Ok since you all may want to hear the rest of the testing story.
So I breadboarded out the old MB which did have multislot slot DDR4 and my son lent an old video card and an old monitor/keyboard to test new memory in another MB, like @Dan_Theman suggested. New memory passed POST on install (but it did that before). It passes a POST reboot, that was iffy before. So we let the old mobo boot from passmarks usb memtest. It passed two runs of it, but I noticed something odd.
The runs took much longer on old board, and memory was stock bios speed on both, and memtest did register the stats on new memory. Yet, on the old board memtest only used 1 CPU core. After some searching, the old board’s BIOS can’t handle multiple CPUs in memtest (known issue). So, how to properly test the suspect dual-channel functionality?
Since old MB BIOS supports XMP, as does new memory, and CPU has clock for it, I turned on default XMP profile hoping that would tell memory to be dual channel. And now, we fail POST on restart with same hard fail (forced shutdown) at QCode 55 (memory initialization). OK. You would think it would POST even if unstable since board/CPU theoretically capable. Sure I don’t have case/fans to control temps but it wasn’t too hot to simply POST. Go back to stock mode. BIOS does now see memory as dual channel. But memtest still is single CPU.
I redid benchmarks, back in stock mode, just to see how it compared to old memory in old board (twice as fast) … but when I switched benchmark to write from read I got this (repeatable) oddity:
I’m now leaning to thinking that this is enough faulty testing artifacts on new memory to at least hold onto new MB until I can test it with a different set of approved new memory. OFC need refund from sketchy new memory to buy …