Hard Drive
Media Transfer Rates

(Cont.)

If you have arrived here through a search engine, and there's no menu to the left click here!

As you have read at the start of this segment, during the last three years we have moved quickly away from what was thought to be the fastest ATA Interface yet, Ultra ATA/66. At the time the Ultra ATA/66 interface was being introduced, faster processors and memory, faster motherboard chipsets and faster hard drives were all emerging on to the market. These speed improvements brought about faster data strobe rates, potential for line noise and the need for better cables, among other things. Now the industry has seen yet another first, matching the burst transfer rate to the speed of the PCI bus. All of these improvements however, are not without problems inherent to the technology used to achieve them.

Issues that Effect Transfer Rates

Some issues that affect transfer rates are inherent to the AT Attachment Interface, while others result from the design and assembly of the hardware system itself by the manufacturer. Other issues arise as the result of oversights on the part of home system builders, or a users attempt at performing upgrades. In addition to these issues, in this final portion of the segment, we will address what is, and isn't, possible (deciphering truth from hype) and then address those more prevalent problems. Let's begin with some of the factual issues surrounding Ultra ATA.

The Facts:

Other Issues:

  1. Mixing Device Types on the same IDE bus -
    The goal of the Ultra ATA/66 specification was to double the burst transfer rate of the interface, from 33 Mbytes/sec to 66 Mbytes/sec, while keeping other significant changes to the Ultra ATA specification to a minimum. The only changes required, other than a Ultra ATA/66 drive and motherboard controller, was the addition of an 80-wire 40-pin IDE cable. This cable contains the exact same pin terminations as the previous 40-pin (conductor) IDE cables, with the exception of an additional 40 “ground” wires interleaved in between the existing wires. These additional ground wires serve to reduce or eliminate “EMI” signals, and allow the data lines to stabilize more quickly after each pulse of transmitted data. Since the lines can stabilize more quickly with the additional ground wires, the amount of delay, or “setup time”, can be halved, effectively doubling the maximum transfer rate.

    Although the Ultra ATA 66, 100 and 133 specifications ensure backward compatibility, the IDE bus is not capable of switching modes on the fly. The motherboards' IDE controller, as well as all of the devices attached to it, including the ribbon cable itself, must be of the same specification, either Ultra ATA/66, 100 or 133. If the ribbon cable, or any one of the individual devices is not compliant, the controller will revert to the speed of the slowest Ultra ATA, DMA or PIO Mode device on the bus.

    For example, let's presume that you have a brand new computer that is fully Ultra ATA/133 capable. The computers motherboard IDE bus controller, hard drive and ribbon cable are fully compliant. The hard drive is the only device on the primary IDE channel (bus), and the CD Rom drive is on the secondary IDE channel. You decide to purchase a second Ultra ATA/133 hard drive, as well as a new CD-ROM burner. Next, you decide that you want the second hard drive on the secondary IDE channel as a master and the new burner on the primary IDE channel as a slave. After installing and checking everything, you notice that when transferring large files, the process is notably slower after the upgrades. Even though the hard drive is Ultra ATA/133 compliant, the original CR Rom drive is not, nor is the CD Rom burner. By placing them on the same bus with the hard drive, the Ultra ATA/133 bus automatically slows down to the speed of the slowest device on the IDE bus, which is ATA/33 for some CD Rom drives, and as slow at 16 Mbytes/sec for others.


  2. Electro-Magnetic Interference -
    Electro-Magnetic Interference is just one of the factors addressed in all electronic components that are installed in today's computers. Maximum EMI levels are mandated by standards and regulations in almost every part of the world. In standard computer design, EMI is usually first controlled at the motherboard level and contained by the chassis or enclosure. The containment of EMI is often referred to as EMC (Electro Magnetic Containment). Aside from the possible health risks posed by EMI emissions, (which are minimized in today's computers) EMI emissions are often a factor when reviewing the performance of various computer components. As in the case with the move from Ultra ATA 33 to Ultra ATA 66, the strobe pulse created across the IDE data path was sufficient so as to require the addition of 40 ground wires to the former 40-wire 40-pin IDE bus ribbon cable to stabilize the path and reduce EMI emissions at the path level.


