Singapore Hardware Zone

Promise Ultra 66 IDE Controller Card
Reviewed by CPU-ZILLA  (25 Aug 99)

Ultra66 Controller Card Specifications

Features

  • 66MB/sec burst transfer rates with Ultra ATA/66 drives
  • CRC checking with Ultra ATA hard drives
  • Dual independent channel PCI IDE bus master controller
  • PCI "Plug & Play" compatibility
  • Coexists with IDE and SCSI controllers
  • Onboard BIOS with LBA translation and Extended Int 13, supporting IDE hard drives up to 128GB
  • Flashable BIOS
Operating Systems Compatibility
  • DOS 5.x and above, Windows 3.x, Windows 95/98, Windows NT 3.x & 4.x
Drive Modes
  • Ultra DMA Mode 1/2/3/4, DMA Mode 0/1/2, PIO Mode 0/1/2/3/4

Physical Dimension

  • 4.75" x 3" (12.07cm x 7.62cm)

<Introduction><The Good><The Test> <The Bad><Conclusion><Rating>

Back to top


Introduction

When you're not feeling well, what do you do? Go to the clinic and consult a doctor, and that's exactly what I'm doing right now, waiting in the long queue for my turn to see the physician. It is extremely boring to sit here staring at everyone, so I took out my Palmtop PC and started typing this article. Anyway, when your PC has problems, you would need to seek a "PC doctor" too. Sometimes, it is strange to see some sending their PC to the "clinic" for upgrading purposes, and I do know of some who constantly upgrade their PC. Well, if you have the money, and willing to spend for a slight improvement in performance, it should be fine. However, I feel that upgrading has to do with a lot of vanity and pride rather than for any useful purposes.

I do have a product that would suit everyone in fulfilling their vanities. Yes, it is the Promise Ultra 66 IDE controller. Although it is not very much of an upgrade, but I'm sure it does feel nice to have one slotted into their PCI slot. Keeping up with today's hardware technology is almost as much as keeping up with today's software updates. People tend to move to the best and latest, sometimes without knowing the exact benefits. I know of friends who keep so up to date with their software updates, even when a minor update is released (probably 0.0.0.1 version higher), they'd be proud and satisfied to have it. Little did they know the patch was only a fix for certain conflict in some obsolete hardware. Don't get me wrong, I'm not saying that adapting to the latest technology is a bad thing, it's just that sometimes it is costly and not worth the investment. Anyway, I suppose a lot of companies do appreciate people like these as they do make up a significant percentage of their customers. On the other hand, wise upgraders would find out the exact benefits before splurging their money on the product. Therefore, make sure you know exactly what you want and how the new product could suit your environment and needs before investing in it. Now, on to the review...

Promise Technology is a well known company for producing controller cards for both IDE and SCSI interfaces, including IDE raid controllers. They have over 10 years of experience in this field, so it is not surprising to see them to be one of the first to ship the UDMA/66 IDE controllers.

We have heard enough of reviews about how little UDMA/66 technology can do for us. Well, this is yet another look into this technology. For those unfamiliar, UDMA/66 is the same as Ultra ATA/66.

The product comes with a detailed user's manual, an 80-conductor 40-pin UDMA/66 cable, a diskette containing drivers for Windows 95/98/NT and the card itself. Let's take a look at how this card performs, shall we?

Back to top


The Good

First of all, the card looks very simple in its build and construction. Installation was easy and quick. The concise and detailed manual explains all the necessary steps to follow in order to install the product. The whole package comes with an ATA/66 cable that's constructed with an 80-conductor cable, but still maintaining the 40-pin IDE interface. The wires are much finer, and somewhat resembles an Ultra SCSI cable. Also, it is stiffer than your usual IDE cables. Anyway, it's good that they have provided a free ATA/66 cable since it could cost a pretty hefty sum should you have to buy it off the shelf. Furthermore, such cables are uncommon and rare. I made a quick check around to see if I could buy the connectors for a home-made ATA/66 cable, but sad to say, they were nowhere to be found.

The card also comes with its own BIOS to allow you to boot from your ATA/66 drive(s). So, if you have one of those speed demons around, you should connect it to the card and use it as a boot device. On the other hand, it is compatible with other non-ATA/66 drives and this card could serve as an additional IDE controller, thus, allowing you to add another 4 IDE devices. You should know by now that IDE drives are still a lot cheaper than drives using the SCSI technology. Hence, this card is quite an inexpensive upgrade as compared to its SCSI counterparts.

Most of the newer hard drives you can find in the market today are actually ATA/66 compatible, although it is still compatible with other modes of transfer. These drives are normally shipped pre-configured for the ATA/66 interface. However, if your motherboard controller does not support this feature, you should disable it yourself. This can be done by using available utilities from the respective hard disk manufacturers. However, if you do not disable this feature, your motherboard will fall back on a slower mode of transfer, and most of the time, it falls back to using PIO Mode 4. On the other hand, with the Promise Ultra66 controller card, you would not need to do anything at all. The card automatically detects whether it is an Ultra/33 or Ultra/66 drive. Furthermore, if you accidentally use a 40-conductor cable, and it detects the presence of an Ultra/66 drive, it informs you to change to an 80-conductor cable during the BIOS initialization.

Another neat feature is the inclusion of BIOS in the card. This allow users the choice to boot their IDE devices using the Ultra66 controller card. What's more, the BIOS is also flashable. This means better product support since some bugs can be ironed out with updated BIOS.

Back to top


The Test

In the tests, I used the Gigabyte dual processor BX motherboard due to its built-in Ultra-Wide SCSI controller. Together with the Adaptec SCSI controller, I've used the IBM Ultrastar 18ES with 2MB cache and rotational speed of 7200RPM as a comparison. On the Promise Ultra66, I've used the IBM Deskstar 22GXP, also with 2MB cache and rotational speed of 7200RPM. The card was tested using both FAT32 (Windows 98) and NTFS (Windows NT) partitions. In addition, only one processor was used throughout the tests, which is an Intel Celeron 466MHz with 128K L2 cache.

Test Configuration

Motherboard(s): Gigabyte 6BXDS w/ built-in UW-SCSI
ECS P6BXT-A+ Rev. 1.2c
Processor(s): Intel Celeron 466MHz, 128K L2 cache
RAM: 1 x 128MB Spectek PC100 ECC SDRAM
Hard Drive(s):

IBM Deskstar 22GXP DJNA-370910 (7200RPM)
IBM Ultrastar 18ES DNES-318350 (7200RPM)

Video Card(s): Creative Labs Riva TNT 16MB
Bus Master Drivers: Windows 98 Bus Mastering Drivers
Video Drivers: NVidia Reference Detonator Drivers 1.88
Operation System(s): Windows 98 Second Edition (build 4.10.2222A)
Windows NT 4.0 Server (Service Pack 5)

Ziff-Davis Benchmark Results (Windows 98 FAT32)

Motherboard/
HD Interface
Business
Disk Winmark99
High-End
Disk Winmark99
Business
Winstone 99

GA-6BXDS /
UW-SCSI

3730 12500 21.6
GA-6BXDS /
on-board UDMA/33
3650 12200 21.9
GA-6BXDS /
Promise (UDMA/33)
3700 12400 22.1
GA-6BXDS /
Promise (UDMA/66)
4170 12700 22.1
ECS-P6BXT-A+ /
on-board UDMA/33
3710 12500 -
ECS-P6BXT-A+ /
Promise (UDMA/33)
3760 12900 -
ECS-P6BXT-A+ /
Promise (UDMA/66)
3750 12900 -

* Note : the on-board UDMA/33 HD Interface means that benchmarks were performed using the built-in motherboard IDE controller. Where the word Promise was mentioned, the Promise Ultra66 controller was used instead.

Comparing the results of the benchmarks, it is certainly no surprise to see the little impact made by the Promise Ultra66. However, we do see slight improvements in all the benchmarks, and it is clear that the Ultra66 card is slightly faster than the normal on-board IDE controller found on all motherboards. I guess the hard disks found in the market nowadays are not as fast as they seem, and it will be some time before hard disks are fast enough to take the full advantage of the ATA/66 technology.

I wasn't satisfied with the results. So, I picked up the Adaptec Threadmark 2.0 benchmark and ran it on the same configurations. I must say I'm pleasantly surprised with the results.

Adaptec Threadmark 2.0 (Windows 98 FAT32)

Motherboard/
HD Interface
Data Transfer
Rate (MB/s)
Average CPU
Utilisation (%)

GA-6BXDS /
UW-SCSI

9.05
32.76
GA-6BXDS /
on-board UDMA/33
4.63
90.86
GA-6BXDS /
Promise (UDMA/33)
8.74
35.88
GA-6BXDS /
Promise (UDMA/66)
8.64
33.31
ECS-P6BXT-A+ /
on-board UDMA/33
4.68
90.34
ECS-P6BXT-A+ /
Promise (UDMA/33)
8.88
34.55
ECS-P6BXT-A+ /
Promise (UDMA/66)
8.85
37.84

At least the picture is clearer now. We can see that the Promise Ultra66 controller is indeed delivering its promise. However, the data transfer rate between UDMA/33 and UDMA/66 mode was still very similar. Although it did not manage to beat the UW-SCSI controller in terms of performance, it did however show that data transfer rates have improved as compared to the normal on-board UDMA/33 controller. Furthermore, it cuts down the average CPU utilisation by more than 50%!! This is a really significant number when you actually consider using your system as a server or for multimedia purposes. Anyway, please note that since both the hard disks are different in size and make, do not take the benchmark scores between the SCSI and IDE controller literally. I've decided to include it as a form of reference only.

You may ask, how about Windows NT 4.0? Does the card work in Windows NT 4.0? Well, I've seen these questions in our PC Clinic before, and I'm glad to say that it is supported in Windows NT 4.0. However, I've yet to see drivers for Windows 2000. Anyway, here are some results performed under Windows NT 4.0 (with Service Pack 5) using an NTFS partition.

Ziff-Davis Benchmark Results (Windows NT 4.0 NTFS)

Motherboard/
HD Interface
Business
Disk Winmark99
High-End
Disk Winmark99
ECS-P6BXT-A+ /
on-board controller (DMA off)
3380 6270
ECS-P6BXT-A+ /
on-board controller (DMA on)
4240 10800
ECS-P6BXT-A+ /
Promise (UDMA/33)
4420 10500
ECS-P6BXT-A+ /
Promise (UDMA/66)
4500 11200

Adaptec Threadmark 2.0 (Windows NT 4.0 NTFS)

Motherboard/
HD Interface
Data Transfer
Rate (MB/s)
Average CPU
Utilisation (%)
ECS-P6BXT-A+ /
on-board controller (DMA off)
3.02 94.64
ECS-P6BXT-A+ /
on-board controller (DMA on)
8.84 11.30
ECS-P6BXT-A+ /
Promise (UDMA/33)
8.87 13.68
ECS-P6BXT-A+ /
Promise (UDMA/66)
7.78 10.50

Well, the scores do look pretty interesting now. The Promise Ultra66 card does seem to perform faster in Windows NT as shown in the Winmark99 scores. However, the performance was not that outstanding. In the Adaptec Threadmark tests, I was surprised to find the score lower when UDMA/66 was enabled. I don't really understand this. I ran the benchmarks repeatedly and it was nowhere near the UDMA/33 performance. This is puzzling and I can't say much here. On the other hand, the average CPU utilization was even lower in Windows NT, almost 9 times lower. Even the generic IDE driver in Windows NT performed well when the DMA transfer mode was turned on. This shows how inefficient is the Windows 98 bus mastering drivers.

A lot of you must be wondering why I ran the test when on-board controller DMA was turned off. I believe a lot of you still do not know that Windows NT turns off the DMA transfer for the on-board IDE controller by default. In addition, there's no way to turn it on, unlike in Windows 98 where you can turn it on under the system device properties with just a click of a button. So, how do you turn it on? Well, Microsoft did ship a small utility in their Service Pack 3 CD, and it is called DMACHCKI.EXE (for x86 systems). What this utility does is modify the Windows registry so that it enables DMA transfers when Windows boot up. However, Microsoft does not guarantee this to work, and if you have a non-UDMA device attached, you could very well end up with the blue screen of death. You can find this utility if you do a search over at Microsoft's web site. Or, if you are adventurous enough, you can add this into your Windows NT registry. Note that you need administrative rights, and whatever happens to your system, it is entirely your fault, not mine. :)

KEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\atapi\Parameters\Device 0 Driver Parameter "DMADetectionLevel=0x1;"

If you have devices connected to the secondary IDE channel, you will need to add this too:

KEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\atapi\Parameters\Device 1 Driver Parameter "DMADetectionLevel=0x1;"

Set the DMADetectionLevel to :-

0x0 if you want DMA disabled.
0x1 if you want DMA enabled.
0x2 if you want to force DMA on.

Please note that the above registry tweaks do NOT apply to the Promise Ultra66 controller as it uses its own bus mastering driver. The above tweak is for NT users who want enable DMA detection in their on-board IDE controller.

Hmm... how did I manage to deviate so far?

Back to top


The Bad

The main let down is that it fails to deliver the performance that everyone expects from it. I guess I should not blame the card but the technology itself since there are no IDE drives fast enough in the market to take full advantage of the UDMA/66 technology. Perhaps in the future, when 10,000RPM drives emerge, this would have been the card to get. Still, I believe it is needed to help drive the market towards that direction.

Another thing about the card is the slow BIOS initialization. After the motherboard's Power On Self Test (POST) sequence, the Ultra66 card initializes and loads the BIOS. It will then detect for devices connected to the card. This adds to the boot-up time, but then again, this is almost like any other SCSI card. In addition, there's no way you can configure the BIOS like some SCSI cards. It would be cool if you could configure the drives in its BIOS, thereby shortening the BIOS initialization time by skipping the detection every time you restart your system.

Back to top


Conclusion

Although the card did not perform as well as we thought it would, it still didn't fail to impress me with its shorter boot time into Windows 98/NT. Moreover, the CPU utilization was well reduced and its performance was quite close to that of an Ultra Wide SCSI controller. Overall, if you have the extra buck to spend, I'd say this is an upgrade worth considering. Alternatively, if you want faster access to your small office server or you are into video editing, this is a must if you intend to cut costs on expensive SCSI cards and drives.

Back to top



MOTHERBOARD RATING

Overall Rating (Out of a maximum of 5 Star)

Installation *****
Performance ***
Price ****
Material Quality ****
Stability *****
Overall Rating ****

Back to top


This Product is provided courtesy of

CORBELL TECHNOLOGY PTE. LTD.

Copyright © 1999 by Singapore Hardware Zone. All rights reserved.

None of the above shall be reproduced, copied and/or
modified without the permission of the WebMaster.