Alan Cox [off-list ref] writes:

quoted

We have written a new block driver for our AHCI based PCIe SSDs. The
main objective of our product is providing high performance. Traffic
through OS storage stack is not to able fully utilize the device's
capabilty. To improve the traffic to the device and hence
showcase/utilize the device's capability, we have come up with this new
block driver. This driver includes
	* utilize device's increased queue depth
	* workaround for hardware errata

We want to get this driver into kernel tree to support the device out of
the box. Attached this driver as a patch for latest kernel. We would
like to get your comments, and also open for discussion.

The kernel starting point would be that we have an AHCI driver. If you
need workarounds for hardware errata then they can go into it and that is
fine. We support NCQ so we can use the queue depths. If there are
extensions then the AHCI driver can be enhanced.

Given the highly parallel nature of these parts, I wouldn't be surprised
if the ahci queue depth of 31 is one of the main bottlenecks.  Can you
think of a way to extend the ahci driver in this manner to accommodate
devices like this one?

Cheers,
Jeff