I just want to post some of my "unscientific" findings with Ryzen.
I have C6H, Prime X370-Pro, and an Asrock B350M Pro4 boards.
I have FlareX 3200MT 2x8GB, and TridentZ RGB 3200MT 2x16GB RAM kits. Both are Samsung B-Die.
I have had Ryzen 1200, 1300X, 1500X, 1600, 1600X, 1700, 1700X, and two 1800X CPUs come through my possession.
All boards have been able to run 3200MT with at least the XMP timings since AGESA 188.8.131.52 on each CPU tested.
My testing was unscientific, and my results are not documented. Only a few of these chips were in my possession as my own, and some of them were only available to me for a few days max. Thus my conclusions are my own and should be taken as just opinion.
Using either the C6H or Prime X370-Pro and any one of the CPUs I have NOT been able to achieve a stable 4GHz overclock. By stable I mean IBT/LinX for several hours at least and 4K HEVC encode workload surpassing 5+ hours of full CPU utilization. The MAXIMUM stable clock I was able to achieve and complete stability testing with any of these CPUs or boards was 3975MHz while keeping under 1.5v as measured by HWInfo64 SVI2 TFN sensor on CPU.
My conclusion is that AMD is already pushing this current generation of it's CPU architecture and IMC design to it's limit. The Ryzen 2 (Zen+) will probably bring some more maturity to both the CPU architecture and IMC, but I don't think we will see any higher clocks out of this generation through motherboard or BIOS updates. Memory compatibility sure, but raw CPU clocks no. Additionally, the Zen architecture is shared pretty much across AMD's entire line-up right now, so I assume the chips are heavily binned with the best wafers going toward Threadripper or EPYC. I have personally seen 4050MHz tested stable on a 1950X, which is an incredible achievement compared to our consumer level chips because of having double the odds of a weaker core as compared to an 1800X.
I know there are golden chips out there and I'm not saying anyone who is claiming 4000-4100mhz isn't telling the truth, but I simply do not trust any OC.net users/YouTubers/Reviewers who claim to have acheived a 100% stable 4GHz or higher overclock unless they post their evidence. Especially the reviewers and YouTubers, because they all claim to have done so yet I have tested 9 different chips and haven't seen one that can do it.
I simply do not believe the vast majority of Ryzen chips can sustain 4GHz on all cores under safe voltage for more than a short period of time. I won't call anyone out for it, but I also won't believe anyone's claims unless they post valid stability tests.
That being said, I am so happy that AMD is back at the table and I am eagerly looking forward to their next revision!
Workstation: Intel 8700k (Conductonaut Delid) @ 5.0GHz w/ Kraken x62 (Kryonaut), ASUS Maximus X Code, 32GB G.Skill RGB DDR4 3200MHz RAM @ 3200MHz (14-14-14-14-28-2T), 2x500GB Samsung 950 EVO RAID-L0 Boot/Main Drive, 2x2TB Hitachi Ultrastar RAID-L0 Storage Drive, 512GB Samsung 950 Pro M.2 Cache Drive, EVGA Geforce 1080 Ti FTW3 Hybrid (Shunt Mod), NZXT H700i Case.
Home Server 1: 2x Xeon E5 2697 v2, ASRock EP2C602-4L/D16 Mobo, 128GB DDR3 1866MHz ECC RAM, 2x256GB Samsung 850 PRO RAID-L1 on Intel C602 Boot/Main Drive, 15x3TB Hitachi Ultrastar RAID-L6 on LSI MegaRAID 9271-8i w/ BBU Storage Drive, 480GB Apacer Z280 M.2 Cache Drive, 2x1000M LAN, 1x40000M Infiniband.
Home Server 2: Dell PowerEdge C2100 FS12-TY 2U Rack Server, 2x Xeon X5650 @ 2.6GHz, 48GB DDR3 1600MHz ECC RAM, 2x500GB SSD RAID-L0 Intel RAID, 12x3TB Hitachi UltraStar RAID-L6 on Dell H700 w/ BBU, 2x1000M LAN, 1x40000 Infiniband.