So i tested Ultra M.2 on X99, where it makes much more sense since CPU bandwidth isn't being affected if you use PCi-E 3.0 lanes for storage. On Z97 if you do this you lose the ability to run 16x on a GPU, unless there is some extra chip added to expand the CPU's lanes. On X99 since there are 40 PCi-E lanes on the top tier CPUs, you can still run GPUs in most configurations without bandwidth limitations.
There are like 2 types of 4x M.2 slots. One type if PCI-E 3.0 which produced 32Gbps, the other way to do it is PCi-E 2.0 and it produces 20Gbps.
There is only really one M.2 drive that can even take advantage of 4x lanes and that is the Samsung XP941 and I got that to 900+ MB/s on both the PCi-E 2.0 and PCi-E 3.0 based X99 boards, but only the sequential read even takes advantage of it.
The rest of the M.2 drives are limited to like 500MB/s to 600MB/s for sequential read and they wont really benefit from anything above normal M.2 standards.
ASRock X99 OC Formula (4x PCi-E 3.0=Ultra M.2): 985MB/s
GIGABYTE X99-SOC Force(4x PCi-E 2.0 = Turbo M.2): 975MB/s
GIGABYTE Z97X-Gaming G1(2x PCi-E 2.0 = Normal M.2): 765MB/s
THose are from memory so I might be a bit off, but you get the gist.
BTW I did have issues booting M.2 on the C7Z97-OCE but other boards were fine for the most part. BTW M.2 drives get hot and I am told that is normal, but damn it was freaking hot.