Original Link: https://www.anandtech.com/show/543
Ultra ATA/66 versus Ultra ATA/33 Comparison
by Scott Gooden on May 16, 2000 12:00 PM EST- Posted in
- Storage
The Ultra ATA/66 interface has been with us for a little over a year now, and most of the hard drives currently available support this standard. With the promise of double the transfer rate over the Ultra ATA/33 interface, Ultra ATA/66 drives seem to be the Holy Grail to our hard drive performance problems. But just how much benefit does the Ultra ATA/66 standard have on the drives that are currently available on the market?
While this comparison is not an attempt to downplay the importance, or need, for the Ultra ATA/66 interface, as current hard drives are becoming increasingly faster demanding more data transfer bandwidth, it does help to show that the choice between these two interfaces has very little to do with performance when using current generation drives.
There is no doubt that the aging Ultra ATA/33 standard will quickly become inadequate to keep up with the ever-increasing performance of hard drives. Industry analysts estimate that hard drive performance will continue to grow at a rate of 40% each year, and with current hard drive performance quickly approaching the Ultra ATA/33 barrier, the Ultra ATA/66 standard is truly needed, as is the future Ultra ATA/100 standard.
Your need look no further than our own review on the Maxtor DiamondMax Plus 40 (Ultra ATA/66) hard drive to see how the current hard drive technology is right at the edge of the Ultra ATA/33 capabilities and will soon need the extra breathing room offered by Ultra ATA/66. Turning in sustained data transfer readings greater than 30 MB/s on the outer tracts, the Maxtor DiamondMax Plus 40 series of drives, would utilize all of the Ultra ATA/33 specs bandwidth.
While we are not trying to downplay the need for the Ultra ATA/66 standard, we do think it needs to be looked at for what it is: a standard, not a performance booster. While the implementation of the Ultra ATA/66 standard does offer slight performance gains in current drives; these gains are mainly due to the burst transfer rate of data, and not the sustained transfer rates. If the requested data happens to be in the cache memory of a hard drive, this data can be transferred in bursts at up to 66 MB/s. This “burst” transfer of data provides a boost for business type applications, where the odds that your next requested block of data has already been pre-loaded into the drives cache are greater.
If your application requires more data than can be fit into your hard drives cache, or if the data requested is not preloaded into the cache memory, then you drives data transfer rate will drop down to the actual physical media transfer rate. Currently, this is below the even the 33 MB/sec of the Ultra ATA/33 standard, although some of the newer drives are fast approaching this threshold. In everyday use of your computer, you will most likely encounter a combination of the above two scenarios, so although you will notice a slight performance gain, it won’t be quite as dramatic as the advertising on the box makes it out to be.
So to summarize, the purpose of this article is not to discredit the need or benefits of using the Ultra ATA/66 standard, but rather to objectively compare the Ultra ATA/66 standard against the older Ultra ATA/33 standard, to see just how much performance the current generation of drives are able to gain by utilizing the newer standard.
Implementing Ultra ATA/66
Before we get too far into our discussion, there are a few concerns you should be familiar with before implementing Ultra ATA/66 into your system. Some of the obvious considerations are that you have an Ultra ATA/66 compatible motherboard or controller and Ultra ATA/66 hard drive, and that you utilize an 80 conductor, 40-pin IDE cable. But there are some other considerations too.
As most of you are well aware, a single IDE channel can support up to two devices maximum. In order to operate your drive at Ultra ATA/66 speeds, all of the devices on that channel must also be Ultra ATA/66 compliant. This means that you will not be able to keep your old PIO mode 4 CDROM drive or DVD drive on the same channel as your new Ultra ATA/66 hard drive without hindering performance.
At first this may not seem to be much of an issue, until you stop to consider that devices on IDE channels are usually arranged so as to provide optimum performance. Let’s say you have one hard drive, a CD/DVD player and a CDRW drive in your system. The optimum configuration has been to place the hard drive as master on the primary IDE channel, the CD or DVD reader as slave on the primary channel, and the CDRW drive as master on the secondary IDE channel. Those of you with a second hard drive would follow the above setup, except the second hard drive would be master on the secondary IDE channel and the CDRW drive would then run as slave.
