Well, it looks as if ATA/133 interface were little by little becoming a "de facto" standard for hard disk drives, even despite some displeased HDD manufacturers. Maxtor was the first one to reach the physical limitations imposed by ATA standard over the maximum HDD storage capacity. Therefore, no wonder that this company appeared the one to offer a solution to the problem, i.e. how this limitation could be overcome. It is evident that all the other hard disk drive manufacturers will have to follow Maxtor, because moving to SerialATA will be much harder for users from the psychological point of view than moving from ATA/100 to ATA/133. Moreover, the chipset supporting Serial ATA will be more expensive (as they will be fresh new) than the existing ones with the ATA/133 support added.

Although there are only two hard disk drives now supporting ATA/133 (both are made by Maxtor) - Maxtor D740X-6L and Maxtor D540X-4G - the manufacturers of external UDMA-controllers and chipset makers have already prepared for meeting them. You may see a lot of products supporting this new protocol selling openly. Among them are: HighPoint HPT372 based controllers and mainboards with integrated RAID chips, Promise Ultra133 and ALi MAGiK 1 based mainboards with M1535D+ South Bridge. In other words, it's high time we paid attention to ATA/133 once again!

Testing Participants

First of all we were very much interested in taking a closer look at Promise Ultra133 controller, because first, this is an external PCI32 card, which can help to improve the current PC configuration (like many other users we still have some i440BX based platforms working here), and second, this is another UDMA-controller made by a very respectful company :)

We are now working on our new test platform for the year 2002 that is why choosing the proper ATA/133 controller is a matter of life and death for us…

We also got hold of HighPoint HPT372 controller integrated into Iwill XP333-R mainboard based on ALi MAGiK 1 chipset (see our Iwill XP333-R Mainboard Review). The results of our tests were used in our Maxtor D740X-6L HDD Review.

The third testing participant is ASUS A7A266-E mainboard on ALi MAGiK 1 chipset. The peculiar thing about this mainboard is the new ALi M1535D+ South Bridge supporting ATA/133 protocol.

Testbed and Methods

• ASUS A7A266-E and Iwill XP333-R mainboards;
• AMD Athlon XP1500+ CPU;
• 2 x 128MB PC2100 DDR SDRAM by Nanya;
• IBM DTLA 307015 system HDD;
• Matrox Millennium 4MB graphics card;
• Windows XP.

To make sure that all the testing participants run in equal conditions we tested Promise Ultra133 controller in a PCI slot of ASUS A7A266-E mainboard. This way all the three controllers were tested on one and the same chipset.

We used the following drivers for our controllers:

• Promise Ultra133 - v.200 build 029
• HighPoint 372 - v2.0.0919
• ALi M1535D+ - v. 1.06

Here are the benchmarks used:

• WinBench99 1.2;
• HDTach 2.62.

As you have already noticed, we didn't include the Intel IOMeter test into the testing set, as it is not "controller-dependent", i.e. the results in this benchmarks (at least in the common patterns) depend neither on the controller type, nor on the UDMA protocol.

For WinBench we used FAT32 and NTFS file systems to format each of the HDDs as one logical drive of the maximum size with the default cluster. All the tests were run 4 times and then the average results were taken for the diagrams.

Performance

HDTach 2.61

Judging by these results, the difference between the controllers turned up only when we checked the read speed from the controller buffer. Let's have a look at the diagram to get a better chance to compare the read speeds from the controller buffer and the average read and write speeds.

As we see, although the read speeds differ by different controllers, it didn't influence the average read and write speeds at all (the best write speed belongs to HPT372).

But, just in case we would like to describe the situation. The indisputable leader is HighPoint HPT372 controller, then goes Promise Ultra133 and the last one in this list is ALi's South Bridge. We would also like to remind you that all the three controllers were tested in equal conditions, as ALi's South Bridge is connected with the North Bridge via the PCI bus.

WinBench99 1.2 FAT32

Well, let's see how well the controller drivers are optimized:

DTR: Beginning

Graph

Graph

Graph

The first thing that catches our eye is the low performance of the ALi South Bridge in Business test. It seems that ALi's software developers were in the first place worried about the stability and compatibility of their solution rather than about it performance. However, it is the stability and compatibility that we usually expect from all the chipsets…

It is a curious thing that the new ALi MAGiK 1 South Bridge yielded quite a lot in Business test, and then caught up with the leaders in High-End tests. And the leaders, Promise and HighPoint controllers, ran nearly evenly, which we have actually seen before.

Now let's pass over to NTFS file system.

WinBench99 1.2 NTFS

Here we again see the ALi chipset falling behind the external controllers by Promise and HighPoint quite significantly. In Business tests its performance is 1.5 times lower than that of the competitors! It's not even funny any more… :(

By the way, in NTFS ALi M1535D+ lags behind the rivals not only in Business benchmarks. If it yielded about 2% in High-End tests in FAT32 file system, then in NTFS this lag made around 6%-7%. Not a tragedy, of course, but anyway, pretty upsetting…

Promise and HighPoint controllers continued the race neck and neck: promise won the laurels in Business benchmark, and HighPoint - in High-End benchmark.

Conclusion

Well, our second trip to the world of ATA/133 seems to have come to an end. We are happy to state that we didn't discover any stability and compatibility problems during our tests.

Judging by the results obtained in this test session as well as in all the previous ones, we can point out that further performance growth under Windows is no longer limited by the controller itself or its drivers (we do not take ALi into account here: its software guys still have some room for improvement). It is small HDD cache and low bandwidth of the bus between the South and the North chipset bridges that are to blame.

So, in the meanwhile ATA/133 protocol doesn't mean any performance gain, but only the opportunity to connect to the PC hard disk drives with over 137GB storage capacity. As we remember, there is only one HDD like that right now: Maxtor D540X-4G (Neptune) with 160GB storage capacity. But there is a new season ahead and it will definitely bring us some new achievements in the per platter data density. That is why those of you who are assembling their PCs right now should better think about such a trifle as high-quality support of large HDDs… SerialATA standard is still quite far away, and ATA/133 HDDs and controller cards are already available in retail stores.