JustPlainGarak

I'm having the same issue as @mightyskull is, but on a Windows Server machine. I have the proper firewall rules in place (7396 for Web Client in from my subnets and 36330 in for Advanced Control), the proper subnets configured in Advanced Control, AND I can connect to Advanced Control from other machines on my LAN just fine (and yes, I have restarted the service multiple times). However, when I try to hit the Web Client from another machine, the page loads to a 401 UNAUTHORIZED page and I can see the 401 message appear in the logs on the server. The behavior is the same even when I have the Windows Firewall completely disabled. Is it possible that remote Web Client access is just busted right now?

As a person who's running a 3700X on a B350 board, no you shouldn't just ignore the QVL. You CAN, but I don't recommend it. I had a heck of a time dialing in stable settings on my FlareX kit on my AB350N Gaming Wifi and it took lots of tuning of voltages (both memory voltages and SoC voltages) and I probably would have had a better time if I had actually consulted the QVL before choosing a memory kit. After the fact, my kit is CLOSE in model number to ones on the QVL, but exact matches are going to give you better compatibility and an easier time dialing in either XMP or custom speeds and timings. Were I you (what with the benefit of hindsight), I'd look at the 2nd Gen QVL for the board, find a kit that meets your requirements for speed and timings and get that kit. I should have followed my own advice, it's what I did when I built computers for family.

Literally today I think I actually got somewhere, and it wasn't what I was expecting to find, either. While I was watching some AHOC videos on Ryzen memory I thought to myself "I wonder if my board just doesn't like the tighter timings with the default voltage." Since I have a Samsung B-die kit (and since you're running Flare X 3200 MHz I bet yours is a B-die kit as well) I figured I would try my luck at a higher frequency with looser timings and more DRAM voltage. Well, as luck would have it, it turns out that seems to be working decently. I've only been running it for a few hours, but all the random Cinebench runs I've done have completed without errors, and I've played some Destiny, SotTR, and AC:Odyssey (both locally and streaming) without any crashes yet. Yet. Now, our motherboards are pretty significantly different, but you may find some success with this as well. You'll want to determine for sure if you've got a B-die kit, but here are the settings I've got working for me at the moment. XMP Profile: Off DDR MT/s Speed: 3600 DDR4 Voltage: 1.45v SOC Voltage: 1.1v Timings set manually as provided by the Ryzen DRAM calculator version 1.6.1. Aside from the timings section in the middle and the two voltages that I tweaked on my own, the only other thing I set manually was the Command Rate at 1T. I recommend you download the DRAM calculator and Thaiphoon (this gets you info on your DIMMs, such as the die type and how many ranks, etc) and see what the application gives you. My big take away is that XMP isn't nearly as fool proof as RAM and mobo manufacturers would want us to believe, at least not on AMD's platform. No wonder Linus is always grumbling about how bad it is! This seems to be working for me, so hopefully you have some luck with this yourself! Screenshot of the timings used below:

A single stick of DDR memory, of any generation, will run at it's rated clock and transfer rate on its own. You need two sticks (or multiple pairs*) of DDR memory to run in a multi-channel configuration, which is separate and distinct from the data rate. *if you want it to work well/correctly

I was also worried about voltage and thermals on my 3700X when I was testing for temps and was even running a negative voltage offset (-0.5v IIRC) when the chip first came out (first time ever being an early adopter, bought from Newegg on launch day after watching/reading a ton of reviews). Now, however, AMD appears to have (from my testing, so this is all A Single Point of Data) tuned the boosting and voltage behavior significantly with AGESA 1.0.0.3ABB and the latest chipset drivers (7/31/19). I'm seeing better temps, more consistent boosting behavior, and even a slight bump in Cinebench R20 scores (4700-something to 4800-something). When pushing the CPU to full load (either with Cinebench or with Handbrake doing HEVC encoding), I will see peaks to 90 C but generally sit around 85 - 88 C. So, if you're worried about voltages and temps at stock with the default Precision Boost 2 (not PBO or AutoOC, which are both distinct and different from PB2), I'd recommend updating first to the latest chipset drivers from AMD and then the latest BIOS with AGESA 1.0.0.3ABB if your motherboard manufacturer has made that available. If anyone sees those temps and starts to sweat, I'm running this CPU in a Silverstone RVZ03B with a Scythe Big Shuriken 3, so these temps are PRETTY GOOD for a LP cooler in a 14 liter case. As most people have noted above, Ryzen 3000 is rated to run anywhere from 0.200 - 1.500v stock depending on load, and even with lightly threaded tasks the voltages you have reported are pretty normal. Tl;dr: This behavior is normal for Ryzen 3000, but it also has improved (for me, at least) with the latest chipset drivers and AGESA 1.0.0.3ABB.

