Feb 19, 2001 BY
GURU: DAVE sMITH
There have been many versions of Integrated Drive
Electronics (IDE) over the years, and all the numbers and specifications can be
a bit intimidating when you're trying to select the correct drive for a
particular system. In this article, I'll answer some of the most common IDE
questions and decipher the standard's acronyms and specs.
Why worry about IDE standards?
Today's IDE subsystems are capable of achieving phenomenal
data-transfer rates, up to 100 Mbps. In order for you to make those high speeds
a reality, however, all three parts of the hardware equation—the drive, the
ribbon cable, and the motherboard or I/O card—have to support that transfer
rate. Unless you understand the IDE standards, you won't be able to ensure that
you're getting the highest possible transfer rates from your hard disks and other
IDE devices. So, if you're unsure about the difference between ATA-2 and Ultra
ATA, or you don't know what kind of cable a UDMA/100 bus requires, keep
reading.
What is IDE?
IDE is a generic term that refers to a disk drive with a
built-in controller. The controller for
the drive is built in to the drive unit itself rather than
on a separate board. Back in the earliest
days of hard-disk technology, the hard disk and its
controller were two separate pieces. With
an IDE drive, however, the drive controller is part of the
package. You've probably noticed
that a hard disk has a circuit board strapped to the bottom;
that's the integrated drive controller. With an IDE drive, when the controller
goes bad, the drive is toast as well.
IDE was a real revolution in PC technology when it first
came out because it meant that drive
makers did not have to ensure that their drives were
compatible with a particular controller
standard. As a result, they could develop higher performance
drive/controller combinations
that spoke their own proprietary language. Today's
motherboards provide IDE connectors, but
the IDE controller is actually in each of the drives. The
motherboard merely communicates
with those controllers. The same is true for any I/O card
you might add to a PC: It merely
adds more IDE connectors and lets the motherboard talk to
them.
Does IDE=ATA?
Is IDE the same specification as Advanced Technology
Attachment (ATA)? Well, yes and no.
It didn't used to be so, but nowadays, for all practical
purposes, they are synonymous. ATA is
also called AT Attachment, a reference to the fact that the
ATA standard was originally
developed for AT computers (of which the 286 was the first
model), rather than XTs.
There have been three different versions of the IDE
interface over the years:
XT IDE (8-bit
ISA): This was the original IDE standard for XT computers (8086
models). It used
a 40-pin connector.
MCA IDE (16-bit
MicroChannel): This was a proprietary standard developed by IBM
for its
MicroChannel Architecture PC bus—a great idea that failed miserably in the
marketplace and
quickly became obsolete. It used a 72-pin connector.
ATA IDE (16-bit
ISA): This is the same 16-bit ISA, originally developed for the 80286
system, that is
still found in systems today. It uses a 40-pin connector but not the same
type as the XT
IDE.
Each of these types of IDE used a different cabling, so you
couldn't mix and match them on the
same bus. Of the three, only ATA remains; the other two are
obsolete. So when you hear
people talk about an IDE drive today, you can bet they're
referring to an ATA drive. The
other two standards have been dead for at least 10 years.
What is EIDE?
EIDE stands for Enhanced IDE. There are two ways in which
the term EIDE is used in the
industry. One way is to refer specifically to an ATA-2 or
ATA-3 type drive. (See the section
"What are the different types of ATA?") Another
way is to refer to any IDE drive that
conforms to any ATA specification other than the original
one—in other words, ATA-2 or
higher. So when you see EIDE in a drive specification, you
can't really assume very
much—the term has become almost as generic as IDE itself.
What are Ultra ATA and Ultra DMA?
These terms are interchangeable. Both refer to a
specification for improving hard-disk speed.
Ultra DMA is often abbreviated UDMA; in this article, I'll
use UDMA. PC manufacturers
will often use Ultra ATA in their system specifications, but
rest assured that it's the same as
UDMA. There are actually three different specs: UDMA/33, UDMA/66,
and UDMA/100.
You'll learn more about them in the "What are the
different types of ATA?" Section.
What is SMART?
SMART stands for self-monitoring analysis and reporting
technology. It's a standard for
predicting the likelihood of impending failure for a hard
disk. The software originated at
Western Digital and was later integrated into the ATA
standard. You enable SMART support
in the system BIOS. It's used by drives of the ATA-3
standard and above.