The advantages to such a setup are that most drive-to-drive transfers could be performed using both IDE channels, as only one device can be active at a time on one channel. So using the above setup, transfers from your hard drive to your CD burner, or transfers from your CD/DVD drive to your CD burner, would utilize both channels and you would realize a greater data throughput. If you needed to install programs off of a CD, you could simply use your CDRW drive as the reader, thus keeping reads and writes on separate channels for slightly quicker installation times. Those of you with the second hard drive on the secondary IDE channel could also see gains by placing your swap file on the second hard drive as well. This would allow your system to access data from your main hard drive and swap file at the same time, increasing performance slightly. This arrangement works out quite well for the average user.
Now enter Ultra ATA/66 into the above configuration and we start to run into trouble. Assuming our system is Ultra ATA/66 compatible and that we are using the proper 80-conductor cables, we still have the problem of a non-Ultra ATA/66 device dragging our performance down. If we had 2 Ultra ATA/66 hard drives, we could put them both on the same channel and then put the CD devices on the secondary channel, but then that negates any of the benefits we previously gained by using the secondary hard drive for the swap file, or for CD to CDRW transfers. While this may not be a problem in certain situations, it does cause grief for someone trying to maintain the optimal setup. Unless you use an add-in IDE controller card, such as those manufactured by Promise, you are limited in your configuration choices, and no matter what you choose, your performance will be less than optimal.
While this article does not intend to provide the magic bullet to solve the above dilemma, it should help to shed some light on just how drastic of a performance hit you will see if you choose one configuration over the other.
A Brief Background
To see why the addition of a slower device prevents us from running at Ultra ATA/66 speeds, let’s take a closer look at the technology behind Ultra ATA/66.
The goal of the Ultra ATA/66 specification was to double the burst transfer rate of the interface, from 33 MB/sec to 66 MB/sec, while at the same time keeping other significant changes to the Ultra ATA specification to a minimum. This goal was met, as the only change required, other than the obvious need for an Ultra ATA/66 drive and controller, was the addition of an 80-wire IDE cable. This cable contains the exact same pin terminations as the previous 40-conductor IDE cables, however, an additional 40 “ground” wires are included. These ground wires serve to eliminate “noisy” signals and to 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.
Since the Ultra ATA specification makes a point of ensuring backward compatibility, and IDE channels are not very good at switching modes on the fly, all of the devices on the cable must be Ultra ATA/66 compatible (including the controller) or else the controller will revert to Ultra ATA/33 speeds.
The Drives
We picked three of the top performing 7200-RPM, Ultra ATA/66 drives that we used in recent testing and ran them through a series of tests, first using the 80-conductor Ultra ATA/66 cable, then again using the standard 40-conductor IDE cable. The use of a 40-conductor IDE cable reverts the Ultra ATA/66 controller to Ultra ATA/33 speeds, as the presence detect conductor in the cable is not broken. It is this broken (or open) signal that the controller looks for, to determine if the proper Ultra ATA/66 cable is being used. If it fails to detect this signal, it automatically reverts to Ultra ATA/33 speeds.
The three drives that we used in this comparison are:
Maxtor DiamondMax Plus 40 (52049U4)
Seagate Barracuda ATA 28.5 GB (ST328040A)
Western Digital Expert 27.3 GB (WD273BA)
|
The Drives |
||||
|
Drive |
Rotational Speed |
Cache |
Seek Time |
Capacity |
|
Maxtor DiamondMax Plus 40 |
7200 RPM |
2 MB |
< 9ms |
20.4GB |
|
Seagate Barracuda ATA |
7200 RPM |
512 KB |
8 ms |
28.5GB |
|
WD Expert WD273BA |
7200 RPM |
2 MB |
9 ms |
27.3GB |
While the results for the Ultra ATA/66 tests are the same as in our recent reviews, this discussion is not about determining the best performing drive. It is meant to illustrate any differences between using the same drive under Ultra ATA/33 or Ultra ATA/66 modes. We just wanted to include a few different drives in the testing to help even out inconsistencies and draw firmer conclusions from the resulting data.
