Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

THROUGHOUT THIS SERIES OF ARTICLES, we have looked at the different design parameters of HDDs, the differences between effective media transfer rates and effective host transfer rates and, moreover, how all these issues tie together and contribute individually to the general idea of hard disc drive performance. One thing we have not talked about yet is the scheme of command queuing, however, with the introduction of Serial ATA Native Command Queuing or short NCQ it is time to close this gap.

Command Queuing

In general, what is Command Queuing? What are the differences between Legacy Command Queuing and SCSI Tagged Command Queuing? Last, what is Command Queuing actually good for and how much, if any, can we expect in terms of performance increases in the desktop sector? We’ll try to go over these questions one by one and then we'll try to put everything into perspective.

Because hard disc drives are electromechanical devices, mechanical inertia is a major limiting factor for performance – access and retrieval of data require head positioning over the appropriate logical block addresses, which involves mechanical movements of the actuator and the spindle. The factors that are defining the delays incurred with respect to head-positioning are rotational and seek latencies.

Close-up and personal with the mechanical parts of a Seagate Barracuda 7200.7 (original photograph courtesy of Seagate Corporation). Note the filigree actuator design for reasons of reduced weight, which directly translates into lower inertia / better responsiveness.

Mechanical latencies can be optimized only up to a certain degree, however, efficacy of the internal workflow can be dramatically increased by intelligent, device-internal management of the workflow. The two key words in this context are “intelligent” and “device-internal”, that is, the drive itself is the only device that actually knows the precise location of the data within. At the same time, it needs to be intelligent enough for an assessment of all positional data involved - in addition to transfer latencies - in order to find the best way for the smoothest execution of the workflow.

The hardware necessary to accomplish this is, in most cases, a queue to hold incoming commands issued by the host bus adapter; within this queue, the individual commands can be reordered and rescheduled while providing a tracking mechanism for completed or else still outstanding portions of the workload. The technical term for this concept is “command reordering based on seek-optimization” or “tagged command queuing”.

A major improvement of command queuing is further that the host can disconnect from the device while the device itself autonomously reorders and then executes the commands while the bus is relinquished. As soon as the device is ready to transfer the data to the host, it can reconnect to the bus and burst the data out. This optimized transfer mode in which data are communicated in a highly optimized fashion with the most efficient utilization of the bus also eliminates unnecessary bus occupancy that can lead to priority conflicts or bus contention.

next page:    => Hard Disc Drive Architecture VI: Different Queing Schemes =>