SMART checks a drive and establishes a threshold for its
performance in several areas. It
can then notify the user when any of those measurements
falls below the performance
threshold, possibly signaling an impending drive failure.
Some of the factors it checks are
head floating height, data throughput performance, spin-up
time, seek error rate, seek time
performance, and drive calibration retry count.
Some utilities, such as Norton Utilities, provide a SMART
status indicator as part of their
Windows-based controls, but this is not really necessary. If
SMART detects a problem, it
will pass an error message to you through your operating
system, so you can back up your
important files before your hard disk fails.
SMART is different from the active disk-checking
utilities that you might run, such as
ScanDisk or Norton Disk Doctor. These utilities check the
logical organization of the file
allocation table (FAT), followed by an optional physical
check of the disk surface. They do
not monitor any of the drive-performance factors, such as
the number of times that the drive
has to retry a certain function before it succeeds. In
addition, these active utilities must be
user-initiated, either manually by running a program or
automatically by using some sort of
scheduling agent, such as the Task Scheduler in Windows 9x.
So you really need both
SMART and a disk-utility program such as ScanDisk for best
data-integrity assurance.
What are the different types of ATA?
An organization called the National Committee on Information
Technology Standards
(NCITS) sets the standards for ATA so that all drive and PC
manufacturers have a common
set of rules to work from. NCITS has come up with six
versions of ATA over the years:
ATA-1: The
original ATA was ATA-1. It was developed in 1988, a few years after
SCSI came out. It
introduced the 40-pin connector and ribbon cable, as well as the
master/slave/cable select configuration options. It also provided
specifications for
signal timing for
Programmed I/O (PIO) and Direct Memory Access (DMA) modes and
Cylinder Head
Sector (CHS) and Logical Block Address (LBA) drive parameter
translations.
Another great feature introduced in ATA-1 was the Identify Drive
command, which
the BIOS setup program could use to auto-detect the drive
specifications.
With earlier hard disks, you had to manually set up the drive type in the
BIOS program,
either by a numeric drive type or by entering the number of cylinders,
heads, and so on.
On ATA drives, however, the drive can send this information to the
BIOS setup
program upon request.
ATA-2: ATA-2 came
out in 1996. It added the capability to use other storage devices
on an ATA
interface, not just hard disks. It also allowed for faster PIO and DMA
transfer modes,
support for PC Card (PCMCIA) drives, and support for
power-management
schemes that allowed the hard disk to spin down after a certain
period of
idleness to conserve power (especially battery power on a laptop). And
finally, it
defined CHS/LBA translation for drives of up to 8.4 GB. Other names for
ATA-2 include
Fast-ATA, Fast ATA-2, and EIDE. Not all vendors agree, however, that
ATA-2 is really
EIDE.
ATA-3: A minor
1997 update to ATA-2, ATA-3 added support for SMART technology,
added a security
mode for password protection, and eliminated the 8-bit DMA transfer
protocols. ATA-3
drives are also referred to as EIDE.
ATA-4: This 1998
revision introduced Ultra DMA (UDMA), with transfer modes up to
33 Mbps. This is
commonly referred to as UDMA/33 or Ultra ATA/33. UDMA/33
operates at twice
the speed of the fastest PIO or DMA mode. To take advantage of it,
your motherboard
or I/O card must support UDMA/33 or higher. UDMA/33 uses
cyclical
redundancy checks (CRC) to ensure that the data being written is correct.
ATA-4 also
introduced support for an optional 80-wire, 40-pin ribbon cable that helped
cut down on noise
resistance (crosstalk) and integrated the AT Attachment Packet
Interface
(ATAPI), which had formerly been a separate standard. This allowed
CD-ROM drives,
tape drives, and other removable storage devices to work under a
common interface
standard.
ATA-5: This
standard, introduced in 1999, offered UDMA/66, providing transfer modes
up to 66 Mbps.
The only glitch (and it's a minor one) is that you need an 80-conductor
cable to take advantage
of the higher transfer rate. If you have only a standard IDE cable,
the drive will
operate at UDMA/33 speeds. The drive automatically detects the type of
cable in use.