If you want more information about any of the drives used in this comparison, please refer to the March 2000 roundup or the Maxtor DiamondMax Plus 40 review.
The Test
The AnandTech storage test bed is currently set up using Windows 98, Windows NT 4.0 Service Pack 6 and Windows 2000. Each operating system is set up using its own 4 GB partition with Windows 98 using the Fat 32 file system, Win NT 4.0 using the NTFS file system and 2 instances of Windows 2000, one using the FAT 32 file system and the other using NTFS. Since Windows 2000 supports both file systems, we decided to set up two platforms to help compare the performance under both situations.
In all instances, the only installed software is the operating system and the benchmark programs. The test drive is being run as the "D:" Drive in all instances, operating as master on the secondary IDE channel. The operating system and all benchmark programs are contained on the boot disk, or the "C:" Drive, which operates as the master on the primary channel.
We followed the same procedures as we do for our normal hard drive testing, with the only difference being that we substituted a normal 40-connector cable for the Ultra ATA/33 tests.
The complete test bed is as follows:
· Intel Celeron 500 MHz CPU
· Intel CA810E motherboard with native Ultra ATA/66 support
· 128 MB Corsair PC100 SDRAM
· Ultra ATA/66 40-pin, 80-conductor HDD cable
· Microsoft Windows 98
· Microsoft Windows NT 4.0 SP6
· Microsoft Windows 2000 (FAT 32)
· Microsoft Windows 2000 (NTFS)
· Ziff Davis Winstone 2000 CC
· Ziff Davis Winstone 99
· Ziff Davis Winbench 99
· Adaptec Threadmark 2 (Windows 98 and NT4 only)
Each hard disk was partitioned and formatted before each suite of tests on their respective operating systems, as to prevent any skewing of the test results. For purposes of consistency, each benchmark was run a total of 5 times, with the final score being the average of those five runs. The machine was rebooted between each test as to minimize the effects of any carryover from the previous tests.
As you can see from this first test, the Ultra ATA/66 interface offers no advantage over the Ultra ATA/33 interface with current drives in the raw data transfer tests. The Maxtor unit is coming close to the limits of the Ultra ATA/33 standard though, and we can see that the faster interface is definitely going to be needed in the very near future.
In the Business Winmark tests, you can see a very slight performance drop when using the Ultra ATA/33 interface versus the Ultra ATA/66. This is most likely due to the cache memory being able to transfer its contents over the bus at faster speeds under ATA/66, but even so, it shows how little the interface affects performance.
Here again we can see that the Ultra ATA/66 interface offers a slight performance boost over the Ultra ATA/33 interface. The results are inline with the Business Winmark tests and can be attributed to the hard drives caches memory able to utilize the higher bandwidth of the Ultra ATA/66 interface.
As you can see, under Business Winstone there is virtually no difference in performance using either the Ultra ATA/33 or Ultra ATA/66 modes.
Under Content Creation we again can see a performance hit using the Ultra ATA/33 interface. It is pretty uniform and again we would suspect it is due to the cache memory being able to utilize the higher bandwidth the Ultra ATA/66 interface allows.
Now under Adaptec Threadmark the results are a bit surprising. The highly rated Maxtor unit shows a big performance hit when running under the Ultra ATA/33 interface, however the other drives show just a slight performance hit.
The results were so surprising that we ran the tests a few more times, than the standard 5, just to make sure it wasn’t something we overlooked. Although we don’t have a definite answer as to the large performance hit, as it seems to be too large to attribute just to the higher bandwidth that the drives cache can utilize, it does seem to scale according to the higher transfer rates.
Under Windows NT4 and Windows 2000 the results followed the same pattern as those under Windows 98, so we will just let you look at the results on your own.
Under Windows 2000 NTFS, the test results fall into the same general pattern as the FAT 32 scores. The tested scores are slightly lower in some cases under NTFS, this being due to NTFS’s slightly higher overhead than with the FAT 32 file system, however the overall pattern is the same. Since the FAT 32 results are basically a repeat of the results from the NTFS tests, we will only list the NTFS scores.
