Good deal man. I had a similar but easier to diagnose situation on Zen2 launch. It's like they felt yields were solid enough to just outright skip QC. I had a DOA 3700X right out of the box, and a 3600 that would only ever see half the ram, one from Amazon and one from Micro Center. Getting them replaced was no biggie, and the replacements have been flawless other than unrelated pains like Mobo/Agesa growing pains. Because of how rare bad CPUs have been for so long, I actually swapped out motherboard first after trying known good PSU and Ram I had on hand since everything was new for the builds. The odds usually on a defective Mobo, PSU, or Ram are astronomical compared to bad CPU.
I think it only affected the first batch to any significant degree though, it's been just going from good to awesome for Zen2 as time goes on, and I don't hesitate to recommend them for most new builds.
I have a Ryzen 3000-series R5 3600 CPU, from an "early batch", got it the night of or the night after initial release date. (What a thrill!!)
After suffering through a number of boards and AGESA update versions (settling on an Asus B450-F ROG STRIX, now listed as "Ryzen 3000 Ready" on their web site, with AGESA 1.0.0.4 patch B), I still have some instability when left at more-or-less "stock" settings. (Auto-tuned OC stuff that it does by default.) I do have 4x8GB GSkill Trident Z RGB DDR4-3600 Hynix RAM, which has always worked OOTB @ XMP settings with the 3600 CPU, in both boards I tested it with. I'm also tweaking the FCLK slightly, to 1800, to be 1:1 with DRAM clock.
That said, until I recently switched to an all-core 4.0Ghz @ 1.3685V OC on the 3600, I was getting more-or-less weekly crashes, video driver crashes, appcrashes, stuff just wasn't
completely happy with me.
I did, at one point when initially testing the CPU, with a PrimeGrid workload (ask
@DrMrLordX about it), get it up to 127C for like two hours, before I decided that maybe the thermal sensors WEREN'T borked, and I was "cooking" my CPU.
So, I may have "weathered" my CPU a bit, and thrown off the voltage/freq. Turbo curves programmed at the factory, so perhaps that's the reasons for the crashing, or perhaps, running the DRAM at 3600, while fine with the CPU, the mobo specs only list up to 3466 for DRAM OC speeds, and then there's the FCLK tweak.
So, maybe I was just running everything a little too close to the edge.
Or, maybe, my CPU being one of the "first batch", maybe the silicon process has matured slightly, and CPUs coming out now, are more stable and better mfg'ed than mine. (But I wouldn't exchange it, I manually cooked it during OC'ing, and I take full responsibility for that, if it fails, it fails, and I buy a new one.)
So far so good, though, with the manual all-core OC to 4.0Ghz. (Which is honestly, pretty mild of an OC, but the auto-turbo thingy, would get to just below 4.0Ghz for an all-core turbo clock anyways.)
So my all-core fixed OC, is in the vicinity of what it was running with the turbo clock all-core.