One bit of news that came out of IDF last week was word from the SATA-IO—the standards body that governs the Serial ATA specification—that it has completed work on a physical layer tweak to allow 6Gbps transfer rates, double that of current SATA devices. The organization decided to go ahead with the release of the 6Gbps physical spec now, while work continues on the full SATA 3.0 spec, which is due for completion before the end of the year. To better understand the upcoming changes to Serial ATA, we spoke with Amber Huffman, Principal Engineer in the Storage Technologies Group at Intel. Huffman is responsible for many storage technology initiatives at Intel, including both SATA and AHCI.
Huffman told us SATA 6Gbps devices and hosts should be backward compatible with devices and hosts based on older SATA revisions, as should current SATA cables. However, the SATA-IO "recommends utilizing quality components to ensure data integrity and robust operation at the fast 6Gbps transfer rate," and Huffman said there may be some tweaks to SATA connectors for 6Gbps. Although achieving 6Gbps transfer rates may require slightly more power use during transfers, Huffman expected that effect to be offset by quicker completion of transfers, placing SATA 6Gbps on roughly equal footing with SATA 3Gbps in terms of power consumption.
Part of the impetus for the move to 6Gbps SATA, Huffman said, is the advent of SSDs capable of saturating a 3Gbps interface, and she predicted SSDs would be the first devices to make use of the new spec. Another impetus for SATA's move to 6Gbps is the fact that Serial Attached SCSI, or SAS, has already made the move to 6Gbps, and the interoperability between the two standards has pushed SATA. Samples of the first SATA 6Gbps devices are expected early next year, with end products coming in the second half of '09.
In addition to announcing the 6Gbps physical spec, SATA will be getting a new logo program, as well. The SATA-IO has certified SATA devices for compliance and interoperability for some time now, but it hasn't had a logo for it. Now, you may have one more thing to peel off of the wrist wrest when you buy a laptop. Certified devices can bear the new "SATA certified" logo.
The SATA-IO is cooking up several new features that may wind up in the final SATA 3.0 spec, including an enhancement to Native Command Queuing intended for streaming data: an abort command. For instance, if the data requested via a read command has become irrelevant during the wait for that request to be fulfilled, the host can send a command to abort the still-outstanding read request. Another much-needed feature is the delivery of power over eSATA, and Huffman told us the main body of work for defining that spec is going on now.
Since we had her cornered, we took the opportunity to quiz Huffman about a few other matters. One of those was the interaction of Native Command Queuing and SSDs. We're familiar with NCQ as a means of dealing with the seek and rotational latency inherent in mechanical hard drives, but wondered what need there was for NCQ with SSDs. (Intel's just-announced SSDs have NCQ listed prominently among their specifications.) She said that in the case of SSDs, NCQ has the primary benefit of hiding latency in the host system. If you look at a bus trace, said Huffman, there's quite a bit of time between the completion of a command and the issuance of another one. Queuing up commands can keep the SSD more fully occupied.
Satisfied with this answer, we moved on to the enduring mystery that is AHCI support. Even at this late date, several years since AHCI first came on the scene, common chipsets from ATI/AMD and Via still don't support AHCI properly, a fact we've pointed out tirelessly. The chipset makers have had few good answers, alway promising a fix in the next driver or chip revision and never—to date—delivering. We've found that the best solution to these problems usually is to disable AHCI entirely, which means one loses ACHI's primary benefits, including SATA hot-plugging, NCQ, and some power-management capabilities.
Huffman was careful not to reveal whatever she might have known about the source of problems for AMD and Via, but she did share some insights about how AHCI compliance is handled. AHCI, or the Advanced Host Controller Interface, is governed by a different group than SATA proper. That group does not collect fees from its members or charge royalties for the use of its creations. As a result, it has no money to conduct an interoperability testing program and doesn't certify AHCI implementations. The thinking here, she explained, is that the participants ought to be sufficiently capable of doing the legwork for themselves.
Ah. Would that it were so.