Author: JT Smith
Blast from the Past
Over the last six months, there has been a plethora of chipset releases from the various chipset manufacturers, such as Intel and VIA. These chipsets have all pushed the bleeding edge, adding more USB ports, 4X AGP, RDRAM, 133MHz FSB support, and all sorts of other new features. And yet, today, I find myself reviewing a board based on the “old reliable” BX chipset, which, while not sporting all of those new features, is still quite a contender.
The Board
BX133 Product Page
The ABIT BX133 is the latest in a long line of BX boards from ABIT. Although the BX chipset only officially supports up to 100MHz front side bus speeds, the BX133, as the name implies, is tuned to be able to run at up to 133MHz, provided you have a video board which can handle the increased AGP bus speed. The 3DFX Voodoo3 used in these tests worked flawlessly. Thanks to the 133MHz front side bus support, the BX133 is fully capable of supporting a 133MHz-bus CPU such as the Pentium III 933 used in this review.
For internal expansion, the board can support up to five PCI cards, one ISA card, and one AGP card, as well as up to eight IDE devices (via two UDMA/100 channels and two UDMA/33 channels). It has three DIMM slots capable of supporting PC100 or PC133 SDRAM. Externally it has two USB ports, as well as two serial and one parallel port. The ATX-standard PS/2 mouse and keyboard connectors are included as well.
For UDMA/100 and Raid 0+1 (striping + mirroring) support, the board uses the High Point HTP 370 IDE Controller. This allows for RAID mirroring and four extra IDE devices, as well as allowing UDMA/100 drives to be used to their full potential. However, I had no UDMA/100 drives at the time of the writing of this column to test it with. Support for this controller under Linux is available and was installed by default as part of the Mandrake 7.1 installation used.
The BIOS on this board is an Award BIOS. The BIOS on the BX133 includes ABIT’s SoftMenu III CPU control system, allowing you to fully control and adjust the settings for your CPU and chipset, including voltage, AGP multiplier, and other features, making it an excellent board for overclockers.
Specifications:
CPU
1. Supports Intel Pentium® III Socket based processor
2. Supports Intel Celeron® Socket based processor
3. Support for future Intel Pentium® III/Celeron processor
Chipset
1. Intel®440BX chipset (82443BX and 82371EB)
2. Supports Advanced Configuration and Power Management Interface (ACPI)
3. Supports AGP 1X/2X (Sideband)3.3V device
Ultra DMA 100
1. High Point HTP370 IDE Controller
2. Ultra DMA 100MB/Sec data transfer rate
3. RAID 0(stripping mode for boosting performance)
4. RAID 1 (mirroring mode for data security)
5. RAID 0 +1(stripping and mirroring)
Memory
1. Three 168-pin DIMM sockets support SDRAM module
2. Supports up to 768MB Max. (8, 16, 32, 64, 128, 256MB SDRAM)
3. Supports ECC
System BIOS
1. SoftMenuTMIII Patented Technology and DIP switches to set CPU parameters
2. Award Plug and Play BIOS
3. Write-Protect Anti-Virus function by AWARD BIOS
4. Year 2000 Compliant
Multi I/O Functions
1. Two Channels of Bus Master IDE Ports supporting up to four Ultra DMA 33/66/100(up to 4 HDD devices)
2. Two Channels of Bus Master IDE Ports supporting up to four Ultra DMA 33 devices(up to 4 HDD devices)
3. PS/2 Keyboard and PS/2 Mouse Connectors
4. 1 Floppy Port (up to 2.88MB)
5. 1 Parallel Port (EPP/ECP)
6. 2 Serial Ports
7. 2 USB Connectors
Miscellaneous
1. 1 AGP slot, 5 PCI slots and 1 ISA slot
2. Hardware monitoring – Including Fan speed, Voltages,System environment temperature
3. Keyboard/Mouse/Password and Hot Key multifunction Power On.
4. Built-in Wake on LAN/Wake on Ring header
5. Built-in IrDA TX/RX header
Board Design
The BX133 is a very well designed board. The IDE and floppy ports are placed on the right side of the board, so that they will not interfere with anything else on the board. The same is true of the ATX power connector, places at the top of the board, keeping it out of the way of the CPU fan. As for the CPU socket itself, the capacitors that surround the CPU are kept far enough away that even a large heat sink such as the one I used in these tests installs and removes quite easily. All expansion slots can accept full length cards, a must for a high-end PC.