Unfortunately, no go. Tweaking the SOC voltage, even up to 1.2, made it, if anything, more unstable at 3200 MHz. I also tried upping the memory voltage a bit too, but that also went nowhere fast. Really just seems like this board just won't go above 3000 MHz memory in Windows. I don't think it's the kit, necessarily, because memtest didn't produce any errors at all at the rated XMP speeds.

That's a solid idea, I'll give that a shot over the next few days and report back on my findings!

I've been doing a lot of searching and have not found any threads specific to this problem (either here or elsewhere on the interwebs) so I figured I'd actually come in here and post a thread. I'm running a Gigabyte GA-AB350N-Gaming WIFI board with a 16 GB kit of G.Skill Flare X 3200 MHz CL14 kit on a Ryzen 7 3700X. I upgraded from a Ryzen 7 1700X with a 16 GB G.Skill Fortis 2400 MHz CL16 kit (on this same motherboard) so I'm not really surprised I've not run into the issue I am about to describe. In the BIOS, when I set the 3200 MHz XMP profile for this memory kit, the BIOS posts and I can get into Windows just fine, but as soon as I login, instability hits and I either have tons of application crashes due to instability and then a BSOD (not the same BSOD, it changes depending on it's mood, I guess) or I just straight up get a BSOD before much can finish launching (this pretty consistently gives a MEMORY MANAGEMENT BSOD). If I leave the XMP profile on but drop the memory speed to 3000 MHz, I have no issues with stability whatsoever. Here are the troubleshooting steps I've gone through to try and get Windows stable with this memory kit: Set VRAM voltage manually to 1.35v in the BIOS (just in case the BIOS was mis-handling the XMP voltage). Set 3200 MHz speed without XMP on (the timings were horrendous, so that was pretty much a non-starter). Use the Ryzen DRAM Calculator to manually set the timings both with XMP turned on and with XMP turned off. Set the standard 3200 MHz XMP profile and ran MemTest86 to look for errors - no errors detected at the rated XMP speeds and timings. Verified that the BIOS is setting FCLK and MCLK correctly automatically to 1600 MHz when the RAM is set for XMP profile at 3200 MHz. Manually set FCLK and MCLK to 1600 MHz when the RAM is set for XMP profile at 3200 MHz. Validated that when I drop my memory speed to 3000 MHz or 2933 MHz with XMP on that the FCLK and MCLK also drop to match automatically (these are both stable speeds in Windows). At this point, I'm kind of at a loss of what I could possibly do to get this kit stable at it's rated XMP speed and timings. I haven't tried loosening to timings to something more along the lines of a CL16 rating because I'll probably get better results just running at XMP CL14 @ 3000 MHz than at 3200 MHz with looser timings. I suspect that the problem is the motherboard and the generally pretty terrible memory compatibility of this particular board. It's technically rated for 3200 MHz memory OC, but I think with the new CPU and this memory kit, it might just not be possible at this time. Just to head this off at the pass, no, the kit I am using is not on the QVL for the motherboard. The kit's model number is close to some verified models, but a LOT of the kits on the QVL for this board are 4x configurations, which says to me they didn't do a ton of actual testing on this SKU since the board only has 2 DIMM slots. I think Gigabyte was doing "close enough" testing for this board, hence my problems today with a 3700X. At this point, I'm fairly willing to chock this up to a substandard motherboard and I'll just have to budget for a B450/X470 board that's Ryzen 3000 ready to get this kit working (don't really have the $$ for X570, or I would have done that already!). See specs below: Windows 10 1903 64-bit (latest Chipset drivers from 7/31/19 are installed) Ryzen 7 3700X Gigabyte GA-AB350N-Gaming WIFI (rev.1) (Running the latest F42a BIOS, yay Destiny, finally) 16 GB 2x8 GB G.Skill Flare X 3200 MHz CL14 Kit (model #F4-3200C14D-16GFX) Gigabyte GTX 1080 GV-N1080TTOC-8GD (modified with a 3rd party cooler was the blower was terrible) Corsair SF600 600W SFX power supply Samsung 850 EVO 512 GB SSD (boot drive) Seagate 1TB HDD (2.5 in form factor, also terrible and slow) Seagate 4 TB USB 3.0 External Drive Thanks in advance!