I'm running Performance Test 9.0 (Passmark) Memory Mark on X299 Dark with I9-7900X.  Memory is TridentZ F4-4000C18D-16GTZ.  The benchmark is 2770 with XMP loaded.
The exact same memory/XMP has a benchmark of 3815 on my Aorus Z370 with I7-8700K.
Does it make sense that the X299/7900X is ~30% lower than the Z370/8700K?  Please advise, thanks
 

Here's my results from Sandra, looks like it's 30% lower...  note this is with BIOS at default less XMP loaded, same as I did with the Z370.
 
Aggregate Memory Performance   40.92GB/s
Integer Memory Bandwidth   40.79GB/s
Float Memory Bandwidth   41GB/s
 
I tried 4x8mb 4266mhz RAM and it wouldn't post in XMP.  At 4000mhz the board would report triple channel, not quad.  Had to go to 3600mhz to get all four modules to show as quad channel.
 
Feels like I have a board with a bad channel...

 
Here's a table where I ran 3 memory sets on 3 boards, Z270, Z370 and X299 Dark.  The PassMark software shows a 30% drop in performance for the X299.  However, Sandra shows an X299 scaling effect based on quad/Mhz increases.  The X299 has a max aggregate score of 63.8 with 4266Mhz memory set at 4000Mhz and 17-17-17-39 timings.
 
So either the PassMark software has a problem with X299 architecture or there's a board issue.
 
Can anyone please run a PassMark Performance 9 Disk Mark on their X299?
 
I'd like to see if you get the X299 2800 score that I get or something similar to the Z270/Z370 3800 scores.  Thanks!
 
 
 
 

So, what were the comparative performance results using Sandra?

I was comparing my results to the ones from Galax and since my scores are higher I am satisfied

The problem you are talking about i had on APEX and R6E many times
on FTW-K only once but fixed it with v
on DARK never