I can comment on this. I recently switched from 1 x MV8 to 2 x M1015
First of all, HDDs can't even reach max speed of SATA II (3GB/s), so SATA III or SATA II, that doesn't affect much.
and there is almost no performance gain in terms of copying stuff to or from unraid when using M1015 or MV8.
HOWEVER, you see the improvement when it comes to Parity Check or Data Rebuild.
M1015 uses PCIe x8, which is able to do 1.6GB/s single direction.
MV8 uses x4, which is able to do only 800 MB/s
so let's say you have 8 drives plugged in to one of these cards. and let's say your hdd's are able to read 110 MB/s max (this is what i saw during preclear for most drives during pre-read).
If you were to use MV8, your speed will be limited to 800 MB/s or less. it can never reaches that high. that's just theoretical speed.
With simple feature stats plug in, you can actually see the transfer speed of all drive during data build / parity check.
I was getting about 500 MB/s. again, this is from SIMIPLEFEATURE STATS. On the index page, i was seeing about 80 MB/s for parity check.
When I switched over to M1015, I was getting about 700 - 800 MB/s total.
and about 100 - 120 MB/s during parity check. (i also removed green drives from my array which increased my speed as well).
SO with MV8, parity check took 10 - 11 hours, with M1015, took about 7 - 8 hours or less. it is a big improvement in my opinion.
So, to sum it up.
In regular usage, there is no difference because you're accessing 1 or 2 drive at time, so it doesn't get limited by PCIe speed.
However, when you're doing the parity check or data rebuild that uses all 8 drives at the same time, you see improvment.