Original Link: https://www.anandtech.com/show/786



Introduction

It has been five months since VIA released the KT133A chipset and most motherboard manufacturers have released their corresponding solutions as well. If you are thinking about upgrading a current system, dropping in a KT133A based motherboard is one of the most cost effective ways to do it, since you will be able to reuse existing PC133 memory.

However, being the most cost effective upgrade right now does not necessarily mean it is the best solution in the long run. As CPU's become more and more powerful, memory bandwidth is getting to be more and more of a bottleneck. We have seen Intel release the i850 chipset with dual channel RDRAM while AMD has gone the DDR SDRAM route with their 760 chipset. Although these archrivals are pushing different technologies, both are designed to provide much higher memory bandwidth than current SDR SDRAM.

VIA's KT133A, with its "old" SDR SDRAM technology, has been able to remain competitive in most cases since the vast majority of today's programs are not able to take advantage of all the memory bandwidth offered by the newer technologies. As seen in our KT133A chipset review, the KT133A performs very closely to the AMD 760, and that's exactly why KT133A motherboards are still the choice for many AnandTech readers, especially since older PC133 SDRAM can be reused.

Furthermore, several problems have recently been discovered with many of VIA's chipset, with the KT133A being one of the most affected. One of the biggest issues is with data corruption when combining a hard drive hooked up to the 686B South Bridge with a Creative Labs Sound Blaster Live! sound card. At the same time, problems with connecting certain USB devices also seems to create some problems. A lot of users have discovered problems taking advantage of multiplier ratio settings on some KT133A motherboards when trying to overclock to a 133MHz FSB. This last one is more of an issue with certain manufacturer's implementations of multiplier controls, and not really a chipset problem.

Despite being chipset issues at heart, they created a lot of discussion about which KT133A is "bug free" and which one performs the best. At the same time, numerous BIOS updates and board revisions have been released that help to correct some the above problems. As such, we thought now would be a good time for us to take a look at all the KT133A boards we have in the lab and see which is best overall.

For this roundup, we are able to gather a total of eleven boards, including the ABIT KT7A-RAID, AOpen AK73 ProA, ASUS A7V133, EPoX EP-8KTA3, FIC AZ11EA, Gigabyte GA-7ZXR (2.2), Iwill KK266-R, MSI K7T Turbo, MSI K7T Turbo-R, Soltek SL-75KAV-X, and Soyo K7TVA.



KT133A: The Final Version of KT133?

The KT133A chipset is composed of two parts, the 8363A North Bridge and the 686B South Bridge. The 8363A North Bridge is exactly the same as the 8363 North Bridge used in the KT133 chipset, except that it has official support for 133MHz FSB operation. With the original KT133 chipset, FSB overclocking was generally not a viable option since it couldn't be pushed any higher than 110MHz - a mere 10% overclock. Therefore, although most manufacturers provided FSB speeds of as high as 166MHz or more, a large portion of these speeds were useless for overclocking.

Fortunately, the new 8363A North Bridge is designed to run at 133MHz and can run reliably at 160MHz or above. You may wonder why VIA did not just support 133MHz FSB from the beginning, and that would be a very valid question. VIA's answer was that they did not want to give too much room for overclocking, so they disabled something in the chipset to prevent such high speed operation.

The increased usable range of FSB speeds gives us a lot more flexibility when it comes to overclocking. In the past, you had to rely mostly on multiplier overclocking, but there is a lot of performance to be gained from faster FSB speeds. With the new KT133A chipset, users can now raise the FSB speed higher and lower the clock multiplier to boost performance noticeably without actually raising the core clock of the CPU.

Like the older 8363 North Bridge, the 8363A chipset also support AGP 1X/2X/4X, so most manufacturers have included universal AGP slots. That means that the AGP slot is not keyed for 1.5V or 3.3V AGP cards only, so you should have no problem installing just about any AGP video card. Some manufacturers have gone a step further and included AGP Pro slots, which follow the same performance specifications, but provide additional electric current for high-end video cards.

The new 686B South Bridge is pin compatible with the older 686A, with the difference between the two being the addition of Ultra ATA 100 support on the 686B. Although at this point no hard drive has a sustained transfer rate of higher than 66MB/sec, it's only a matter of time before hard drives pass this barrier, at which point Ultra ATA 100 will become a more important feature.

Also like the 686A, the 686B features two USB root hubs, supporting a total of four USB ports. Two traditional USB ports are usually mounted at the back of the board within the ATX I/O panel, while a header for the second set of USB ports are usually located on the left side of the motherboard. You'll need a USB bracket to take advantage of these ports, but fortunately manufacturers have started to include such a bracket in the box.



What to look for in a KT133A motherboard

In order for everything to function properly under Windows 98/ME or Windows 2000, you should be sure to install the latest VIA 4-in-1 drivers. At the time of publication, that is 4.30, which includes several bug fixes and performance enhancements. You should also check closely to see if there are any BIOS updates for your motherboard. Most of the time, you can find this information on the manufacturer's homepage.

When it comes to overclocking, multiplier ratio settings are no longer a luxury, but rather an expected and necessary feature. Now that there is official support for FSB speeds of 133MHz, FSB overclocking ability has become a major focus.

Other overclocking features such as voltage tweaking, are also something we have to consider, since they contribute to the overall stability of an overclocked system.

As mentioned previously, the KT133A officially supports a 133MHz FSB, and can push 160MHz in overclocked form. With that in mind, traditional PC133 SDRAM is often not good enough to handle such a high FSB. Typical PC133 SDRAM should be able to run around 143MHz at CAS 3.

Unfortunately, there is no industry standard memory specification higher than 133MHz, so any "PC150" memory you see out there is likely nothing more than a PC133 DIMM with faster chips installed. Fortunately, most of these higher speed modules do indeed work fine at the speeds they are rated for and are a must if you want to push your FSB to the max.