Other Devices
If you managed to survive reading through all of those charts, you have probably come to the same conclusion that we have; although definitely needed for future bandwidth capacities, the Ultra ATA/66 interface only offers a slight performance boost to current generation drives, and then only when running certain applications.
Another question that we frequently get asked here at AnandTech is if running a non-Ultra ATA/66 device on the same IDE channel as your Ultra ATA/66 hard drive will affect its performance much. We outlined some situations you might encounter when setting up your system at the beginning of this article, and we will now take a brief look at how much of an impact a non-Ultra ATA/66 device on the same IDE channel as the hard drive will have.
By using the information gained in the Ultra ATA/33 versus Ultra ATA/66 testing, we can already see that if there is a performance hit when running in the above outlined configuration: it will be minimal and would only affect us under certain situations. While this is true for the current generation of drives, as we can see by the Maxtor DiamondMax Plus 40 unit, hard drives will be utilizing the Ultra ATA/66 bandwidth more fully, in the very near future.
Since the Maxtor unit was the only one that came close to using all of the Ultra ATA/33 specifications bandwidth, we decided for the purposes of this brief test we would only use it to demonstrate the affects a non-Ultra ATA/66 device on the same IDE cable as a Ultra ATA/66 drive has.
The following tests are the exact same as in the above testing, with the exception that a Panasonic CD-ROM drive, operating in PIO Mode 4, was added to the same cable as the Maxtor hard drive. The hard drive was acting as Master and the CD-ROM drive as Slave. The system was configured to run in Ultra ATA/66 mode identically as in the above Ultra ATA/66 tests, including use of an 80-conductor IDE cable. Again, the only difference is the inclusion of the CD-ROM drive on the IDE cable.
Since the results are fairly consistent among all of the operating systems, we have trimmed this next comparison down to just the Windows 98 SE and Windows 2000 NTFS results. We did run the test under all 4 configurations as usual, but felt it redundant to include listings under all of the operating systems, as it is basically the same data being repeated.
As you can see from the above results, with the exception of ZD Content Creation, the addition of a CD-ROM drive to the IDE cable seems to have little to no effect on performance under Windows 98.
The same results as under Windows 98, not much of an impact at all is noticed by sharing the IDE channel with a non-Ultra ATA/66 device.
Final Words
While the above tests are not meant, nor do they, to represent all applications one would use, and results will vary depending upon the application you are running, we do feel they give a good indication of the general performance you can expect by utilizing the Ultra ATA/66 interface over the Ultra ATA/33 interface. You can also see that you will not realize too big of a hit, if any, on performance by placing a Ultra ATA/66 hard drive and non Ultra ATA/66 device on the same IDE channel.
In general, if you are still using an Ultra ATA/33 system, with the current generation of drives there is not really an incentive yet to switch over to the new interface. We do expect the forthcoming drives to start utilizing the increased bandwidth the Ultra ATA/66 interface affords and defiantly see the need for its implementation. However, as you can see from our testing, there is currently little performance benefit using one interface standard over the other.
As mentioned previously, the Ultra ATA/66 specification should be viewed more as a way of allowing room for growth in hard drive transfer bandwidth, than it is for a current performance booster. To put it in simpler terms, it’s akin to adding 4 more lanes to a highway that currently is able to handle 5000 cars per hour. If that highway only has a current peak usage of 5000 cars per hour, and all the cars are able to maintain a 60 MPH speed, then the addition of those extra 4 lanes will not be immediately noticeable. But in 5 years time, when the traffic has doubled to 10,000 cars per hour, the average speed will be able to be maintained at 60 MPH instead of reducing the traffic flow to a crawl had the changes not been previously made.
Although there are other factors in traffic flow patterns besides the number of lanes, and hard drive interfaces are not highways, we hope that you can see the point in the analogy. While the Ultra ATA/66 specification is definitely needed, and in some limited situations, you can currently see a slight benefit, it is definitely more of a technology that allows room for growth in the future than it is for offering enormous performance benefits today.
Facebook
Twitter
RSS
