Yeah, that's the issue that brought up my question.
The Z68 I believe is a 6 series board while I'm currently on a late x58. My hope was that with write-back caching enabled, the RAID chipset would write directly to cache and return as if the IO completed already (this is how it's described in the manual) which being in a mirror set, I would've imagined the data IO would be processed by the fastest drive that can complete the request. Even with a 1 TB mirror setup across 2 normal HDDs the performance is doubled both read and write unless there's a specific reason why the SDD would wait on the HDD to complete its read request instead of passing it through and vice-versa for writing. As far as the capacity, I was planning on just putting the mirror on the outer tracks somewhat like short-stroking it to maintain the best speed.
The writing performance I imagine would degrade to the SDD speed level once the cache had been saturated but I'm afraid I don't see why read performance would drop along with it. If mirroring allows performance gains in read by being able to access multiple drives simultaneously then why would the chipset request the same data from both disks and wait for both of them to retrieve it before completing the IO request? If it were, then I would imagine my RAID1 testing wouldn't have doubled in read speed. I guess I would expect the performance the be somewhere in between the two drives performance. Generally I would just rely on backups although I am trying to gain more space on my system disk. Shame Win8.1 Enterprise didn't ship with ReFS or I would just use its hierarchical storage.
Oh well. I do intend to try and break up one of the RAID1 mirrors and see what the result is in the least. I'd like to see the result first since they have no data on them at the moment and I just lost a 2TB archive of 15 years of history I am trying to avoid at all costs. 