Several companies have released PC150 memory, and the one we have been using in the AnandTech labs is Mushkin's High Performance Rev. 3 memory. The memory chips are still rated as 7.5ns (133MHz), but they are handpicked and pretested at 150MHz CAS 2. Our tests show that this memory is able to run at 163MHz CAS 3 reliably. This should be more than enough to overclock your KT133A motherboard without having the memory become a bottleneck.

One last thing to look for in terms of overclocking is the multiplier "issue" that has been encountered on some KT133A motherboards. The topic is a bit intricate, so read our complete article if you want all the details and a solution. Basically, the problem arises on boards that have a 100/133MHz jumper, in conjunction with a 100MHz FSB CPU overclocked to 133MHz FSB with a lower multiplier. The result is that the system does not POST at all when powered on. We'll let you know in this roundup which boards have this "issue" and which have solved it.

Of course, we will also look at our standard stability tests, which compare the number of crashes in 24 hours while running a system stress loop. Remember that not all crashes in the stress test are fatal, and many would go unnoticed by the user. Also remember that this is a stress test and is much tougher on the system than anything a single user could apply in the same time period.

For each individual motherboard, we will also consider its unique features including layout, expansion slot configuration, extra features like IDE RAID controller and diagnostic tools. Now without further ado, let's start looking at the candidates.



ABIT KT7A-RAID


Click to Enlarge

ABIT KT7A-RAID

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Highpoint HPT370 IDE RAID (optional)

Form Factor
ATX
Bus Speeds

100 - 183MHz (1MHz increments)

Core Voltages Supported

1.100 - 1.850V (0.025V increments)

I/O Voltages Supported
3.2 - 3.9V (0.1V increments)
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (4 full length)
1 ISA Slot (1 Shared)

On-board Audio
N/A
BIOS

Award Modular BIOS 6.00PG


The KT133 based KT7-RAID won our Editor's Choice Award back in July 2000 for its brilliant design, performance, and stability. The new KT7A-RAID is based on the same design and delivers on the high expectations placed on the follow up to such a high quality product.

ABIT has never geared their products towards the OEM market; rather they have designed their products for all the hardware enthusiasts out there with a strong focus on overclocking features, which is exactly how ABIT has made their name. Probably the most attractive feature of the board is SoftMenu III, which ABIT has been using since the BX133-RAID. With SoftMenu, you can perform every overclocking option right inside the BIOS, so you don't have to go through the hassle of shutting down the machine and setting jumpers or dipswitches. While ABIT was the pioneer of jumperless CPU setups with the original SoftMenu, they are no longer the only game out there. Fortunately for ABIT, SoftMenu III continues to be one of the best, easiest to use, and most flexible jumperless utility out there.


The famous SoftMenu III setup

Provided that your chip is unlocked (either from the factory or by reconnecting the L1 bridges), you can easily change the CPU's clock multiplier right inside the BIOS, where you can also select any FSB speed between 100MHz and 183MHz in 1MHz increments. ABIT pioneered the idea of FSB speeds in 1MHz increments with the original SoftMenu III and has now become a hallmark of the feature. Inside the BIOS, you can also tweak the CPU core voltage as well as I/O voltage for added stability when overclocking.

Stability of the board has been very good as well. Although it is not the most stable KT133A we have tested, it only crashed twice in 24 hours, wihch is still very impressive among the boards we have here.

The layout of the board is virtually identical to the one on the KT7-RAID, with the biggest difference being that ABIT decided to put a full HSF unit on the 8363A North Bridge. This seems to be a good idea since the 8363A seems to generate considerable amounts of heat during normal operation, much less overclocked situations. The power connector is also very nicely placed so that power cables will not run over the CPU and memory. Unfortunately the otherwise excellent layout is marred by the placement of the CPU socket too close to the right edge of the board, making CPU removal difficult once everything is installed in a case.

The expansion slot configuration is one of the best we have seen. With six PCI slots, you should have plenty of room for all your expansion cards. The extra ISA slot will be welcomed by those who are still hanging onto those old legacy devices. You may have noticed that ABIT almost never includes AMR/CNR/ACR slots on their boards, mainly because of their target market where these cards are almost never used.

Last but not least, ABIT has also included the Highpoint HPT370 Ultra ATA 100 IDE RAID controller. It supports RAID 0, RAID 1, and RAID 0+1, the three most commonly used RAID setups out there. In fact, the KT7A-RAID is actually one of the only two boards that support these three modes. RAID 5 is excluded, of course, because it would require much more processing power than can be provided by a single chip solution like this. Of course, if you don't need or want RAID functionality, you can choose to use it as just a second IDE controller, for a total of four IDE channels on the board, supporting up to eight IDE devices. If you don't even want to use it as a secondary IDE controller, ABIT offers a non-RAID version of this board, known as the KT7A, that is otherwise identical.

Overall, the KT7A-RAID is a very nice KT133A motherboard. Although it doesn't stand out quite as much as its predecessor, mainly because the other manufacturers have caught up, ABIT set the standard for the KT133A market we see today. Of course, the overclocking options are truly phenomenal, as we've come to expect from ABIT.



AOpen AK73 ProA


Click to Enlarge

AOpen AK73 ProA

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Form Factor
ATX
Bus Speeds

100 / 102 / 104 / 106 / 107 / 108 / 109 / 110 / 111 / 112 / 113 / 114 / 115 / 116 / 118 / 120 / 124 / 129 / 130 / 133 / 136 / 140 / 145 / 150 / 155 / 160 / 166 MHz

Core Voltages Supported

1.100 - 1.850V (0.025V increments)

I/O Voltages Supported
3.2 - 3.5V (0.1V increments)
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
5 PCI Slots (5 full length)
1 AMR Slot

On-board Audio
Analog Devices AD1885 AC'97 CODEC
BIOS

Award Modular BIOS 6.00PG


In the past, AOpen had become one of our favorite motherboard manufacturers thanks to outstanding products that won more Editor's Choice Awards than any other manufacturer. Unfortunately, they've been less impressive of late and, quite frankly, the AK73 ProA was quite disappointing. It may be a good candidate for the OEM market, but it simply lacks a number of features that hardware enthusiasts want.

