AX370-Gaming 5 General Instability and Virtual Capability Lacking

[totally_random]

Have been putting off raising issues to allow the environment around the Ryzen setups to settle down, but I can't seem to make any real progress after 3 months now.

The main issues I'm currently hitting:

• RAM is not stable above the default 2133MHz. I recently managed to get 2666MHz manually configured and somewhat stable, but still had 2 lockups and a Kernel Mode Exception Trap (KMET) BSOD over the course of 3-4 weeks while on Windows 7.
• VMware Workstation 12.5.6 still reports a CPUID bug trying to power on any VM, despite me disabling SMT and manually tweaking things like IOMMU. I was able to boot into the ESXi 6.5 installer with no issues; I would assume this would be an application issue, though googling reveals people running on Ryzen setups without problems. Trying VirtualBox briefly, it would only let me use 1 CPU, similar to how machines that don't support VT-d appear - and the VM wouldn't boot-to-install anyway, winload.exe saying the CPU is not compatible with 64-bit mode.

Alas I bought this setup to play games, perform development, and to run a beefy Linux VM so I could decommission my KVM and second desktop, none of which I can perform optimally at present

CPU
    Ryzen 7 1800X (Stock) with Noctua NH U12S SE-AM4
RAM
    2x 16GB Corsair Vengeance LPX 3200MHz - CMK32GX4M2B3200C16
Motherboard
    Gigabyte AX370-Gaming 5 with BIOS F4, F5, F5d, currently F6d
Video Card
    AMD Radeon R9 290 4GB
PSU
    be quiet! Dark Power Pro 750W
Disks
    Samsung 256GB M.2 NVMe (boot)
    Samsung 1TB 840 EVO
Operating System
    Windows 7 Ultimate SP1 x64 (currently Windows 10 Pro x64)

Today, I blew away the Windows 7 install and put Windows 10 Creators Update on; within the space of 6 hours I got another KMET BSOD. I then disabled the XMP Profile, putting the memory back at 2133MHz and full defaults - and I got another total lockup after no more than 15 minutes (screen goes back, fans spin higher, zero response from anything until powered off). This did oddly coincide with opening Firefox 10-20 seconds earlier, which was open when the other BSODs occurred too.

HWMonitor confirms all temperatures are rock solid, and the PSU has been used with an Intel Core i5-2500K and AMD Phenom II X6 1090T using the same graphics card, and I never had any issues over years.

I suspect all the problems could just be down to the RAM (all info I've researched points these to being Hynix), but I'm not willing to spend another £200+ on something which may make zero difference; likewise for getting another motherboard and comparing with that.

Any suggestions?

[BoNfiRe]

What voltage on the CPU and RAM? I'm running Corsair RAM too, and with Hynix memory chips. Mine is only 3000Mhz. I can run it at 3200Mhz with loosened timings. So I'm sure it's not the RAM, you could try memtest?

[totally_random]

What voltage on the CPU and RAM? I'm running Corsair RAM too, and with Hynix memory chips. Mine is only 3000Mhz. I can run it at 3200Mhz with loosened timings. So I'm sure it's not the RAM, you could try memtest?

At stock, the RAM is running with 1.2V - I've had to raise it to 1.35V to reach 2666MHz, which is actually the voltage stated on the DIMMs - but it appeared more stable at ~1.375V.

CPU has been running at stock with the sole exception of the SoC, which was bumped to 1.35V as well. Wouldn't post with RAM at anything other than defaults unless this was done.

I have done a memtest, both within Windows and the classic bootable iso, with nothing flagged.

Reading some of the other similar issues (bugcheck is a double fault), it would appear that even more voltage is recommended. Not particularly fussed with overclocking, but would like to get the RAM somewhere close to its stated speeds 

[absic]

Which BIOS version are you running?

[BoNfiRe]

Which BIOS version are you running?

The poster did say that they tried them all.

If memtest isn't failing, I can't see it being the RAM. Your system is similar to mine apart from RAM speed, but not a lot. I'm using the 1600X and bar the 2 cores, they're exactly the same, and my system is rock solid. All I can think of is a software issue, rather than hardware and I'm thinking it could be VMWare... possibly.

EDIT: Have you tried reflashing the BIOS? Also, do it in the BIOS and not through Windows.

[totally_random]

Am happy to report that I've found the issue in relation to the virtualization; SVM Mode was set to Auto, but I had to manually set it to Enabled.

Both VMware Workstation and VirtualBox both work without any issues now, all features available, including with SMT re-enabled - wouldn't take affect after the first save+exit, but the second time round, it was enabled. Seem to recall this being a common thing with people attempting overclocks.

Just going to see if I can keep the system stable for now. I've re-enabled 2666MHz @ 16-18-18-18-36 with tRC 64, but bumped the voltage to 1.38V, and increased the VCORE SoC to 1.2V. Fingers crossed 

[BoNfiRe]

Glad it's working for you now, so it was software, rather than hardware, but auto is a fickled SOB. As for the RAM timings and voltage... Mine is at 15-17-17-17-36 @ 1.35 and I can get it to 3200 just fine. I wouldn't want to raise the voltage that much as I would imagine you should get it to run @ 16-16-16-16-36 on 1.35v and 3200Mhz (which is stock settings) with no issues at all.

[totally_random]

Glad it's working for you now, so it was software, rather than hardware, but auto is a fickled SOB. As for the RAM timings and voltage... Mine is at 15-17-17-17-36 @ 1.35 and I can get it to 3200 just fine. I wouldn't want to raise the voltage that much as I would imagine you should get it to run @ 16-16-16-16-36 on 1.35v and 3200Mhz (which is stock settings) with no issues at all.

Yeah, is a strange one - looking at the BIOS now, the only options for it are Enabled and Disabled; assume it's tied into other settings possibly..

Just tried again to get 3200 or higher than 2666 to no avail; never POSTs until it resets the BIOS. Have tried using the timings as presented in system information viewer, attached - timings are maybe not as 'standard' as possible. Will give all 16's a try, can't hurt.

Can live with 2666 really if it was truly stable; just had another lockup while trying to install a VM, and had just come into this browser and started typing. Not a BSOD, so no information about it sadly - just screens suddenly going blank and nothing responding. I'm a heavy multi-tasker, and never knowing when the system is going to fail is stopping me from normal workflow 

EDIT: Just had another BSOD, page fault in non-paged area. Windows 10 does seem to be more susceptible than Windows 7 was, amusingly. I've reverted back to all-defaults, 1.2V RAM @ 2133MHz (which is 15-15-15-15-36), no SoC or any other modifications - is the only thing I've had zero issues with for more than a month so far