Again, the motherboard or I/O card must be capable of UDMA/66 in order
to get the top
speeds. By the way, the card has only the normal 40 pins. The extra 40
wires are grounds
placed between each data wire to provide shielding. In theory, the
normal 40-pin
wire could work if it was of sufficient quality and shielded, but it's not
cost-effective.
Note that the 80-conductor cable is much more delicate than the standard
40-conductor
cable because the wires have to be finer to fit in the extra 40 wires.
ATA-6: This was a
brand-new standard in 2000, offering UDMA/100 with transfer
rates of up to
100 Mbps. As with ATA-5, you must have an 80-conductor cable;
otherwise, the
drive will revert back to UDMA/33 performance. And again, the
motherboard or
I/O card must be UDMA/100-capable; if it isn't, the drive will revert to
the top speed
that it does support: UDMA/33 or UDMA/66.
How does rotational speed fit in?
Some drives are advertised according to their rotational
speed—that is, the speed at which
the drive spins its disks or platters, in revolutions per
minute (rpm). Generally speaking, a
higher rpm means higher drive performance. An rpm of 5600 is
typical of a UDMA/33 drive,
while an rpm of 7200 is commonly found on UDMA/66 drives.
Because a drive advertised as 7200 rpm typically comes with
an 80-conductor ribbon cable,
you might get the idea that the two factors are somehow
connected. Not so. A 7200-rpm drive
will operate at 7200 rpm regardless of the type of ribbon
cable it uses. The rotational speed
is a separate factor from the UDMA mode (that is, the data
transfer rate). A 7200-rpm drive,
however, is generally a much better performer than a
5600-rpm drive in terms of seek time
(the amount of time the drive takes to get to the data) and
often in actual transfer rate.
Increases in data density could conceivably result in a
lower-speed, higher-density drive that
has a faster transfer rate than a lower-density,
faster-speed drive.
Can I mix and match ATA versions in a system?
You can mix and match ATA versions in a system, but you
might cheat yourself out of the best
performance. Whenever all the conditions are not met for a
particular level of ATA
performance, the drive drops back down to the performance
level that the entire subsystem
can mutually agree upon. (Each IDE connector on the
motherboard or I/O card is a separate
subsystem.)
Suppose, for example, you have a brand-new UDMA/66 drive. In
order for it to function at
UDMA/66 levels, all of the following must be true:
The motherboard
or I/O card must support UDMA/66.
You must be using
an 80-conductor ribbon cable.
There must not be
any other drives on the same cable that have a lesser ATA
specification.
If any one of these factors is not in place, the drive will
still function but not at UDMA/66
levels.
That last item on the list—the requirement that the device
not share a cable with a device of a
lesser ATA specification—can be a real kicker, especially in
a system that's full of IDE
devices. If your motherboard has only two IDE connectors and
you have four IDE devices
you need to use, you might not have a choice about the hard
disk sharing a cable with some
other device. If the highest UDMA performance is important
to you and another device on the
chain is slowing things down, you might want to buy a
separate I/O card for some of your IDE
devices so that the UDMA/66 hard disk can be on its own
cable. If you absolutely must put a
fast drive, such as an UDMA/66 drive, on the same cable as
another device, choose a
CD-ROM drive rather than a lesser hard disk and make the
lesser device the slave.
How can I maximize IDE performance on an existing
system?
The key to maximizing IDE performance on an existing system
is to identify the bottleneck in
the system and then eliminate it. Your first step should be
to identify the motherboard's native
IDE support mode. Does it support UDMA/33, UDMA/66, and/or
UDMA/100? Perhaps the
documentation will tell, or perhaps there will be something
in the BIOS Help about it. The
chipset on the motherboard determines this; for example, the
Intel 810e and 820 chipsets
support UDMA/66, and the Intel 815 chipset supports
UDMA/100.
Next, check the drive that you want to use with it for the
same support capability. Check the
BIOS under IDE devices for the model number. If that fails,
look on the outside of the drive
for a label that might provide a clue or look up the drive's
model on the manufacturer's Web
site to get its specifications.
If your motherboard is going to hold the drive back (for
example, a motherboard that supports
only UDMA/33), consider adding a high-speed PCI-based IDE
interface card that matches the
capability of the drive. Don't forget, too, that you need an
80-conductor cable for UDMA/66
and UDMA/100.