The layout of the board is based on the KT133 based AK73-1394 and is pretty decent. Unlike that board, AOpen didn't include the IEEE 1394 Firewire controller on our evaluation sample. However, we expect a version with Firewire to appear shortly since the silkscreening already in place for the controller.

As far as overclocking goes, the multiplier ratio settings are implemented through a set of dipswitches, meaning that you will have to shut down the system to modify the multiplier. The same goes for the I/O voltage controls except that they are controlled by jumpers instead of dipswitches. CPU core voltage, on the other hand, is adjustable in the BIOS.


Dipswitches for multiplier ratio settings



Multipliers are not adjustable in the BIOS

Like many other AOpen motherboards, FSB settings are semi-jumperless. You can select your desired FSB speed inside the BIOS, but the available speeds are determined by the position of two jumpers on the board. Even then, the choices are limited, with a total of just 27 entries. If you look at FSB speeds higher than 133MHz, you will notice that there are quite a few gaps that could prevent you from reaching the true limit of your CPU and/or memory.

Stability of the board is not particularly impressive either, crashing a total of four times in 24 hours, which was second worse among this group. This was somewhat surprising since AOpen has always been known for its quality and stability, but that's just not the case with the AK73 ProA unfortunately.

The AK73 ProA features five PCI and one AMR slots, which is pretty standard today. However, AOpen has yet to implement features such as a RAID controller on any of their boards, although the Firewire option may be more useful to some. AOpen does continue to include their Die-Hard BIOS, which is very similar to Gigabyte's Dual BIOS feature. Basically, there are two BIOS chips on the board, and in the case of corrupted BIOS, you can switch and boot from the backup to recover your system. In the case of Die-Hard BIOS, you have to physically move a jumper to make the change.

In short, we are quite disappointed with AOpen's AK73 ProA, since it not only lacks the necessary overclocking options for users, but the stability was also below par.



ASUS A7V133


Click to Enlarge

ASUS A7V133

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Promise PDC20265 IDE RAID (Optional)

Form Factor
ATX
Bus Speeds

90 / 95 / 100 - 166 MHz (in 1MHz increments)

Core Voltages Supported

1.10 - 1.85 V (in 0.05V increments)

I/O Voltages Supported
3.35 / 3.56 V
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
5 PCI Slots (3 full length)
1 AMR Slot

On-board Audio
VIA VT1611A AC'97 CODEC
BIOS

Award Medallion BIOS 6.00


Most KT133A boards are strongly based on their original KT133 counterparts, usually with the same PCB, features, and to a large extent BIOS. ASUS is never one to follow the status quo, however, and the A7V133 is one of the few boards here that received significant improvements over its KT133 predecessor. The board physically looks more or less the same as the A7V, including the unique vertically mounted voltage regulator setup. The only obvious change is the addition of a nicely designed HSF that sits atop the 8363A North Bridge.


ASUS has yet another unique design, with the vertically mounted PCB on the A7V133

Without a doubt the most impressive improvement on the A7V133 is the migration to 1MHz increments in FSB speeds. With the old A7V, we complained that the lack of 1MHz increments hurt overclocking potential for those looking to maximize the performance over the CPU. It's clear that ASUS listened to their customers and made the corresponding changes.

Multiplier settings are also nicely implemented in the BIOS this time around, so you should have no problem overclocking on the fly, without shutting down your system. Further, a wide spectrum of CPU core voltages is also available in the BIOS. The only non-jumpered option on the A7V133 is the I/O voltage, which is still controlled through a jumper.

Stability of the A7V133 has been outstanding in our stress tests. In fact, the A7V was among the most stable motherboards we have tested and did not crash once in 48 hours of stress testing. This is quite impressive since we generally encountered at least one crash for every 24 hours.

Of course, it seems that no board today would be complete without onboard IDE RAID, so the A7V133 comes with an optional Promise Ultra ATA 100 RAID controller. The old Promise controller used on the A7V acted as just a standard IDE controller, without RAID support, but that's been upgraded here. Note, however, that ASUS uses what they call a Lite BIOS for the RAID setup, which only supports RAID 0 configurations - RAID 1 and RAID 0+1 are completely missing. Fortunately, RAID 0 is the setting that the vast majority of people are likely to use. If you're still not satisfied with that, there is already information on the web that explains how to get full support for the missing two RAID configurations. However, keep in mind that you are modifying the BIOS and there is a chance that you could ruin the BIOS.


The newer Promise controller allows RAID 0 setups only

Since ASUS is one of the largest OEM suppliers of motherboards, they have to please their customers in those markets as well, so sometimes their designs are not fully geared towards hardware enthusiasts. The board features five PCI slots and one AMR slot - not very impressive in the face of boards with six PCI slots, especially since the AMR slot is not useful for most of us.

Unlike the A7V, ASUS took away the additional three USB ports provided by an onboard USB hub chip, leaving the standard four USB ports supported by the 686B South Bridge. This was probably done as a cost cutting measure, and it may still be possible to order the board with the chip installed as an option.. Last of all, like most other recent ASUS boards, the A7V133 features an AGP Pro slot.

It is safe to say that the A7V133 is one of the best KT133A boards in the market. The addition of 1MHz increments when adjusting FSB speeds makes the board a complete package for overclockers, while the excellent stability under all circumstances will satisfy everyone. We were able to reliably hit 153MHz FSB in the AnandTech labs.



