As we have already mentioned several times in our reviews, manufacturing of hard disk drives with Ultra DMA/66(Ultra ATA-66) interface began earlier than the manufacturing of mainboards supporting it. Besides, the storagedevices strive for perfection (i.e. cheapness) in a really impressive tempo. That's why most of us, users, are nowthe happy owners of HDDs with Ultra DMA/66 interface and are also enjoying a mainboard based, shall we say, on BXchipset, which is exactly what we need in all possible respects but unfortunately doesn't support the data transferrate of 66MB/sec. Of course, there is always a way-out and you can change your mainboard but again we would like toremind you of the minimal gain you get to the data transfer rate! Moreover, you will definitely wish you never had tospend this money and besides, the mainboard you have been using for so long has undoubtedly become very dear to yourheart. On the other hand, you will never refuse to make a hundred percent use of your harddisk. And if dreams likethat start occurring to you then you definitely belong to those users for whom ABIT designed its Hot Rod 66 - IDE UltraDMA/66 controller, which is none other but a PCI expansion card. Buy, install and enjoy! :-) Here is the list of thetechnical features announced by the manufacturer:

• Chipset
  1. HPT366 Ultra DMA/66 IDE controller
• PCI bus
• Features
  1. Ultra DMA 66MB/sec data transfer rate
  2. Two independent ATA channels
  3. 256 Byte FIFO per ATA channel
  4. Concurrent PIO and bus master access
  5. Compliant with Plug & Play
  6. Up to 4 disk drives support
• Supported drive modes
  1. Ultra ATA 4/3/2/1/0
  2. PIO 4/3/2/1/0
  3. DMA 2/1/0
• BIOS supports
  1. Disk type autodetection and configuration
  2. Ultra Mode (ATA/EIDE) autodetection and support
  3. Up to 128GB disks detection
• Advanced Data Features:
  1. Supports new CRC enhanced data protection for Ultra ATA drivers
  2. Supports dual data channels, which allow separate device timings for Ultra ATA and EIDE devices
  3. Supports ACPI functions
• Software
  1. Microsoft Dos 5.x and up
  2. Microsoft Windows 95/98
  3. Microsoft Windows NT 4.0
• Approval
  1. CE
  2. FCC

In fact, the device is based on HighPoint HPT366 controller. Besides it, the mainboard is also equipped with aFlash Rom microchip with BIOS, two IDE connectors and a connector for light emitting diode, which signals thatdata throughput is in progress. Got curious? OK, let's open the box.

Installation and First Impressions

The colorful box contains the board, a very beautiful user's manual, two colorful (with multicolor connectors)80-wire cables and an absolutely gray (among such vivid and bright colors) floppy disk decorated with a black-and-whitesticker, which includes the drivers for Windows 95/98 and NT as well as BIOS flash utility.

At first we read the manual very attentively. As usual it provides the whole bunch of necessary info about the cardinstallation and the drivers. We were curious to find out that for instance the blue cable connector should go into HotRod,the black one - into Master device, the gray - into slave. We wish we never forgot to try to make it vice versa! To tell thetruth we considered it to be pretty strange that the booklet finished with the drivers installation, which made us suspectthat manual configuration was practically neglected. But we'll see if there is any real cause for concern.

We install the board, connect all the cables according to the manual's instructions and turn on the PC. In our computerwe have ASUS P2B mainboard, 1008 BIOS revision and Windows 98. Great! The system is happy to inform us that it has detecteda PCI Mass Storage Device and requests the drivers. Well, let's satisfy its demand and install the software. Restarting… Wow,we get some new SCSI-controllers. We feel very inclined to take a look at the new options we get: what can we set there? Andas a bolt from the blue there comes the first disappointment: nothing! Well, let's take a look at the changes, which couldhave occurred in the HDD Settings page. Now we are dealing not with an abstract GENERIC IDE TYPE XX drive but with a concretefirmware version. Seems a trifle but very pleasing one, anyway. And which is actually quite natural, the DMA checkbox has becomeempty. Of course, it is somewhat disappointing that there are no manual disk settings (we understand that everything should bedefined automatically, but you can't imagine how many messages we get from those, who for some reason don't get the properresult...). Instead we get a number of SCSI options, which you'd better not touch.

It is very pleasant to realize that the new controller doesn't replace the integrated one but just adds its contribution.So, you can connect up to 8 IDE devices if the power is enough. We don't know what about eight devices, but five ones workedjust perfectly, we checked it. The NT drivers installation also ran without any problems. In other words, the first impressionis highly positive. The only drawback here is the total absence of manual settings.

