal ram make the cpu require more voltage even on lower clocks?
I don’t have a real answer, but theoretically the only voltage that should be relevant is the memory controller which I believe is part of the SOC.
FWIW I’m running 32GB (8x4) of Corsair CMK16GX4M2B3200C16 Hynix die (so I suppose it would be called suboptimal) at 2933MHz but I have done nothing beyond the standard XMP profiles. I do not believe it has affected my CPU overclock.
I mentioned this in my first post in this thread: 3.95GHz at 1.375v with LLC set to 4 (of 5, 5 being the highest). Vdroop drops the voltage under load to 1.362v as reported by HWiNFO.
Have you tried overclocking the CPU and leaving RAM at 2133/2400/2667MHz?
I tried just to see if it lets me put voltage at lower settings, I assume that I tried something like 1.3 and 3.7ghz, it failed. But then I realized I really don’t want to run a cpu OC without having good frequency on the ram because of how important ram speed is nowadays.
I actually got close cinebench results with 3.4ghz and 3200mhz ram to 3.7ghz and lower ram. Not sure if this is for other reasons though.
It has more to do with eliminating variables than long term usage. If you think RAM is the culprit, then run it at stock speeds and see if the CPU responds. You might just have to push a little more voltage, and AMD has stated that up to 1.45v is safe (even though I’m not comfortable at that voltage personally).