I have two Samsung 830 ssd drives in an Intel Raid 1 system on an intel x79 chipset (Intel 3930K cpu) using the most recently officially released Intel rste drivers (version 184.108.40.20620). Running Windows 7 x64 sp1.
Gpt fdisk (a diagnostic utilility, link at the end) will report the gpt backup partition table is corrupted (CRC check fails), and after repair, it will report the gpt backup partition is correct. Then ten or fifteen minutes later, gpt fdisk will once again report corruption in the gpt backup partition table (CRC error).
I have discussed this with the author of the gpt fdisk utility, and he believes that the intel raid drivers may be dynamically updating raid metadata and corrupting the gpt backup partition in doing so. Can someone else confirm this, or offer any advice on testing this out? Thank you. I don't see any documentation on the Intel raid metadata structure for windows.
I notice that another utility, mbrwizard, also shows that the gpt backup partition is "invalid"; something is definitely writing into that area.
One theory I have is that SSDs using unpartitioned (unallocated) space at the end of the drive for manual overprovisioning may be incompatible with the full GPT spec. (perhaps the ssd thinks it can use that location near the very end of the drive, where the GPT backup partition information is, as a scratchpad?).
If anyone with GPT partitions on SSD drives or intel RAIDed drives or both can run mbrwizard, gpt fdisk, or some other gpt-compatible utility, to see if their gpt partitions, both main and backup, are valid (passing crc checks), that would be helpful to know. Typically I see a cycle within half an hour of valid followed by invalid, and it's hard to figure out what is going on.
Any other suggestions are welcome. (chkdsk and raid verification report no problems--only the gpt backup partition is getting corrupted.)