EPoX EP-8KTA3


Click to Enlarge

EPoX EP-8KTA3

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Form Factor
ATX
Bus Speeds

100 / 102 / 104 / 106 / 107 / 108 / 109 / 110 / 111 / 112 / 113 / 114 / 116 / 118 / 120 / 124 / 127 / 130 / 133 / 140 / 145 / 150 / 155 / 160 / 166 MHz

Core Voltages Supported

-0.1 / -0.075 / -0.05 / -0.025 / +0.025 / +0.05 / +0.075 / +0.125 / +0.150 V

I/O Voltages Supported
+0.05 / +0.10 / +0.15 / +0.20 / +0.25 / +0.30 / +0.35 V
AGP Voltages Supported
+0.1 / +0.2 / +0.3 / +0.4 / +0.5 V
Memory Slots
4 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (3 full length)
1 ISA Slot (1 Shared)

On-board Audio
VIA VT1611A AC'97 CODEC
BIOS

Award Medallion BIOS 6.00


The EPoX EP-8KTA3 is one of the few KT133A motherboards that actually has physical layout changes. Fortunately, these changes appear to be for the better. Compared to their KT133 solution, the EP-8KTA2, the most useful change is the migration of multiplier ratio settings from dipswitches to the BIOS. As we have discussed before, having the multiplier ratio settings inside the BIOS makes overclocking much easier for people with an unlocked CPU.

However, FSB speed options are something EPoX did not take seriously. Although the FSB is adjustable inside the BIOS, 1MHz increments are not available. Indeed there are only six settings that are above 133MHz, which means that there are gaps of at least 5MHz between settings, which could prevent you from maximizing your CPU's overclocking potential.

On the other hand, EPoX did a good job in the voltage tweaking department, by not only including CPU core and I/O voltage settings, but also AGP voltage controls. Our tests show that by raising the I/O and AGP voltage, you can make the board more stable in overclocked situations, so the added feature is definitely a plus. The AGP voltage adjustments mainly come in handy when overclocking the AGP bus, which occurs whenever the FSB is also overclocked.

The EP-8KTA3 is much more stable than the EP-8KTA2, crashing just two times in 24 hours of stress testing, tying the ABIT KT7A-RAID. Once again, this is not the best result we have seen with KT133A boards, but is definitely very stable among all boards we have ever looked at.

The expansion slot configuration of the EP-8KTA3 is quite impressive with its six PCI slots, one ISA slot, and no AMR slot. This type of setup is probably most desired among hardware enthusiasts out there.

The board we received did not include onboard IDE RAID, but EPoX does offer the option and it comes with the EP-8KTA3+, which is otherwise identical. The EP-8KTA3+ includes Highpoint's HPT370 IDE RAID controller, the same that ABIT uses on a number of their boards, including the KT7A-RAID included here. Like the ABIT KT7A-RAID, there is support for RAID 0, RAID 1, or RAID 0+1 functionality or it can be used as just a second IDE controller for connecting an additional 4 drives.


The silkscreen for the Highpoint IDE RAID controller

For diagnosing problems, EPoX also used the release of the EP-8KTA3 to introduce their new diagnostic tool, the P80P LED Debugging Display. On the front of the board, you can see two 8 segment LED displays. During boot-up they will show two hexadecimal values representing the state of the system. If the machine fails to boot correctly, you would be able to use this two digit code to find out what the problem is associated with. The advantage of this setup is that it can store a total of 256 states, which is considerably more than the 16 states represented by most current diagnostic tools out there which use four LED's.


EPoX's new diagnostic tool

To enhance overclocking and improve stability, EPoX also installed a full HSF unit on the 8363A North Bridge, as well as heatsinks on the voltage regulator's MOSFETs closest to the CPU.

In general, the EP-8KTA3 is a solid board, with quite a few overclocking features for us. The downside is the lack of 1MHz increments for FSB speed selection, which affects our overclocking results, and the performance of the board, which is merely average.



FIC AZ11EA


Click to Enlarge

FIC AZ11EA

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Form Factor
ATX
Bus Speeds

90 / 95 / 100 / 101 / 102 / 103 / 105 / 107 / 109 / 110 / 111 / 113 / 115 / 117 / 120 / 125 / 130 / 133 / 135 / 137 / 139 / 140 / 143 / 145 / 148 / 150 / 155 / 166 MHz

Core Voltages Supported

1.500 - 1.850 V (in 0.025V increments)

Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
5 PCI Slots (5 full length)
1 CNR Slot

On-board Audio
Audio Sigmatel STAC9708T AC'97 CODEC
BIOS

AMI Simple BIOS Setup 1.24b

The FIC AZ11EA very closely resembles their second KT133 solution, the AZ11E, which was a very decent motherboard. The same design is carried over to the AZ11EA, which also delivers decent performance and quality.

Probably the most annoying thing about the board is the long power cable for the HSF unit sitting on the 8363A North Bridge. Since the AZ11EA is based on the AZ11E, which has only two fan headers, FIC was forced to use the fan header sitting on the left side of the board for the HSF unit on the 8363A North Bridge for cooling. Together with the necessary fan header for the HSF unit on the CPU, that leaves no additional headers for more fans to be attached to the board directly.


The long wire of the HSF unit on the 8363A North Bridge

The layout of the board is good for the most part, except for the power connector that is placed too close to some capacitors. This makes removing the power cable a bit difficult. Additionally, some MOSFETs and capacitors are quite close to the CPU socket as well, possibly preventing the use of larger HSF units.

The performance of the board was pretty average, except in Quake III, where it came in last.

Multiplier ratio and CPU core voltage settings can be adjusted by two sets of dipswitches, which is not as friendly as some other KT133A boards we have here. Meanwhile, I/O voltage tweaking is not available, which could affect overclocking results negatively.


Multiplier ratios and CPU core voltages are adjusted using dipswitches

FSB speeds are not available in 1MHz increments either, although FIC did a good job of providing a variety of values so the gaps in between settings for speeds higher than 133MHz are not too big. Overclocking of the board was not too impressive, possibly due to the lack of I/O voltage tweaks. We were able to push the board to 145MHz before the board started to lose stability.