Performance

The first thing we would like to point out deals with the system boot. The matter is that the priority belongs tomainboard BIOS, of course. That's why if at least one harddisk is connected to IDE controller of your mainboard, youwill be able to boot from the disk connected to Hot Rod only in case the mainboard BIOS Setup implies that the systemshould boot from the disks connected to SCSI-controller. In fact BIOS doesn't care if it is an Ultra DMA/66 or SCSIcontroller, the only thing that really matters is that it is an external one. Besides, it is also possible to choosethe disk out of those connected to the mainboard, from which the system will boot, as well as the opportunity to bootfrom Zip/LS-120 and CD-ROM (if they are connected to the mainboard controller). These BIOS settings work properly anyway.

It was not for nothing that we complained about the absence of the manual settings for Hot Rod controller! If we connectIDE devices only to this controller, it will still be possible to choose the booting device. Hot Rod will boot from C drive,i.e. from the device connected as a primary master, if it is absent them from primary slave, etc. In this case you can't bootfrom Zip/LS-120 or CD-ROM connected to Hot Rod. In other words, there is not so much space left for maneuvers while booting.The most suitable way for us seems to be the following. DVD-ROMs, CD-ROMs, CD-Writers, Zips and other ATAPI-devices aremostly connected to the mainboard controller, and the hard disk drives - to Hot Rod.

By the way, Ultra DMA/66 controllers are said to work with ATAPI and Ultra DMA/33 devices pretty badly. Besides, wedoubt if the PCI bus is stable enough while working at higher frequencies.

Well, as far as the Ultra DMA/33 devices are concerned, everything is in order here. IBM DTTA-371010 grad disk driveworks just perfectly without any serious problems, and even performs nicely in the benchmarks.

As for the higher working frequencies, the things are also OK. This time we have every right to state this. The matteris that having set the FSB frequency to 83.3MHz (PCI frequency - 41.65MHz) and having run the tests for about 8 hours,we forgot to set the frequency value back to the initial one and kept on testing for another three days at higher frequency:no problems.

Regarding ATAPI-devices (such as CD-ROM, etc.), we had to stop the testing after 8 hours because of the lack of time.On the other hand, Hot Rod controller remained in our testing system even after we finished and Hitachi GD-2500 DVD-ROMdrive is constantly connected to it. We faced no problems during the tests as well as in the further work. Of course, itdoesn't mean that the notion "problems" simply does not exist for this device, but the situation proved not that awful,as it might seem at first sight.

And now a few words about the cables. You all probably know that each Ultra DMA/66 cable has 80 wires and are 18 incheslong at the most.

Unfortunately, you have to use these particular cables. If you try to connect an Ultra DMA/66 device to Hot Rod with anordinary cable, it will be recognized as Ultra DMA/33 device. And as for the complicated connection scheme, it is notnecessary to follow the connector colors. We checked it: everything works without any problems (which is not surprising,actually).

And at last a few words for the DOS lovers. You don't need any special drivers to work in DOS. And you really get UltraDMA/66 (maximum read rate from the cache as to HDDSpeed 2.1 - 60.1MB/sec). But the whole thing works quite strangely, tosay the least of it. HDDSpeed we have already mentioned honestly reports that a 504MB device is something unknown (notIDE and not SCSI) and its size doesn't depend on the disk type connected to the controller. And Norton Commander (DOSversion) 5.0 shows the disk size about 2GB and announces the whole disk to be free. However, at the same time thisCommander shows all the info saved to the disk and successfully copies it from/onto the disk. So, you may work inconditions like that, but it seems a bit uneasy for us, for instance.

And now let's pass over to the most interesting thing. You are probably wondering was this matter really worth it.Well, why don't you look at the tests then?

Benchmarks

The testing system was configured as follows:

• ASUS P2B mainboard (Intel 440BX, BIOS version 1008);
• Intel Pentium II 400MHz CPU;
• 128MB SDRAM SEC PC100;
• IDE Ultra DMA66 Abit HotRod controller;
• ASUS V2740 (Intel 740) graphics card;
• Windows 98.

For our benchmarks we used two Quantum HDDs with Ultra DMA/66 interface - Quantum Fireball Plus KA 9.1 with 7200rpmrotation speed and Quantum Fireball CX 20.4 with 5400rpm. By the way, we haven't yet tried the CX family that's why we willalso take a closer look at this harddisk in comparison to its "group-mates". In order to check the performance of HotRod 66 controller with Ultra DMA/33 disks we resorted to IBM DTTA-371010 (7200rpm).

