I'm sort of doing a very, very small 'come back' to overclocking myself (coincidentally, with a Ryzen 5 1600 as well). I haven't yet OC'ed the CPU itself (although I did just try the OC Tuner option in the BIOS, activated it, restarted and * boom *, magically the CPU sits at 3.5Ghz; but I reverted the change, I'm not gonna get +30 FPS in my games with a 300Mhz OC on a modern CPU nor would it have been the case even 10 years ago). However, I DO want to, and almost have to overclock the Memory, due to how Ryzen (more than) loves high Memory frequency and how apparently one 'gimps' their system hard time if they just run Ryzen with a "mere" 2133Mhz Memory frequency (and I can attest, not only from my own results so far but from tests all over the web, it IS true that the overall system performance is significantly better by just going from default Memory speeds to "just" 2666 or 2933 or of course as high as 3200; it's more snappier, smoother, more responsive, just overall faster and games DO gain a pretty decent FPS boost).
With this said, what I've recently learned about Memory OC'ing with Ryzen is the following (and take this with a grain of salt, I'm no expert but I'll 'report' what I've read):
1) The absolute best memory (manufacturer) for Ryzen Memory OC'ing is apparently Samsung (the famous "B-die"). They're known to 'yield' the best results, and have the best potential out there (they're usually the ones that bring you to the Memory company's 'rated' timings at specific voltages out of the box). The other memory 'types' (Hynix, etc) are working, but they're the ones that you'll "fight" with most of the times and might even get 'stuck' at default speeds forever with those.
2) Apparently, there's a thing called 'Power Phase' to take in consideration on SOME motherboards (which ones exactly I don't know). When it's a problem, the known issues are: * You can't OC the Memory IF your CPU frequency runs above the "base" speed, and if it does with the default 'boost' frequency system then the Memory OC will 'fail' and it causes issues such as freezes, crashes, etc. * In those cases when the Power Phase issue pops-up, the only way to OC the Memory is to somehow manually make sure that the CPU runs at a FIXED speed and never, ever moves up from there (so basically turning the auto boost system off completely; in AMD's case I believe it's called Turbo Core or Turbo Boost and it's indeed possible to turn that thing off when need arises). Now, how does one actually identify if their motherboard model is 'affected' by the Power Phase-related issue... now that is one heck of a good question. On a side note, I've read about this issue on ONE single post and I can't remember on which forum it was (I think it was on HardOCP), and heard about it on one single YouTube video (about how the guy in question in the video couldn't OC his Memory at all whenever his CPU happened to be OC'ed manually or when he let it run with that Turbo system On).
3) VDDCR SOC Voltage is sometimes * required * to be tempered with (increased) to attain stability with Memory OC'ing, even when regular RAM Voltage is increased (say, going with RAM voltage of often-rated 1.35v, but being unable to get past POST or still having instability issues; that's when SOC voltage comes into play, usually). I don't know much about SOC voltage yet (but there's a good amount of info about that one all over the place). I've literally read about SOME of that just today and yesterday (since I've finished building my system this very week). All I know is that as I type this (nice coincidence) I'm currently attempting to run my Memory at its 'rated' 3000Mhz speed. I've had minimal success so far, sometimes getting to the Desktop and staying there and doing my normal things for a while, until eventually I get instability and weird system behaviors and unresponsiveness issues.
However, my latest attempt (and I'm typing from my Desktop on that very attempt) was to leave the regular DRAM on Auto and JUST touch the SOC Voltage (always good to test things separately from each other, that way I know which one fails; if I had increased both and it failed I wouldn't have known which one would be the culprit). However, the SOC Voltage is a VERY risky territory, it's extremely sensitive from what I've read. Basically, do NOT, under any circumstances set that thing higher than 1.150v for safe 24/7 system use (and I've read that going below that mark is still recommended, to try and stay stable below 1.1xx is more than recommended, say something like 1.080, 1.020 instead, even below that if you're lucky enough). The gist is, from whatever default value your SOC is set to (usually around 0.9xxx volt), your increments should be minimal and done slowly, step by step with multiple reboots to test if necessary (suggested increments in my own ASUS BIOS is no more than 0.00625v, that's right). My own current voltage on SOC is set to 1.037 and so far it seems stable (or, stable enough to be in the Desktop and type this while listening to music and doing a Windows Defender system scan... only doing small things first, then doing with more potent tests like Prime95 and MemTest).
4) With all this said, there IS one thing that stuck with me from all the way back when I used to actually overclock (years ago, back on my Intel Duo E6600). When you have a system crash, a game crash (or just screen freezes), or when you get a BSOD or "sudden" system unresponsiveness or screen flashes (or Monitor going dark with the GPU seemingly stopping out of nowhere even without any obvious overheating issues), etc. When that kinda stuff happens and you just KNOW it shouldn't have happened, AND, if you know your system IS actually overclocked in ANY way (CPU, Memory, both or separately), then the GOLDEN RULE that I forced on to myself at the time (and always payed back eventually) is to do a FULL CMOS clear of your BIOS, first. That's the absolute first step. You turn the system off completely, you physically unplug the power cord from your Power Supply, you remove the motherboard's Battery and you locate the two pins used for a CMOS reset and short them with something like the tip of a screwdriver (those two pins are usually shown on your motherboard's manual with actual step by step documentation on how to reset the CMOS). You let the mobo without its battery that way for something like a minute or two (the pins shorting part takes like 5 seconds).
When that part is done, you place the battery back in, restart your system and access the BIOS immediately (don't just go straight to the Desktop). From the now-reset BIOS, immediately go to your "Optimized Default" settings wherever they happen to be, and Load those Defaults settings (do NOT make any changes to your BIOS right after a CMOS clear, that's what a lot of people do right away but they always forget to load the Optimized Defaults first). THEN, after you do that and reset the system, you go back in your BIOS a second time and THEN from that point you can adjust things.
HOWEVER, in your specific case, you NEED to do that good ol' elimination process to find out what caused your crashing / freezing issues. You KNOW that your CPU isn't manually OC'ed, so leave that portion alone. But you KNOW that you did OC your Memory. So, with optimized defaults loaded, I'd highly recommended to LEAVE everything, and I do mean absolutely * everything * untouched, at default and stock values and go to the Desktop and do your normal daily use of your system first, for maybe 1 or 2 days. It's not a race, it's not a rush, there's no point to not want to do that; believe me, been there done that in the past, patience IS a key. Now, if NORMAL (non-stressed) use of your system seems fine (playing games, streaming maybe if you do, watching movies, listening to music, doing some multitasking, etc... normal stuff) and you experience no crash immediately, THEN start doing those stress tests, Prime95, etc. Do them overnight for a start (that's usually the amount of time recommended for a 'bare minimum', something like 8 to 12 hours). And then, if those tests ARE stable, at THAT point you should know that your Memory overclock indeed most likely WAS the culprit. But you do have to do the whole elminiation process by starting from square one (default, stock values on everything). Otherwise you'll perpetually stay in a state of not knowing which exact component(s) or which exact timing value(s) or which one of the voltage values would be responsible, and if you changed a grand total of 12+ things you'll NEVER, ever get out of that loop.
So yeah, step by step, isolating and small increments has always been keys in successful overclock AND in successful investigations in finding causes to my OC issues whenever they showed up. I do happen to be doing all of that again after all those years, albeit only for the Memory, but still... I'm in that boat again now. Hope any of this helps a bit.
Facebook
Twitter