Overall, the AZ11EA is still a very good board, but it's just not the best for hardware enthusiasts. FIC failed to include some of the extra features that help with overclocking.



Gigabyte GA-7ZXR (2.2)


Click to Enlarge

Gigabyte GA-7ZXR (2.2)

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Promise PDC20265R IDE RAID (Optional)

Form Factor
ATX
Bus Speeds

133 - 150MHz (in 1MHz increments)
95 / 160 / 200 MHz

Core Voltages Supported

1.500 - 1.850 V (in 0.025V increments)

I/O Voltages Supported
3.3 / 3.4 / 3.5 V
AGP Voltages Supported
1.5 / 1.6 / 1.7 V
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (6 full length)
1 AMR Slot

On-board Audio
Sigmatel STAC9708T AC'97 CODEC
BIOS

AMI Simple BIOS Setup 1.24b


The Gigabyte GA-7ZXR revision 2.2 is the last KT133A motherboard we reviewed and we were absolutely shocked at just how much they could do to a product in a five-month span. The first revision of the board was actually a KT133 board, and we concluded in our November 2000 KT 133 Roundup that the board was far from satisfactory.

Not long after the roundup, we got word from Gigabyte that they were working on including more features in a future version of the board. The first thing they did was add multiplier ratio settings, something most other manufacturers had already included. Then they added quite a few more FSB speeds as well as CPU core voltage tweaks, so that the board could compete with other competitors. We expected these two changes, but the list doesn't stop there. Gigabyte has actually become only the second manufacturer to include both I/O and AGP voltage tweaks, something we did not expect at all.

When it comes to FSB speeds, Gigabyte again surprised us by including 1MHz increments for more precise overclocking. You can choose speeds freely between 100MHz and 120MHz, and then between 133MHz and 150MHz all in the BIOS.

Similar to the ASUS A7V133A, Gigabyte also uses the Promise IDE RAID controller on the GA-7ZXR (2.2). Unlike ASUS, Gigabyte has included the complete Proimse RAID BIOS, which supports both RAID 0 and RAID 1 configuration, giving you more flexibility. For some odd reason, however, RAID 0+1 support has not been included.



The Promise controller supports RAID 0 and RAID 1

Gigabyte even went in and changed the expansion slot configuration as well. Instead of five PCI and one ISA slots found in the first revision, the current version features a total of six PCI slots, but the ISA slot has been dropped.

Overall, the GA-7ZXR (2.2) is a very nice board when it comes to features in overclocking, expansion, and IDE RAID. Unfortunately, the actual overclocking potential of the board was not as good as we expected. The same goes for the performance, which was not quite up to par.



Iwill KK266


Click to Enlarge

Iwill KK266

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

AMI AG 80649 IDE RAID (Optional)

Form Factor
ATX
Bus Speeds

100 - 166 MHz (in 1MHz increments)

Core Voltages Supported

1.525 - 1.850 V (in 0.025V increments)

I/O Voltages Supported
Auto Detect / + 5% / +10%
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (3 full length)
1 ISA Slot

On-board Audio
C-Media CMI-8738 3D 4.1 Hardware Sound
BIOS

Award Modular BIOS 6.00PG


Iwill is one of the smallest motherboard manufacturers in Taiwan which explains why a lot of people have not heard about them before. But that's all changing now as Iwill has begun producing a number of top notch boards. A prime example of this is their KT133A solution, the KK266, which shows everyone that even a small company can make something outstanding.

The success of the KK266 lies not only in the mostly jumperless design, but also in the excellent overclocking ability. The only jumper on board is the 100/133MHz setting, which may in fact cause problems in certain cases, so make sure you read our KT133A Chipset Issue article. The layout of the board is very similar to that of the ABIT KT7A-RAID, which is a very decent one, as we have discussed previously.

All the overclocking features can be found inside the BIOS. Multiplier ratio settings are implemented nicely, and FSB speeds are available from 100MHz to 166MHz in 1MHz increments. Iwill also includes the appropriate range of CPU core and I/O voltages in the BIOS.

Like the ABIT KT7A-RAID, the KK266 features six PCI slots and one ISA slot, showing that Iwill is still gearing their products more towards hardware enthusiasts rather than OEM builders.

Performance of the board is very good - among the top boards here - and more importantly, the board overclocks extremely well. We were able to reach FSB speeds as high as 161MHz, a substantial 21% overclock. This is so far the best overclocking result we have seen among all KT133A motherboards.

Furthermore, the board was very stable as well, tying with the ASUS A7V133 for the top spot by not crashing a single time in 48 hours of our stress tests. Even at FSB speeds of 162MHz, the board was able to maintain the same level of stability.

Iwill also offers another version of the board, the KK166-R that adds integrated IDE RAID. Unlike most other manufacturers here, Iwill is the only one going with AMI's Ultra ATA 100 IDE RAID controller. The RAID controller supports the three most popular modes of operation, RAID 0, RAID 1, and RAID 0+1.



The silkscreen for AMI IDE RAID controller found on the KK266-R

The Iwill KK266 is clearly one of the best KT133A motherboards on the market, if not the best. The KK266 should show the world just how good a small manufacturer like Iwill can be. It offers an almost jumperless design, all the required overclocking options right inside the BIOS, excellent performance and overclocking, and it is also one of the most stable boards we have seen so far. What more can you ask for?



MSI K7T Turbo-R


Click to Enlarge

MSI K7T Turbo-R

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Promise PDC20265R IDE RAID (Optional)

Form Factor
ATX
Bus Speeds

100 - 166 MHz (in 1MHz increments)

Core Voltages Supported

1.300 - 1.850 V (in 0.025V increments)

I/O Voltages Supported
3.30 / 3.45 V
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (4 full length)
1 CNR Slot

On-board Audio
VIA VT1611A AC'97 CODEC
BIOS

Award Modular BIOS 6.00PGN

MSI was the winner in our November 2000 KT133 roundup, and their K7T Pro2 also earned our Editor's Choice Award for its outstanding performance and stability. So we of course had high hopes for MSI's KT133A solution, the K7T Turbo series. The board we received was the K7T Turbo-R, which includes the optional onboard Promise IDE RAID controller.