The contents of all the disks tested was absolutely identical. We cloned it with Norton Ghost utility. And as for thetests, the set was rather standard: WinBench 99 (Business WinMark and Disk Inspection Tests) and Adaptec ThreadMark. DiskCPU Utilization benchmark may be carried out in two different regimes: with the fixed read rate and with the maximallyachievable one. All the benchmarks were run with the speed fixed at 4000000Byte/sec and repeated at least 5 times (if thedifference between the results achieved each time didn't exceed 3%). So, the tables contain only average values. The WindowsDMA support was enabled. The benchmarks were undertaken only for "cold" disk drives, i.e. we let the harddisk cool downtill the room temperature before starting the next test. The system was reinstalled every time before running a new benchmark.

So, at first WinBench 99. The results achieved are given in the table below. The numbers in brackets denote the CPUutilization for each particular benchmark. In the column "Ultra DMA/33" are given the benchmarks results when the drive isconnected to mainboard controller, and in the column "Ultra DMA/66", when it is connected to Hot Rod.

Frankly speaking, there isn't much to comment on, because we got what we actually expected to get: a slight performanceincrease following a slight increase of the processor utilization. Again, the results are very illustrative. They show thatthe applications, which have to process large files (sound editor - Sound Forge 4.0, graphics editor - Adobe Premiere 4.2,well-known Adobe Photoshop 4.0), can hardly boast any significant gain (and the loss you have probably noticed by FireballCX in Sound Forge is the result of averaging the values). And as for the applications working with smaller files, theyappear to have greater gains. In this respect C++ compiler is situated somewhere in-between. And which was really pleasant,it's the fact that if the device does not support Ultra DMA/66, Hot Rod 66 works at least as it usually should and not alittle worse.

Data transfer rate graphs are given only for Quantum Fireball CX, because, as we have already warned you, we haven'tyet tested this drives family. You may easily notice that the two graphs given hardly differ, i.e. the platter read rate isstill quite within Ultra DMA/33.

Adaptec ThreadMark. As usual, we offer you the average data transfer rate in MB/sec according to 64 tests carried out. Thesetests include read and write operations for blocks of 512Bytes, 1, 2, 4, 8, 16, 32 and 64KB. These operations were carried outthrough one, two, three or four threads simultaneously calling Win32 API in a standard way. And the second row shows CPUutilization during the benchmarks.

To tell the truth, the situation appears rather paradoxical: the higher is the data transfer rate, the slower is thetransfer itself. In fact, there is nothing to be surprised at. The thing is that in order to win relative independencefrom the operation system cache, the test uses very long files (doubled by the system memory size - in our case 256MB).So, Ultra DMA/66 brings almost no gain to the general performance. And the losses can be explained by the fact that theintegrated controller works much better with the multithread input/output. Unfortunately, we are unable to carry out amore detailed analysis because this test can't be regarded as highly informative.

And in conclusion, we would like to say a bit about the features of Quantum Fireball CX compared to its "group-mates",such as IBM DJNA and WD Caviar. Well, the situation was very typical for Quantum: the best platter read rate, a slightfalling behind in Business Disk WinMark 99 and approximately equal performance in High-End Disk WinMark 99 (due to thesmaller buffer). Its leadership in Adaptec ThreadMark is just indisputable (due to the read rate). All in all, it appeareda really worthy representative of Fireball family. If you are willing to compare the numbers on your own, please, here youare.

Conclusions

Well, nothing unexpected happened. The device really works and really provides up to 66MB/sec data transfers.Quite another thing is that this controller seems to be getting ahead of the time. However, we would like to drawyour attention to the fact that the data density and hence the platter read rate as well as the cache-buffer sizehave been increasing pretty fast recently. That's why we may expect that in the near future Ultra DMA/66 featureswill be very widely used everywhere. And in the meanwhile you have to decide if this gain is worth the money youshould pay for it. So, highs and lows:

Highs

• The performance increase for the devices supporting Ultra DMA/66
• "Moral upgrade" of your PC (getting ready to use the future storage devices) without changing the mainboard
• Up to 8 IDE devices support
• Reversed compatibility with any IDE devices

Lows

• small performance gain
• the restrictions dealing with the booting devices
• the money you anyway have to pay :-)