Well that's very odd, because I would have thought that our BIOSes would be quite similar. Here you can see mine with every core at 100%, yet it still has every core at 3.8GHz:
Now in my experience with past mainstream quads, Turbo has never worked very well, but it did at least try. For example, the 2600k supposedly did 3.8 with a single core loaded, but in practice I only saw 3.7 or below, and it was all over the place. But with every core loaded, it always clocked down to 3.4 (or was it 3.5? - I think the base clock was raised 100MHz with turbo enabled). This one is doing the opposite.
Regarding the Turbo Boost 3.0, I know mine is enabled, because I can see the core prioritizing aspect of it working. UT2004 is a good candidate for testing, and here you can see it constrained to the "best" core as defined by intel in the core list of the Turbo CP.
(in case it isn't clear, that last graph in task manager is pegged at 100% - usually you'd see the load distributed across multiple graphs because windows is constantly shifting the thread)
So I don't know why the 4GHz clock is missing even though Turbo Boost 3.0 is clearly operating.
One other thing: With Broadwell-E, are the VID voltages being reported by HWiNFO actually VCORE? Because I thought VID was the target voltage level set by intel, and static for each clock state. These "VID" values are constantly fluctuating (subtly).