  3. In addition to specifying 40 additional grounds wires for the Ultra ATA 66 (as well as Ultra ATA 100 and 133) IDE ribbon cable, engineers also specified that the cable not be longer than 18 inches (457.2 MM). Although this presents a problem for today's taller tower cases, limiting the length of the ribbon cable also isolates, to a degree, the EMI signals that might affect data traversing the cable. Longer lengths, in lay terms, turns the ribbon cable into a form of antenna for signals emanating from sources other than the hard drive or motherboard.

  4. Redesigning the ATA 66, 100 and 133 Cable - There have been many attempts at shielding longer ribbon cables in order to enable their use without suffering a performance hit in the process. Technicians have tried just about anything you can think of in order to use their longer ribbon cables, even wrapping them in foil. Even when the rounded ATA 66, 100 & 133 cables first surfaced, everyone thought they might be the answer. To a certain extent they were, but even those first rounded cables were problematic at longer lengths, until recently that is.


  5. The early rounded cables were a great attempt at improving system cooling over flat ribbon cables, and to a certain extent, they were an improvement in EMI reduction, however extended lengths were still problematic. However, a recent redesign of the rounded cable has added additional shielding as well as a specific grounding strap. See Figure 1.
    Figure 1


    The unique design and construction of these high-performance cables takes shielding to a new level. Data signal wires and ground signal wires are arranged in an alternating pattern so that data wires are completely surrounded on all sides by ground wires. See Figure 2.

    Figure 2


    As you can see in Figure 2 above, each layer of wires is separated additional aluminum shielding, and then the entire matrix is wrapped in a grounded metal mesh and protective resin. These new cable designs eliminate, for the most part, the additional cross-talk resulting from extended length IDE cables. These are ideal for the performance-minded computer user. You'll find these and other performance items at our on-line store.

  6. Does a larger drive cache make a difference? -
    If you're not familiar with the term cache, or drive cache, we are referring to the memory cache built into today's newest hard drives. Not that long ago drives didn't have a memory cache, and now today cache sizes range from a half megabyte to eight (8) megabytes and more. The purpose of this cache, in overly simplistic terms, is to improve the transfer of data, which is the subject of this segment.

    It was long thought that by increasing the size of the on-drive cache, substantial improvement of media transfer rates would be the result, however this doesn't seem to be the case in many instances. Soon after Western Digital released their their "special edition" drives, we tested two of them. The first, a WD800JB, which was an 80GB 7,200 RPM drive with an 8MB buffer (cache) and the second, a WD800BB, which was an 80GB 7,200 RPM drive with a 2MB buffer (cache). These drives range in capacities from 40GB to 200GB and 20GB to 200GB respectively. Without allot of discussion, these drives are essentially identical with the exception of the buffer (cache) size. In truth, whether on-drive caches are beneficial or not depends largely on the system design, not merely whether or not a drive has a memory cache.

What does the future hold?
Among other things, a complete redesign of the IDE interface. Serial ATA is looming on the immediate horizon with a theoretical burst rate of 150 MB/s in 2003, and 300 MB/s by 2005. In addition to increased burst rates and speed, Serial ATA brings with it an entirely new interface, one that will permit "hot plugging", more commonly referred to as hot swapping very much like that used in today's mid-range and high end servers. For a more thorough review of ATA/ATAPI history, visit Technical Committee T13 and Leroy's Engineering Web Site. For more information on Serial ATA, you can visit this section of our web site devoted to Serial ATA, or the Serial ATA development section of Maxtor Corporation's web site.

 

Notice: Windows® 95, Windows® 98, Windows® NT, Windows® 2000, Windows® XP and
Microsoft® Office are registered trademarks or trademarks of the Microsoft Corporation.
All other trademarks are the property of their respective owners.

Copyright ©1995-2002 DEW Associates Corporation. All rights reserved.