The K7T Turbo is based closely on the K7T Pro2 in terms of layout, but that board did not feature an onboard RAID option, so there were some significant changes made to the south side of the board, while the north part remains identical. MSI put the Promise controller in front of the fifth and sixth PCI slots with the two IDE connectors placed vertically in front of the controller. Although they don't block the PCI slots, the connectors may be too close to the bottom of the case when the board is installed, which might prevent the IDE cables from reaching the hard drive bays.

Regardless, the Promise IDE RAID controller supports both RAID 0 and RAID 1 configuration, like the Gigabyte GA-7ZXR (2.2). Interestingly enough, RAID 0+1 is again missing from the list, possibly to get a better deal from Promise on the actual chips by agreeing to include a slightly crippled BIOS.

Like the K7T Turbo, most of the overclocking features are available within the BIOS. One thing to watch out for is that the original engineering sample we received did not support the full range of CPU clock multipliers (5 to 12.5 in 0.5 increments). Instead, only multipliers higher than the default multiplier of your chip were selectable, which would be a problem when trying to lower the multiplier of your chip in order to raise the FSB. MSI has stated that a new BIOS would solve the problem, but we have yet to receive it, so make sure you check up on this before purchasing if it affects you.

The FSB speed ranges from 100MHz to 166MHz in 1MHz increments and is selectable within the BIOS, a nice feature that will help ensure that you can push your chip to its limit. Moreover, all the CPU core and I/O voltage options are also available in the BIOS so you can overclock easily. The only down side to the overclocking feature set, if any, would be the 100/133MHz jumper, which means that this board is affected by the KT133A Overclocking Issue.

Performance of the board was a little bit below average, but that's not the main focus of MSI. They have always focused on making sure their products are of the highest quality and stability, just as we found with the K7T Turbo. Together with the ASUS A7V133 and the Iwill KK266, the K7T Turbo was the third board arrived in our labs to survive 48 hours of our stress tests without crashing.

In short, MSI continues to live up to expectations with the K7T Turbo. Although performance and overclocking results were not the best here, MSI does continue to deliver high quality and very stable products to hardware enthusiasts.



MSI K7T Turbo-R Limited Edition


Click to Enlarge

MSI K7T Turbo-R Limited Edition

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Promise PDC20265R IDE RAID (Optional)

Form Factor
ATX
Bus Speeds

100 - 199 MHz (in 1MHz increments)

Core Voltages Supported


1.550 - 1.850 V (in 0.025V increments)

I/O Voltages Supported
3.30 / 3.45 V
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
6 PCI Slots (4 full length)
1 CNR Slot

On-board Audio
VIA VT1611A AC'97 CODEC
BIOS

Award Modular BIOS 6.00PGN

Just in time for this roundup, we received the K7T Turbo-R Limited Edition from MSI. The board is virtually identical to K7T Turbo-R, but with some interesting new features. The most obvious is the red PCB that caught our attention as soon sa we opened up the box.

More importantly, however, several settings in the BIOS also caught our attention. For starters, the FSB range increased from 100-166MHz on the original board to 100-199MHz on the Limited Edition. Although the highest FSB we have successfully achieved on the KT133A is 162MHz, the added values above 166 should provide more headroom for extreme overclockers.


You can choose speeds between 100MHz and 199MHz now

We also noticed multiplier values between 13 and 15 are now available, something that no other KT133A board can do to this point. It will be interesting to see if those values work and whether the board would will accept the new Palomino core Athlon's. Unfortunately, at the time of publication, those chips were not readily available in the lab for testing. It's clear, however, that MSI is looking to the future with this board.



Multipliers between 13 and 15 are now available

This board also brings us the introduction of a new MSI innovation - PC2PC allows you to use one of the four USB ports provided by the 686B South Bridge to connect two PCs together for Internet sharing and data transfer. In order to take advantage of that, MSI includes a special USB bracket to connect to the USB header and a long USB cable to connect the two PCs together. One machine must be the MSI machine with the special USB bracket, while the second can be any computer with a USB port. Once the two PCs are physically connecting, installing the software is easy and you end up with a USB network between the two machines. MSI still includes a traditional USB bracket in case you don't want to use the PC2PC feature.


The new splash screen seen during boot up

The only problem we ran into with the board was stability, which was not particularly bad, but not up to par with the standard K7T Turbo-R. The Limited Edition crashed twice in 24 hours of stress testing compared to none in 48 hours for the standard board. This could be due to a BIOS that is still beta, but it's hard to tell until the final version is available.

For the Limited Edition boards, MSI also includes a set of new manuals with more information on the board, as well as some new software. In general, the Limited Edition board is an enhanced version of the K7T Turbo-R, with more features and even higher quality construction. Of course the board comes with a slightly higher price tag, but if you want the absolute best KT133A product from MSI, this is a board you can't miss.

Other than what we have noted here, the remaining features are the same as the K7T Turbo-R, so you can refer to that board's description in this article for more detail.



Soltek SL-75KAV-X


Click to Enlarge

Soltek SL-75KAV-X

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Form Factor
ATX
Bus Speeds

100 - 183 MHz (in 1MHz increments)

Core Voltages Supported


1.550 - 1.850 V (in 0.025V increments)

I/O Voltages Supported
3.3 / 3.4 / 3.5 / 3.6 V
Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
5 PCI Slots (3 full length)
1 ISA Slot

On-board Audio
ICE 1232 AC'97 CODEC
BIOS

Award Modular BIOS 6.00PGN

The Soltek SL-75KAV-X was actually the first KT133A to arrive in the lab, and was the board we used for our original KT133A chipset review. Unfortunately, after extensive testing on all KT133A boards, we found out that the SL-75KAV-X actually performs relatively poorly, coming out as one of the slowest KT133A boards we've tested.

