Wooo thanks for the review @wendell. Been looking forward to this one.
Just placed an order for my 32GB RAM Kit - 2x packs of Corsair Dominator Platinum Series 16GB (2 x 8GB) DDR4 DRAM 3200MHz (PC4-25600) C16 (CMD16GX4M2B3200C16)
@comfreak Good point, curious to hear about that too.
I haven’t had much luck finding ECC RAM beyond 2400 MT/s - if you do, let me know.
Only had two ecc sticks. One bank was enabled one was not. It works fine tho in quad channel if you have 4 matching sticks.
I recall you posted a video on increasing the timings to improve the infinity fabric for slower sticks, i.e. a 2400 MT/s kit could perform similar to 3200 MT/s. Would you recommend this approach given there aren’t much options for ECC RAM past 2400MT/s?
I haven’t experimented with it much. Unless the ecc has really right timings you could probably bump from 2400 to 2666 or 2666 to 2800 but that’s about it.
If seems to have some latencies of it’s own for sure. So it kinda works fudging it a bit.
So if I understood it right, the populated bank will properly work in ECC mode, despite the log message?
The log message describes both banks? Ecc on in one bank off in the other bank is present in the log message.
So yes?
Thanks! The reason why I asked was that I thought from the message that it will only work in ECC at all, if both banks are populated.
But I wouldn’t leave the second bank empty anyway 
The ryzen memory controller situation is for sure a bit weird. each die gets its own dual channel controller. I guess. Vs “true” quad channel memory on the other platform.
It does scale really well with serverish workloads, though, vs the other approach
Video promised the IOMMU groups in text… can’t see it anywhere but also did you try with ACS patch as well or not?
You can refer my listing until @wendell updates the article
I’ve got a GTX1070 in the second PCIe slot (running in 8x); it’s just a temporary situation.
Thanks @bsodmike - does the ACS patch improve that any? I notice the USB controllers are all bundled in with other stuff and ideally I would like to pass through a separate GPU and USB controller to each VM (probably 3 running at same time).
Gigabyte Aorus X399 Gaming 7 VRM details.
Main Cpu Vcore vrm.
SOC vrm:
Memory vrm:
Pardon my ignorance, what’s this ACS patch you’re referring to? /me has been under a rock for a while…
Hi all - I’d like to confirm that the Corsair Dominator Platinum 16GB Kit (2x 8GB) at 3200MT/s v4.24 sticks work flawlessly on this mobo. I added 2x sticks for 32GB and it booted into XMP straight away.
CMD16GX4M2B3200C16
https://www.amazon.com/Corsair-Dominator-Platinum-3200MHz-PC4-25600/dp/B018GK2G9S
Handle 0x0046, DMI type 17, 40 bytes
Memory Device
Array Handle: 0x002E
Error Information Handle: 0x0045
Total Width: 64 bits
Data Width: 64 bits
Size: 8192 MB
Form Factor: DIMM
Set: None
Locator: DIMM 1
Bank Locator: P0 CHANNEL D
Type: DDR4
Type Detail: Synchronous Unbuffered (Unregistered)
Speed: 3200 MT/s
Manufacturer: Unknown
Serial Number: 00000000
Asset Tag: Not Specified
Part Number: CMD16GX4M2B3200C16
Rank: 2
Configured Clock Speed: 3200 MT/s
However, the universe wants to rain on my parade so my Threadripper box is down with a PSU OCP loop issue, so I’m sorting that out now.
EDIT — I spoke too soon.
Having set VCORE SOC to 1.125vdc, CPU Core voltage to 1.4vdc, both DDR4 Channel Voltages to 1.4vdc, I’m only able to get a stable 2933MT/s OC on the Corsair sticks. Guess the v4.24 kit are not B-die after all.
Handle 0x0046, DMI type 17, 40 bytes
Memory Device
Array Handle: 0x002E
Error Information Handle: 0x0045
Total Width: 64 bits
Data Width: 64 bits
Size: 8192 MB
Form Factor: DIMM
Set: None
Locator: DIMM 1
Bank Locator: P0 CHANNEL D
Type: DDR4
Type Detail: Synchronous Unbuffered (Unregistered)
Speed: 2933 MT/s
Manufacturer: Unknown
Serial Number: 00000000
Asset Tag: Not Specified
Part Number: CMD16GX4M2B3200C16
Rank: 2
Configured Clock Speed: 2933 MT/s
Seems the ‘flawless’ boot up at 3200MT/s was a fluke. I can’t get it to do it again…
@wendell Were you able to get Windows 10 installed?
I had to disable IOMMU in the UEFI to get the installer running.
Once I get to the ‘custom install’ option in the installer and I choose my target SSD (connected to mobo SATA header), I’m able to select it. When trying to create a new partition or attempt the installer, it complains saying that the hardware controller needs to be enabled in the BIOS.
AHCI is enabled in UEFI. Do you need to load any particular drivers at this point?
Now that was strange - once I disconnected all drives, apart from the one I wanted to install to - it installed smoothly. 
Yep. Is why we recommend that. Windows can see other uefi disks and assumes it needs to modify the boot loader for that
That makes sense now. It was winging that my Fedora boot was a gpt partition. Sorted now.
I plan to forward this SSD via IOMMU in the 'next ’ phase from Fedora itself.
Hi all,
Thought I’d post my progress in overclocking the Corsair Dominator Platinum DDR4 RAM (CMD16GX4M2B3200C16) - I’ve included a link to Amazon.
It took a lot more work getting these to run stable at 3200 MT/s given that the RAM I received are v4.24 E-die. Here’s the config that worked for me -
CPU VCore 1.4vdc
VCore SOC 1.125vdc
DRAM Channel 1.48vdc
XMP - disabled
Memory manually set to 3200 MT/s
Timings manually set to 16-18-18-18-36
I noticed in the UEFI the DRAM Channel voltage does fluctuate and touch 1.5vdc. It’s good that I didn’t set it to 1.5vdc directly, as that may have pushed a bit too much through the RAM.
XMP had to be turned off; with XMP I just cannot get it to POST - it may be lowering the voltage, although I’m not sure.
