So... I've been trying to optimize the speed of my developing in Lightroom... and I've been running tests to see what adjustments/formats work fastest. I'm using a Canon 5D3 and recently discovered that mRAW develops significantly slower than RAW. What I can't understand is why converting to DNG would prove slower for developing, when it's meant to be optimized for fast loading. It seems to "load" initially 10% quicker, but with additional developing adjustments applied and when moving from photo to photo, it proves 15% slower...
I converted with no jpegs embedded, with fast load data enabled & I'm using a 6-core CPU (12-core with multi-threading), and 32GB RAM, so I can't see any reason for the slowness.
Your computer certainly seems to have the specs such that you should expect ultimate performance from Lightroom or any other program as far as that is concerned. But having all that RAM and high specs doesn't seem to be relevant with Lightroom. I have read numerous reports of people with both Windows and Mac computers with "extreme" specs who experience performance issues. I'm using a very basic Windows 7 computer with 8 GB RAM, and I don't have any performance issues whatsoever. No special graphics card or anything.
I too have read of the performance boost DNG files are supposed to provide. I have experimented and compared using DNG's versus my NEF files. And, like you, have noted a slowdown if anything when using DNG. I know this isn't answering any of your questions but I don't think there is a definitive answer to your dilemma.
The only factor that I can think might influence this is that I'm working with Lightroom 4.4. Can anyone else think of a reason why DNG files with fast load data would perform slower than native Canon 5D3 RAW files? Initially the perform faster, but with any adjustments enabled, they perform slower.