In term of features the board is pretty decent, offering 1MHz increments on FSB speeds between 100MHz and 183MHz, the same range available on the ABIT KT7A-RAID. It also includes CPU core voltage tweaks in the BIOS, but I/O voltage settings are only available using a set of jumpers. Multipliers are adjustable using a set of dipswitches on the board.


Dipswitches for changing the multipliers of the CPU

It's interesting to note that only one serial port is included in the ATX I/O panel because Soltek uses the same PCB to produce a KM133 version of the board. A short ribbon cable and bracket are included if you need that second serial port.


The second serial port is missing from the I/O panel

The layout of the board is pretty decent as well, although a few components on the right side of the board might get in the way of mounting some HSF units. The power connector is found along the right edge, so you won't have to run the power cables over the CPU or memory. The two IDE connectors are placed vertically in front of the first PCI slot, but in such a way that they do not block the use of full length PCI devices.

Unfortunately, the performance of the board is far from ideal and the stability is also below average, crashing a total of five times in 24 hours - the worst among KT133A boards we have tested.

Further, overclocking was not as successful as you might expect from the feature set, as we found that the board couldn't get past 140MHz at all, which is not even a 10% overclock. Soltek definitely needs to work on the overall stability of the board as well as the quality, as we found that the second DIMM slot actually did not work on our sample.

All in all, we can't recommend the SL-75AV-X at this point due to its serious quality and stability issues. Performance and overclocking are other factors that will prevent most AnandTech readers from choosing this board as well.



Soyo SY-K7VTA Pro


Click to Enlarge

Soltek SL-75KAV-X

CPU Interface
Socket-A
Chipset

VIA KT133A
VIA 8363A North Bridge
VIA 686B South Bridge

Form Factor
ATX
Bus Speeds

100 / 101 / 102 /103 / 105 / 107 / 109 / 110 / 111 / 113 / 115 / 117 / 120 / 125 / 130 / 133 / 135 / 137 / 139 / 140 / 143 / 145 / 148 / 150 / 155 / 166 MHz

Core Voltages Supported


-0.025 / -0.050 / -0.075 / -0.100 / + 0.025 / +0.050 / +0.075 / +0.100 / +0.125 / +0.150 / +0.175 / +0.200 / + 0.225 / +0.250 V

Memory Slots
3 168-pin DIMM slots
Expansion Slots

1 AGP Slot
5 PCI Slots (3 full length)
1 ISA Slot

On-board Audio
Audio Sigmatel STAC9700T
BIOS

Award Modular BIOS 6.00PG


Our last contestant is the Soyo SY-K7VTA Pro, another example of a board that's based on the company's KT133 design. Like its older brother, the SY-K7VA Pro also performed pretty well in our tests.

Like a couple other manufacturers, Soyo also included a full HSF unit on the 8363A North Bridge for better cooling. Soyo added another fan header specifically for this fan. The layout of the board is based Soyo's KT133 solution, where the power supply connector is placed next to some capacitors and the I/O panels, whjch can make unplugging the power cable a difficult task.


The power supply connector is very close to some capacitors

The K7VTA Pro features a total of five PCI slots and one ISA slot, which is pretty average these days. The stability of the board is also pretty average in this group, crashing two times in 24 hours, as is performance.

Changing the multiplier is done through a set of dipswitches, but CPU voltage and FSB speed are both selectable inside the BIOS. Like several other boards we have here, there are some gaps between FSB speeds higher than 133MHz, which limited our overclocking results.

Our overclocking tests showed that the SY-K7VTA Pro could reach an FSB as high as 145MHz before losing stability - around a 10% overclock, which is a bit below average.

In general, Soyo continues to provide solid motherboards for general users, but for hardware enthusiasts, the K7VTA Pro might not be the best choice.



How and What to Compare

Now that we have looked at the eleven boards individually, we have to find a way to compare them from as many perspectives as possible. Performance is definitely a factor, although not a huge one since the variation among boards is only a few percentage points. Many of these boards were reviewed back in early January, and since then a variety of new BIOS revisions have been released which could affect performance noticeably. So, we gathered up the most recent BIOS updates from all the manufacturers and retested the performance to ensure the comparison between the boards is fair.

Next, prices have certainly changed since those initial reviews and this comparison offers an opportunity to look at a snapshot of prices that will give us a fair comparison. However, keep in mind a few things. First of all, these prices are by no means the cheapest you can find out there, but are meant to represent typical prices that you can find relatively easily.

Of course, don't forget that picking any product solely on price is probably not a good idea. Each board has a different set of features, performance, stability, overclocking ability, and quality, so a more expensive board that delivers more features that are useful to you may actually be a better buy for you. Thus, it is critical that you find the right balance between price and features that meets your needs.

We will also include a table showing the results of our stability tests, given in number of crashes within a 24-hour period of stress testing. As we have stressed several times before, our stability tests put much more stress on the system than any single user ever could manage. Thus, when we say a board crashed X times in 24 hours, that does not translate into the board will crash for you X times everyday. It does, however, provide a good tool for comparing the stability of different boards.