Documentation
The documentation included with the BX133 is excellent. It contains full instructions on installing components as well as a detailed BIOS guide, giving you instructions on every feature of the BIOS. The documentation is well written and will be extremely helpful to those wanting to take full advantage of ABIT’s SoftMenu III technology. It even includes instructions for the usage of the HPT370 IDE/RAID controller. Many manufacturers have decided to go with CD-ROM or even online documentation. It is nice to see that ABIT chose quality over cost by including full printed documentation.
The Tests
To test the BX133, I used a Pentium III 933MHz processor. For comparison to a board based on a newer chipset, I included the results from the Soyo SY-7ISM, based on the Intel i815 chipset.
System Configuration
Pentium III 933EB MHz CPU provided by Intel.
128 Megs PC133 Memory
Western Digital 7200RPM 10.2 gig IDE Hard Drive
Sound Blaster Live! Value
3Com 3C905TX-C PCI 10/100 NIC
300 Watt AMD-Approved ATX Power Supply
The above components were purchased from Specialty Tech.
3DFX Voodoo3 3000 16Meg AGP
Operating System: Mandrake 7.1 with 2.2.15 Kernel.
Kernel Compiles
To test the motherboards, I ran a variety of CPU, I/O and memory intensive benchmarks. The first benchmark is a set of timed Linux kernel compiles. Compiling a kernel is a common action for a Linux user, making it a very valid benchmark for a Linux system. To do the testing, I used a Mandrake 7.1 installation. I configured the kernels by typing “make config” and selecting the defaults (holding down enter works nicely). I type “make clean; make dep; time make bzImage” in order to time the kernel compiles.
| 2.2.16 | 2.4.0-test5 | |
| 933MHz | 2:09 | 3:15 |
| 980MHz | 2:02 | 3:02 |
| 2.2.16 | 2.4.0-test5 | |
| 933MHz | 2:23 | 3:40 |
| 980MHz | 2:18 | 3:30 |
| 1015MHz | 2:13 | 3:21 |
As you can see, the ABIT results, while strong, could not beat the SY-7ISM.. However, for a chipset that is two years old, these are very respectable numbers.
Distributed.Net
Next, I used the Distributed.Net client to perform benchmarks using “dnetc -benchmark.” This benchmark uses highly tuned, CPU specific computation cores to achieve the best results possible. Because the core is small enough to run in cache, this test gives you numbers specific to the raw performance of processors of the same type. So, while not fair for cross-platform comparison, it is valid when comparing similar CPUs, or in this case, the same CPU at different speeds, on different motherboards.
| RC5 kkeys/sec | OGR nodes/sec | |
| 933MHz | 2,624,660.80 | 4,920,687.34 |
| 980MHz | 2,766,203.19 | 5,186,969.09 |
| RC5 kkeys/sec | OGR nodes/sec | |
| 933MHz | 2,632,697.12 | 4,937,961.83 |
| 980MHz | 2,776,423.15 | 5,203,178.37 |
| 1015MHz | 2,881,804.87 | 5,403,375.60 |
Here we find the the BX133 beats out the SY-7ISM by a slight margin, perhaps because Distributed.Net’s clients are less memory-intensive than a kernel compile, and once you remove the I/O performance increase of the i815 chipset, the boards are on more even ground.
HDParm
Using HDParm measures the performance (cached and un-cached) of hard drives under Linux. This shows the potential performance of a hard drive under this particular motherboard. Comparison with the SY-7ISM is again included.
| Device Read Timings (-t) | Cache Read Timings (-T) | |
| 933MHz | 4.21MB/sec. | 142.22MB/sec. |
| 980MHz | 4.39MB/sec. | 150.59MB/sec. |
| Device Read Timings (-t) | Cache Read Timings (-T) | |
| 933MHz | 4.82MB/sec. | 76.19MB/sec. |
| 980MHz | 4.87MB/sec. | 77.58MB/sec. |
| 1015MHz | 5.00MB/sec. | 81.01MB/sec. |
In this case, we find that while the BX133 wins the uncached speed, the SY-7ISM wins the cached speed. This does not, however, reflect normal usage. Cached speed is only relevent to tasks such as file servers or imaging workstations where the same data is frequently reused. This performance gap is caused by the UDMA/66 controller present on the SY-7ISM, which roughly doubles the maximum bandiwdth of the UDMA/33 controller on the BX133. Interestingly, results taken with the HPT370 controller seemed inconsistent with this, scoring lower than the ABIT’s native UDMA/33 controller. As such, I believe it was a configuration problem and will update this review when I find out more information.