Motherboard
Multiplier Settings in BIOS?
FSB Speed in 1MHz increments?
Voltage Adjustments
Overclocked FSB results
RAID
PCI/ISA/AMR/CNR
Crashes in 24 hours
Street Price (USD)
Core
I/O
AGP
ABIT KT7A-RAID
Yes
Yes
Yes
Yes
No
154MHz
Highpoint HPT370
(0,1,1+0)
6/1/0/0
2
$136
AOpen AK73 ProA
No
No
Yes
Yes
No
140MHz
No
5/0/1/0
4
$123
ASUS A7V133
Yes
Yes
Yes
Yes
No
153MHz
Promise PDC 20265 (0)
5/0/1/0
0
$117
EPoX EP-8KTA3
Yes
Yes
Yes
Yes
145MHz
Highpoint HPT370
(0,1,1+0) (Optional)
6/1/0/0
2
$99
FIC AZ11EA
No
No
Yes
No
No
145MHz
No
5/0/0/1
3
$84
Gigabyte GA-7ZXR (2.2)
No
Yes
Yes
Yes
Yes
147MHz
Promise PDC 20265R (0,1)
6/0/1/0
2
$131
Iwill KK266
Yes
Yes
Yes
Yes
No
162MHz
AMI AG 80649 (0,1,0+1) (Optional)
6/1/0/0
0
$115
MSI K7T Turbo-R
Yes
Yes
Yes
Yes
No
152MHz
Promise PDC 20265R (0,1)
6/0/0/1
0
$98
MSI K7T Turbo-R (Limited)
Yes
Yes
Yes
Yes
No
154MHz
Promise PDC 20265R (0,1)
6/0/0/1
0
N/A
Soltek SL-75KAV-X
No
Yes
Yes
Yes
No
140MHz
No
5/1/0/0
5
$99
Soyo SY-K7VTA Pro
Yes
No
Yes
No
No
145MHz
No
5/1/0/0
2
$103


The Performance Test

In recent times, choosing a motherboard cannot be completely determined by a Winstone score. Now, many boards come within one Winstone point of each other and therefore the need to benchmark boards against each other falls. Therefore you should not base your decision entirely on the benchmarks you see here, but also on the technical features and advantages of this particular board, seeing as that will probably make the greatest difference in your overall experience.

Test Configuration

Processor(s):
AMD Athlon (Thunderbird) 1GHz
RAM:
1x 128MB Mushkin PC133 SDRAM
Hard Drive(s):
Western Digital 153BA Ultra ATA 66 7200 RPM
Bus Master Drivers:
VIA 4-in-1 v4.24 Service Pack
Video Card(s):
NVIDIA GeForce 2 GTS 32MB DDR
Video Drivers:
NVIDIA Detonator 5.22
Operation System(s):
Windows 98SE
Motherboard & BIOS Revision:
ABIT KT7A-RAID Revision 1.0 (BIOS YH)
AOpen AK73 ProA Revision 1.04 (BIOS 1.07)
ASUS A7V133 Revision 1.04 (BIOS 1002 A1)
EPoX EP-8KTA3 Revision 0.3 (BIOS 1417)
FIC AZ11EA Revision (BIOS VDC45)
Gigabyte 7ZXR Revision 2.2 (BIOS FA6)
Iwill KK266 Revision 1.1 (BIOS 34240A)
MSI K7T Turbo Revision 3 (BIOS 10F)
MSI K7T Turbo-R Limited Edition Revision (BIOS 2.6B13)
Soltek Revision (BIOS Q4)
Soyo SY-K7VTA Pro Revision (BIOS 2AA3)


Gaming Performance

From the above graph of Quake III scores, we can see that the difference between the fastest and slowest board is about 7%, a relatively small gap, but one that's larger than the one we found in our November 2000 KT133 motherboard roundup. This is probably due to the fact that some manufacturers haven't tweaked the KT133A chipset to the fullest or maybe even the fact that some manufacturers redesigned their PCB's while others reused KT133 designs.

Business Performance - Content Creation

When dealing with office applications, we once again see a 7% differential between the fastest and slowest boards.

Business Performance - SYSMark 2000

With SYSMark 2000, the gap between the boards is down to 3%, probably because SYSMark 2000 is not as stressful as Content Creation Winstone 2000. Interestingly enough, the Soyo K7VTA Pro, which was one of the slowest in Content Creation Winstone 2000, jumped to second in SYSMark 2000. On the other hand, Soltek scores the worst in both business benchmark suites.



The Winners and Conclusion

This roundup is very different from the one we had a little over six months ago with the KT133 chipset. After almost one year of release, the KT133 chipset and its successor, the KT133A, have become very mature chipsets. Manufacturers have had quite a bit of time to familiarize themselves with all the ins and outs and are able to squeeze the most out of them.

The result is a very close race at the top of the list. Both the ASUS A7V133 and the Iwill KK266 are excellent examples of what a truly exceptional KT133A motherboard should be. The ASUS A7V133 is excellent for its fully jumperless design, while the Iwill KK266 shows us the true limits of extreme overclocking.

Moreover, both boards performed extremely well also, scoring first and second in most of our benchmarks. They are also of very high qualities at a relatively reasonable range of prices. Both boards offer an optional onboard IDE RAID controller, although the Iwill KK266 stands out a bit here by supporting RAID 0, RAID 1, and RAID 0+1.

One thing we want to stress is that many of the other boards here also placed just behind our two winners. The ASUS and Iwill boards were just able to offer all the small features to put them slightly ahead of the others, and that's the why we've chosen them as the winners. Most other contenders are still delivering very nice features and performance at a decent price.

It is just impossible for us to give out a single Editor's Choice Award with this roundup, while at the same time we can't give the award to too many boards either. The ABIT KT7A-RAID, ASUS A7V133, Gigabyte GA-7ZXR (2.2), Iwill KK266, and MSI K7T Turbo are all very good boards. Each of them has their own pros and cons and we're going to leave it up to you to pick your own winner.

The fact that so many boards are so good may actually be a good thing since you will have more flexibility in choosing a board that fits your needs, rather than wasting time finding a board that works as you expect it too. At the same time, if your favorite board is not available to you, it should be easy to find a suitable alternative. From our perspective as hardware enthusiasts, we recommend choosing between our two winners, the ASUS A7V133 and the Iwill KK266, if possible.

Log in

Don't have an account? Sign up now