Overclocking
For an overclocker, the BX133 is a top of the line board. With control over voltage, AGP multipliers, bus speed, AGP transfer mode, cache latency and other features essential to succesful overclocking. These features are what made it possible to get this board up to 1015MHz, since getting it to that high of a speed required a boost from 1.65v to 1.70v to provide stable results. The bus speeds supported by this board are 66, 75, 83 and 84 ~ 200MHz in 1 MHz increments.
About ABIT
ABIT’s Website
ABIT Corporation was established in 1989 as manufacturer of computer mainboards. Since then, it has been a leader in the industry, producing some of the most popular boards to date – Their BX6 and BH6 line of boards were the most widely praised BX boards, thanks to their jumperless operation and their ability to control the voltage going to the CPU, something that can be a huge bonus to overclockers.
Conclusions
The BX133 is a fine board, made by a company with an excellent reputation for expandability and performance. Though it is bested by newer i815 based boards in I/O performance, it is still an excellent board, and the RAID ability makes it a definate winner. I highly recommend this board to anyone looking for a simple RAID 0/1 solution (such as for entry level workstations), or for someone looking for a board with a high capacity for drives.
Full Disclosure: CPU provided by Intel, SY-7ISM provided by Soyo, BX133 provided by ABIT.
Category:
Author: JT Smith
Tending a trade show booth eight hours a day for three days is a wearing task. It’s fun the first few times you do it, but after that it’s simply tiring. Many of the people at this show were veterans of computer shows in general and Linux shows in particular. After a while you recognize faces, and I saw many familiar ones here.
A growing number of companies apparently believe it is good policy to regularly send sysadmins or programmers to shows like LinuxWorld to attend sessions with titles like An Introduction to Writing Linux Device Driver Modules or Porting Windows NT Server Software to Linux. These tutorials are the true heart of this kind of show. They are where the serious learning takes place. There are keynote speeches too, of course, and they were as well-attended at this show as at any I’ve attended, but when you come right down to it they tend to have a certain sameness to them after you’ve heard 10 or 20 of them. The only experienced Linux show attendees I know personally who go hear these speeches at every single show are journalists who are expected to write stories about them, and they usually do not look forward to this duty.
The exhibit floor is where most of the milling-around is done. That’s where you run into old friends and make new ones. Bars and hotel lobbies near the convention site serve a similar purpose, with the added advantage of offering a non-work atmosphere that encourages conviviality in a way a formalized, salesperson-filled exhibit hall cannot.
The summer LinuxWorld show grows every year. A tidbit I picked up from a conversation between two employees of IDG, the company that puts on LinuxWorld, indicated that they turned down 45 major corporate exhibitors who wanted in on this year’s show but inquired too late; all available space in the San Jose convention center was filled to the point that the show keynotes were held in a hall across the street. This overflowing, not the lack of convenient places in San Jose to hold large all-ages parties, is the main reason next year’s summer LinuxWorld will be held in San Francisco, not San Jose.
But we can discuss that later. Right now I am tired, and so are many other people here. There is always a strong feeling of letdown as one of these events comes to an end, especially one where the heartbeat group at its center is not the business crowd but a clannish bunch of people who are motivated more by community feelings than by money, as is the case with the inner core of the Linux movement.
The next stage, of course, as the LinuxWorld-type shows become increasingly marketing-dominated and less friendly to the original Linux people (and are held in increasingly expensive places) is for the inner cult crowd to put on its own series of smaller, less commercial shows. These shows could either be non-profit events or under the aegis of a Linux-oriented company that remembers that the LUG-based community of developers and users, not the people wearing company logoed polo shirts standing in $500,000 trade show booths, is the real reason for the whole get-together.
Category:
