Duron Is The New Star
Today AMD offers the best low-cost processor with its Duron. It outperforms Intel’s Celeron by far and it is easy to overclock. The new Athlon for Socket A has no reason to hide either. At the same clock frequency it is still cheaper than its’ cousin Pentium III.
You need the right motherboard for AMDs new Socket A CPUs though. At the moment the installed base is still rather small. For this reason we checked the features and performances of 10 Socket A motherboards with VIA’s KT133 chipset.
Let us mention one thing right at the beginning: Not every motherboard is suitable for easy overclocking, but overclocking friends should have fun with our test as well.
How Expensive are Socket A Boards with KT133 Chipset?
Many test magazines take a fairly one-sided look at the price factor “CPU” forgetting to take the cost of a motherboard into account as well.
Platform | Prices |
Intel 440BX, Socket 370 | up from $65 |
VIA Apollo Pro 133A, Socket 370 | up from $75 |
VIA KT133, Socket A | up from $95 |
Intel 815/Solano, Socket 370 | up from $120 |
The prices for Socket A motherboards have come down a lot in the past weeks. Especially the combination Duron/KT133 has become a quite interesting option. Despite its lower price is AMD’s Duron significantly faster than Intel’s Celeron at the same clock frequency. The affordable AMD processor achieves this mainly with a faster Front Side Bus (FSB) of 100 MHz (200 MHz DDR). On the Celeron the FSB is specified to 66 MHz. Additionally Duron profits from its supreme floating-point performance. Last but not least Duron is also a very good choice for overclocking friends. With a few tricks this CPU can be overclocked to up to 150 % of its specified clock.
Current Processor Prices
Until recently the general opinion was that Socket A systems for AMD CPUs were more expensive than Socket 370 systems for Intel processors. It is easy to see why, if you just compare the official retail prices in Dollars published by Intel and AMD for quantities of 1000 units. Street prices are often quite different however. Listed below are the processor prices we found at large online retailers like Price Watch. We chose the currently most popular segment for clock frequencies: 700 MHz.
Low-cost CPUs | Prices |
AMD Duron 700, Socket A | up from $90 |
Intel Celeron 700, Socket 370 | up from $150 |
Mainstream CPUs | Prices |
AMD Athlon 700, Socket A | up from $125 |
Intel Pentium III 700, Socket 370 | up from $175 |
If you add up motherboard and CPU prices you’ll notice that the combination Duron/KT133 motherboard offers the best price/performance ratio today. Using an Athlon is also more cost effective than using a Pentium III.
Manufacturers of Socket A Chipsets
The chipset selection for AMD’s socket processors Athlon and Duron is quite meager, as there’s currently only AMD’s 750 and VIA’s Apollo KT133 chipset. Only a few manufacturers – e.g. Gigabyte’s GA-71XE – intend to use grandfather Irongate (AMD750). Most other board makers despise Irongate, since its restrictions to PC100 SDRAM and AGP 2x are not up-to-date anymore. VIA’s KX133 chipset turned out to be unsuitable. Initially it was intended for Slot A and Socket A, but in practice the AMD CPUs only ran stable up to 800 MHz. Consequently the manufacturer dropped the KX133 for Socket A and thus you won’t find any Socket A boards with the KX133.
While there is a variety of chipsets for Intel’s socket processors (Pentium III and Celeron), including 440BX, 810/Whitney, 815/Solano, 820/Camino, 840/Carmel, and VIA’s Apollo Pro 133 (A) as well as models from SiS and Ali, VIA’s KT133 is pretty much the only alternative for Thunderbird and Duron. Two months ago most SocketA-motherboards only existed as samples, and only a few users were able to get their hands on a board back then. Meanwhile VIA is shipping a sufficient amount of KT133 chipsets to the motherboard manufacturers and customers can chose from a wider selection of boards. The current situation with practically only one chipset for Socket A motherboards, VIA’s KT133, should change soon. As you can see in the following table, AMD, VIA, SiS and ALi are developing new chipsets for AMDs Socket A processors.
Chipset | Alias | System Bus 1 | AGP | Integrated graphics | SMP | Memory | Available |
AMD750 | Irongate | 100 MHz x 2 | 2x | no | no | PC100 SDRAM | yes |
VIA KT133 | VT8363 | 100 MHz x 2 | 4x | no | no | PC133 SDRAM or VCRAM | yes |
AMD760 | 133 MHz x 2 | 4x | no | yes | PC133 SDRAM or PC2100 DDR SDRAM | Q4/2000 | |
VIA KM133 | VT8365 | 100 MHz x 2 | 4x | yes | no | PC133 SDRAM or VCRAM | Q4/2000 |
VIA KX266 | VT8367 | 133 MHz x 2 | 4x | no | yes | PC133 SDRAM or PC2100 DDR SDRAM | Q4/2000 |
VIA KM266 | VT8367 | 133 MHz x 2 | 4x | yes | no | PC133 SDRAM or PC2100 DDR SDRAM | Q2/2001 |
SiS730S | 100 MHz x 2 | 4x | yes | no | PC133 SDRAM | Q4/2000 | |
ALi MAGiK 1 | M1647 | 100 MHz x 2 | 4x | no | no | PC133 SDRAM orPC2100 DDR SDRAM | Q4/2000 |
1All new socket processors from AMD use the EV6 protocol. The real clock is 100 and 133 MHz, respectively. Double Data Rate offers an effective bandwidth of about twice the clock frequency (200 and 266 MHz, respectively). The Athlon and Duron CPUs currently available are specified for a ‘double-pumped’ system bus clock of 100 MHz.
The Features of VIA’s Apollo KT133
The Northbridge of the KT133 is the VT8363. This chip handles communication with the processor, memory and the graphics card (up to AGP 4X). VIA specifies 100 MHz as the upper limit for the system bus. In practice the VT8363 can deal with a much faster Front Side Bus. Four of the motherboards we tested offer overclocking functions that allow increasing the clock to over 150 MHz. Such settings are a total overkill however. Even though the Northbridge does not have any problems with beyond-spec FSB-clocks, no AMD processor is currently able to handle more than 110 MHz. So to say, most motherboards are equipped for the future, as AMD plans to offer processors with 133 MHz Front Side Bus soon.
The KT133, or VT8363 respectively, allows an asynchronous memory clock. For example at 100 MHz Front Side Bus the memory clock can be set to 133 MHz, which generates a higher performance from PC133-SDRAM modules.
The Southbridge of the KT133 is an old acquaintance. You can find the VT82C686A on Pentium III/Celeron compatible motherboards with the VIA Apollo Pro 133 and 133A as well. Besides the AC97 sound compatibility the VT82C686A also allows the UltraATA/66 mode. VIA is slightly behind with this Southbridge. Intel already offers chipsets with UltraATA/100 support. VIA plans to introduce the pin -compatible VT82C686B shortly, which should also come with UltraATA/100 functionality.
I) AMR-Slot
One year ago nobody really knew about those small AMR slots that are on many motherboards now.
AMR stands for Audio Modem Riser. This slot is for a card with an audio codec chip and/or a modem circuit. At least if Intel has its’ way, but Intel’s AMR-Specification almost resembles suicide note for manufacturers and users. The first page of the spec contains the important sentence:
Note that the AMR Specification does NOT define an aftermarket standard I/O expansion slot. |
Intel is shooting itself in the foot with this note because there are no plans to offer AMR cards for the retail market. The manufacturers are not interested in it either. Theoretically the AMR cards are supposed to be an “affordable solution”, but no manufacturer can offer a cheap AMR card because the costs do not leave any room for adequate support. The only way out is to buy a regular modem and sound card. Of course the end user gets the short end of the stick again. The additional manufacturing costs for the AMR slot are handed down to the customer. He has got to spend money for a useless slot. It is a similar situation with the so-called CNR slots that are found on 815/Solano motherboards. It is supposed to allow network functionality besides audio and modem, but no enduser will be able to take advantage of it. In our evaluation of the motherboards we do not criticize any manufacturer if his board lacks an AMR slot, so he can offer his product at a lower price.
II) Infrared Interface
It would be really nice if it were possible to have a wireless connection between your PC and your PDA or notebook. The infrared interface is the ideal solution because most laptops or PDAs come with a built-in IR interface. Transferring data would be much easier. Unfortunately users have to fight with the cable collection at the back of the PC instead, with the motto: “Don’t touch anything because you might pull off a cable by mistake.” We are annoyed because not even a single manufacturer includes an infrared connection kit with his motherboard for integration to the front of the PC. The user has to buy it from an electronics store and does not even know whether that specific infrared kit works with his motherboard.
III) Second USB Connector
Many people do not even know it. Every motherboard we tested offers a second USB-connector on the motherboard besides the two external USB ports on the ATX connector (picture above). If you are a USB fan and have already connected keyboard and mouse to this interface, you quickly run into problems if these devices do not allow a pass-through. The number of USB peripherals is growing! More and more digital cameras, monitors or audio systems have these interfaces. Asus is the only manufacturer in this comparison that offers three additional USB ports on his A7V motherboard.
Problem I: Early BIOS of the AOpen AK33
All motherboards worked immediately with a new installation of Windows 98 or 2000 and the driver update to VIA’s All-in-One-Version 4.23, apart from the AOpen AK33. We observed only low frame rates under Windows 2000. The reason: Two numbers of the device ID of the KT133 chipset had been switched in the BIOS.
Note the second number just before the ‘PCI-PCI bridge’ in the picture above. The AK33 reports “0583”. This is the wrong value for the KT133, and so Windows 2000 will not activate the AGP driver. This is irrelevant under Windows 98, however, because the device ID does not matter. Therefore Windows 98 users will not encounter this problem.
This is the correct version: The KT133 chipset reports “8305” to the operating system. We contacted AOpen and promptly received a new BIOS that you can download from our FTP server. You only need an update if the boot screen shows a date before August.
Problem II: VIAs Busmaster Driver
VIA greatly improved the installation procedure for the drivers. As we already mentioned in the previous paragraph there are no more problems with the initial installation of the 4-in-1 drivers. We tested an update of the Busmaster driver from version 580_2147 to 580_219. Windows always asks for a file catalog that is not in the driver collection.
The file viadsk.cat simply isn’t there. It’s actually not needed! Pressing the Skip File button (three times during the installation) remedies the problem. The UDMA mode can be activated later without any negative consequences. The following scenario creates the same problem: Some BIOS updates change the resources of the motherboard. In this case Windows switches to the “Winstupid” mode, which is similar to temporary amnesia. For example the hardware detection starts up and distributes the resources again. In our tests VIA’s Busmaster driver often made problems. On many occasions the driver switches to the lame PIO mode without any warning. You’ve got to use VIA’s IDE DMA tool and don’t forget to check with e.g. HD Tach of TCD-Labs. It should show throughputs of way above 16 MB/s. Sometimes the DMA Tool has to be switched to PIO mode to actually get DMA mode.
AC97 versus PCI – The Audio Functions of the Motherboards
Nine of the ten motherboards we tested came with audio functionality. Most are equipped with only a small codec chip. For those the processor has to divert some computing power to the audio part, at least in theory.
The Chaintech 7AJA and the IWill KV200-R use the PCI sound chip CMI8738. It requires its’ own interrupt. Contrary to a cheap codec chip, the CMI8738 allows assigning functions to the inputs and outputs on the mentioned motherboards. For example it is possible to play DVD videos over SPDIF with 3D sound. All it takes is ‘reprogramming’ one audio connection (usually line-in).
Gigabyte uses a sound chip from Creative on the GA-7ZX. Similar to the CMI8738 it needs an interrupt. If you install the audio drivers you will notice that it is the same chip that sits on the Soundblaster PCI128 sound card. To enable 3D sound you can connect four speakers to the motherboard.
In this case line-in becomes the second lineout if the function is activated. Despite the more popular brand name the Creative chip offers the same performance as the CMI8738. We compared the two PCI sound chips with the codec chips, and we expected that the PCI chips do not strain the CPUs as much. In the test we had to find out that the differences between PCI sound and audio codec are minimal. The codec chips are definitely sufficient for regular stereo sound (without 3D effects). The PCI chips CMI8738 and Creative CT5880 offer more features however.
ABit KT7
BIOS-Version: July 28, 2000
Board Revision: 1.01
The KT7 follows the long ABit tradition and comes with all kinds of gimmicks for overclocking. The so-called Soft Menu III combines all necessary functions like settings for FSB clock, multiplier (up to x12.5) and voltage supply. For successful overclocking all L1 bridges on the processor must be closed. One simple but effective tool is a graphite pin because the material is a good electric conductor.
This board is also available with UltraATA/100 RAID controller. ABit uses the controller chip HPT370 from Highpoint. This type of board is called KT7 RAID and costs a little bit more.
There is not much to criticize about the layout, just that the DIMM clips obstruct the graphics card. Therefore the clips must be closed before installing a graphics card. Quite unusual is the fan on the Northbridge chip that is supposed to keep the temperature of the chipset component in a safe range even during overclocking.
Unfortunately the board does not possess a sound codec, and one of the PCI slots must accommodate a sound card. However, since the board has 6 PCI-slots a lack of resources should not be a problem.
Loading the BIOS with “Optimized Defaults” causes a peculiar AGP mode setting. In this case AGP only runs in 2x mode. This does not affect the general functionality but the graphics performance may suffer a little.
AOpen AK33
BIOS-Version: r103 August 16, 2000
Board Revision: 1.01
We really liked AOpen’s Website when we searched for a current BIOS. The pages are well structured, clear and can be viewed in different languages.
The BIOS version from July 21 generates an incorrect device ID for the AGP of the KT133 chipset. As we already mentioned this causes Windows 2000 to address the graphics card without the AGP driver, and the graphics card only runs at PCI level. To avoid this problem you should load the latest BIOS version 1.03.
Five PCI slots and three DIMM sockets should be sufficient for most applications. The codec chip from Analog Devices enables an AC97 compliant sound system.
An interesting feature is the so-called Dr. LED. Eight LEDs permanently display the system status. Especially during booting this function can quite often act like a debug card and simplify the search for errors significantly.
Luckily AOpen decided against the AMR slot. Appropriate cards are hard to come by anyway. To make everything picture perfect the board only needed the function Wake On Modem that should wake the system from sleep mode for an incoming call. Similar to Chaintech and Epox AOpen specifies a maximum memory of 768 MByte even though the chipset allows twice as much.
Asus A7V
BIOS-Version: 1003 (July 21, 2000)
Board Revision: 1.01
Asus is the only manufacturer in our test that includes an adapter cable for the three additional USB ports, so you can actually use them. The shipment also contains a software program called Asus Probe, which allows monitoring the system status (fan, CPU and casing temperature, voltages) under Windows quite comfortably.
This board also possesses a sound system that works with a AC97 codec.
The board is also equipped with an UltraATA/100 controller chip from Promise. If you want to use a hard disk with pre-installed Windows NT 4.0 or Windows 2000 you must connect it to the standard IDE controller first. Afterwards you install the driver for the fast controller chip to enable Windows NT or 2000 to address the new controller. Now you can connect your hard disk to one of the UltraATA/100 channels.
The Asus board outperforms all other competitors. Generally there are two board versions but they cannot be distinguished by the version number. One comes with DIPswitches and one without. Today you will mainly find the version with the DIPs in retail stores. This one is one of the best platforms for overclocking AMD processors. If you close the L1 bridges on the Duron or Athlon with a SMD soldering iron or with a simple graphite pin, you can use the DIPswitches and often get more performance from the CPU.
If you want to know how to upgrade an A7V without these DIPswitches, please read our separate article: Overclocking with the A7V.
Chaintech 7AJA
BIOS-Version: August 2, 2000
Board Revision: S102
While we were looking for a more current BIOS once again, we discovered information on Chaintech’s Website stating that the board is available with the VIA Southbridge 686B and UltraATA/100. This was not the case with our test sample. It works with the 686A – just like all other candidates.
Chaintech equips the 7AJA with two ROM chips. If the main BIOS is corrupted by a virus or an unfinished update procedure, the board can boot from the backup BIOS. This is a very effective way to guarantee that the system will always work.
The sound system is based on the PCI chip CMI8738 from C-Media that supports AC3 instead of simple codecs. The board has a digital input and output (SPDIF), and offers one internal audio input – a second input would be nice.
The menu option “Frequency/Voltage Control” in the BIOS is for overclocking. Unfortunately it only lets the user change the clock frequency but not the voltage supply of the CPU.
DFI AK74-SC
BIOS-Version: August 4, 2000
Board Revision: A
DFI’s nomenclature is quite similar to AOpen: Both the AOpen AK33 and the DFI AK34 use the KT133, AOpen AK72 works with the KX133, and the DFI AK74 with the KT133 again. Obviously this is very confusing, especially for the customers. Ordering hardware by mail-order for example will almost guarantee mistakes.
The AK74-SU uses Creative’s sound chip ES1373. We ‘only’ tested the smaller version AK74-SC that is based on a codec.
Starting at the initial value the CPU voltage can be changed in 0.025 Volt steps with a maximum of 0.100 V. The system clock can be set to any value between 100 and 132 MHz, and the frequency is entered directly as a number. All settings are done in the BIOS, so you do not have to deal with the pesky jumpers anymore.
The board comes with three internal audio line-ins allowing connecting a TV or an MPEG decoder card to the sound system without any problems.
Epox EP-8KTA
BIOS-Version: August 14, 2000
Board Revision: 1.2
The Epox 8KTA had a few problems during our test. After mounting a CD Windows 98 crashed completely, but Windows 2000 ran without any flaws. The error can probably be eliminated with a new BIOS.
The AGP slot is quite a different story because installing the graphics card requires a certain amount of skill. In other words: If the card does not fit exactly it does not work.
Only ABit and Epox offer six PCI slots. Both also offer an ISA slot – in case you have an older card. The processor voltage can be adjusted between 1.3 and 1.85 V via DIPswitches.
The rest (FSB) is comfortably done in the BIOS. Officially Epox allows a maximum memory of 768 MB. Other boards take advantage of the full potential of the chipset and allow up to 1.5 GB. There was nothing wrong with the performance.
Gigabyte GA-7ZM
BIOS-Version: July 17, 2000
Board Revision: 1.0
The GA7ZM is a Micro-ATX board and the little brother of the 7ZX so-to-speak. Because of its compact size it only has three PCI slots. But the rest of the equipment is very complete: 3 DIMM sockets, AC97 sound system with 2 line-ins and configuration via DIPswitches.
Under Quake III the system crashed a few times. Since it was not possible to reproduce these crashes exactly they are probably caused by the main memory. We exclusively used SDRAM from the American company Crucial for our comparison tests, one of the most renowned manufacturers.
It is possible to increase the system clock up to 132 MHz (1 MHz steps up to 112 MHz), but the CPU voltage cannot be altered.
Micro-ATX boards are mainly interesting for office PCs because they are usually only equipped with just the necessary components.
Gigabyte GA-7ZX
BIOS-Version: July 17, 2000 (F8)
Board Revision: 1.0
The 7ZX offers an excellent sound system that is based on Creative’s CT5880 chip. This chip has all the functions of the Sound Blaster PCI 128. If surround sound with four speakers is selected the line-in becomes the output for the speaker pair in the back. The chip also supports the AC3 format; in this case line-out is changed to SPDIF-out. But there are no Soundblaster PCI128 drivers for Windows 2000 yet. This means that the chip runs in Soundblaster compatible mode but with a limited number of functions.
The board is also available as GA-7ZX-1 without the ISA slot, with only one BIOS chip and without the sound system. The standard version possesses the so-called Dual-BIOS(tm). It has a second BIOS for safety reasons. If the main BIOS fails the backup BIOS takes over.
Gigabyte uses a clip that secures the AGP card after insertion. This feature makes sense if you need to transport your computer a lot. Usually you should get along quite nicely without this feature.
The labeling of the connectors for power-on, reset, power-LED, and so on is not quite clear. Only a look into the handbook clarifies the situation.
IWill KV200-R
BIOS-Version: July 31, 2000
Board Revision: 35100A
Iwill is one of the first manufacturers that plans to ship a motherboard with DDR support based on the ALi MAGiK 1 (KA266-R). Until this actually happens you can fall back on the KV200-R. The R stands for RAID because the board comes with the MG80649 from American Megatrends. Also on board is a sound system based on the CMI8738 from C-Media.
There is a mistake in the labeling of the IDE channels: The two connectors on the outside belong to the VIA chipset, the two on the inside are controlled by the AMI chip. The labeling says the opposite!
According to the benchmarks the board delivers the expected performance but it still lands in last place. Apparently IWill mainly focussed on the implementation of the RAID chip because we discovered another problem. Activating the AGP Fast Writes in the BIOS always caused a crash.
The approach is good but performance and reliability could use some more work.
Soltek SL-75KV
BIOS-Version: August 5, 2000
Board Revision: Q2
We were surprised when we checked the voltage supply for the Athlon 1000. The BIOS showed 1.89 V (under voltage default) – even though the standard value is 1.75 V.
The frequency steps for the system bus are not very transparent. In principle the PLL can be adjusted in 1 MHz steps. But unfortunately this happens with an addition to a base clock by +/- 28 MHz. Both parameters must be changed separately.
Furthermore there is no reason to run the board at the lower clock of 75 MHz because all of AMD’s Socket A CPUs are officially specified with 100 MHz (200 MHz DDR).
Performance and system behavior are satisfactory. The equipment is complete as well because it also offers an AC97 sound system.
Test Configuration
Hardware | |
CPU | AMD Athlon 1.0GHz |
RAM | 1x 128 MByte SDRAM (Crucial) PC133, 7ns, CL2 |
Hard Disk | Seagate Barracuda ATA, ST320430A 20 GBytes, UltraDMA/66, 7200 U/Min |
Graphics Card | Asus V7700 AGP, nVIDIA GeForce2 GTS, 32 MB DDR SDRAM |
Drivers & Software | |
HDD/Mainboard/ AGP-Driver | VIA All-in-One 4.23 (Win 98 and 2000) |
Graphics driver | NVIDIA Reference driver 5.23 (Win 98 and 2000) |
DirectX Version | 7.0 |
OS | Windows 98 SE, Version 4.10.2222 A Windows 2000, Version 5.00.2195, SP1 |
Benchmarks and Settings | |
Quake III Arena | Retail Version command line = +set cd_nocd 1 +set s_initsound 0 Graphics detail set to ‘Normal’, 640x480x16 Benchmark using ‘Q3DEMO1’ |
Expendable | Downloadable Demo Version command line = -timedemo 640x480x16 |
Sysmark 2000 | Patch 4B 1024x768x16 |
ViewPerf | Version 4.1.2 1280x1024x16 |
Refresh rate | 85 Hz for all tests, V-Sync = off |
Sysmark 2000
The Asus A7V dominates, significantly under Windows 98, only slightly under Windows 2000. It is always astounding how regularly Asus is at the top of our benchmark charts. The Chaintech board clearly loses ground under Windows 2000, all other candidates are close together.
Quake 3 Arena Demo001
Asus and AOpen lead the Quake benchmark, IWill is in last place. ABit’s KT7 ends up in the middle of the field.
Expendable
Viewperf 6.1.2 AWadvs-04
The Advanced Visualizer kind of shows the sequence of the game benchmarks. The Gigabyte boards look significantly better while the Abit KT7 ends up in last place.
Viewperf 6.1.2 DRV-07
In Design Review the middle field is composed slightly different. Again, Asus is in first place. In a motherboard comparison the SPECviewperf benchmark is not used to determine winners and loser but more for detecting instabilities. If the system has weaknesses the viewperf benchmark will crash under Windows 98 and Windows 2000.
Viewperf 6.1.2 DX-06
The winner in Data Explorer is the Soltek motherboard, but the distances between the individual candidates are very small. Unfortunately the ABit clearly falls behind again.
Viewperf 6.1.2 Light-04
Again ABit delivers the worst result. The Soltek board SL-75KV is at the top with a small lead.
Viewperf 6.1.2 MedMCAD-01
MCAD also only shows minimal differences.
Viewperf 6.1.2 ProCDRS-03
The first and only time the Asus A7V is not able to get one of the top spots and lands in the middle field. In this benchmark the performance differences are more significant than in the previous disciplines. The AOpen board does also not do as well as in the game benchmarks and only comes in second to last.
Overclocking
The overclocking features are important characteristics to differentiate motherboards, and many manufacturers have already recognized this. Only about half of the companies master it satisfactory. Two conditions must be fulfilled.
a) Adjusting of the Front Side Bus (system bus) in 1-MHz steps
The current Athlon and Duron processors are officially specified for a system clock of 100 MHz. Our tests showed that most processors can take up to 110 MHz. Therefore the motherboard should allow adjustments in 1 MHz steps, especially in the range between 100 and 110 MHz. Values that are above or below this range are only valuable for marketing purposes.
b) Adjusting the core voltage (VCore) in 0.05 Volt steps
For successful overclocking the motherboard must also offer the option of at least 0.05 Volt steps. Increasing only the Front Side Bus can quickly lead to instabilities. Only if the voltage is raised slightly the system becomes stable again.
Here are the standard-voltages for available Socket A CPUs:
Socket A CPU | Standard Core Voltage (VCore) |
Duron 600 – 700 | 1,50 V |
Athlon 700 – 850 | 1,70 V |
Athlon 900 – 1100 | 1,75 V |
No motherboard offers more than 1.85 Volts. This certainly makes sense because otherwise the processor could be destroyed very easily. Be careful when you increase the voltage of Duron. 1.85 Volts can quickly become dangerous. The motherboards Abit KT7, Asus A7V, DFI AK74-SC, Epox EP-8KTA and Soltek SL-75KV offer 1 MHz settings and VCore settings.
Overclocking Masters: Abit and Asus
Most overclocking friends had already dismissed the possibility of changing the multiplier of modern processors, after the Mendocino Celeron 300 Intel and AMD had the multiplier set to a fixed value (multiplier lock). The user was only able to change the Front Side Bus. As we already described in the installation guide for the Asus A7V, it is possible to break the multiplier lock in Socket A CPUs. A good example is the Duron 700 that often can be overclocked to 950 MHz. At a FSB of 100 MHz the multiplier is changed from x7.0 to x9.5. The core voltage has to be increased as well, of course. Currently two Duron and Athlon version are available:
Here the L1 bridges on the processor surface are open.
These are the better overclocking CPUs – the L1 bridges are closed. You can close the open L1 bridges on the first CPUs fairly easily. Silver conducting paste does the trick, and it is available in electronics stores. With a little bit of skill everybody can do that.
Let’s now take a look at our favorite overclocking boards. If you insert a Socket A CPU with closed L1 bridges in the Abit KT7, you can set any random multiplier in the Softmenu of the BIOS. The Front Side Bus can also be changed. The Abit KT7 is clearly our favorite. As a second candidate the Asus A7V is quite convincing. Here you should note that there are two retail versions available. The “old” revision does not have multiplier DIPswitches. For a few bucks you can solder the DIPswitch and IC-chip onto the board with a soldering iron, following the instructions in our guide. Then you can change the multiplier again, but only if the L1 bridges are closed. After our installation guide was published Asus decided to ship all new A7V boards with DIPswitches for the multiplier again, thus allowing overclocking. For the customer it is difficult to differentiate between both A7V board versions. The exact location of these DIPswitches is also described in the article.
Conclusion
The winner of the test is the Asus A7V. This board convinces with its excellent performance, leaving all competitors far behind. The updated version comes with DIPswitches that allow changing the multiplier of Socket A CPUs (with closed L1 bridges). This makes this board ideal for overclocking. The A7V also possesses an UltraATA/100 controller that drives newer hard disks to peak performance.
A secret tip for overclocking fans is the Abit KT7. The multiplier can be changed in the Softmenu of the BIOS if the L1 bridges on the Athlon or Duron are closed. The KT7 is also available in a more expensive version with IDE RAID controller. For what it’s worth, the KT7 lacks sound functionality. The AOpen AK33 and the Soltek SL-75V are solid products in the medium range. Both boards are well-equipped, run stable and offer good performance. Regarding performance the IWill KV200-R gets last place. It couldn’t convince in this category.
Equipment Table, Part I
Manufacturer | Abit | AOpen | Asus | Chaintech | DFI |
Product | KT7 | AK33 | A7V | 7AJA | AK74-SC |
Platform | Socket A | Socket A | Socket A | Socket A | Socket A |
Chipset | VIA KT133 | VIA KT133 | VIA KT133 | VIA KT133 | VIA KT133 |
Form Faktor | ATX | ATX | ATX | ATX | ATX |
Memory | |||||
Max. Memory (MByte) | 1536 | 768 | 1536 MByte | 768 | 1536 |
# DIMMS | 3 | 3 | 3 | 3 | 3 |
Slots and Interfaces | |||||
# ISA slots | 1 | 0 | 0 | 1 | 1 |
# PCI slots | 6 | 5 | 5 | 5 | 5 |
AGP 4X | 1 | 1 | 0 | 1 | 1 |
AGP 4X (Pro) | 0 | 0 | 1 | 0 | 0 |
AMR (Audio Modem Riser) | 0 | 0 | 1 | 1 | 1 |
parallel | 1 | 1 | 1 | 1 | 1 |
seriell | 2 | 1+1 | 2 | 2 | 2 |
USB ports external | 2 | 2 | 2 | 2 | 2 |
USB slots internal | 1 | 1 | 1 | 1 | 1 |
# IDE connectors (UltraATA/66) | 2 | 2 | 2 | 2 | 2 |
Fan connectors | 4 | 2 | 4 | 2 | 3 |
Integrated (Onboard) Features | |||||
AC97 Sound | no | AD 1885 | Cirrus CS4299 (Codec only) | CMI8738 PCI | ICE1232 |
Audio external: Line Out/In/Mic |
no/no/no | yes/yes/yes | yes/yes/yes | yes/yes/yes | yes/yes/yes |
Audio internal: Line-Ins |
no | 2 | 3 | 1 | 3 |
Wake on LAN (WOL) | yes | yes | yes | yes | yes |
Wake on Modem (WOM) | yes | no | yes | yes | yes |
Infrared | yes | yes | yes | yes | yes |
Intruder sensor | no | no | yes | yes | no |
extra UltraATA/100 IDE connectors | no | no | 2x Promise PDC20295 | no | no |
Piezo speaker onboard | no | no | no | no | no |
Frequencies and Settings | |||||
Freq. adjustable with | BIOS | BIOS | DIP or BIOS | BIOS | BIOS |
FSB for DDR system clock | 100 – 155 MHz | 66 – 147 MHz | 90 – 145 MHz | 100 – 160 MHz | 100 – 132 MHz |
Freq. adjustable in 1 MHz steps | 100 – 155 MHz | no | 100 – 111 MHz | 106 – 112 MHz | 100 – 132 MHz |
CPU core voltage adjustable with | BIOS | no | Jumper or BIOS | no | BIOS |
CPU core voltage | 1.10 – 1.85 V | n/a | 1.10 – 1.85 V | n/a | 1.10 – 1.85 Volt |
CPU core voltage adjustable in 0.05V steps | yes | n/a | yes | n/a | yes |
BIOS and Revisions | |||||
BIOS | Award | Award | Award | Award (Dual) | Award |
BIOS version | 28.7.2000 | 14.8.2000 | 1003 | 2.8.2000 | 4.8.2000 |
Board revision | 1.01 | 1.01 | 1.01 | S102 | A |
Comments | |||||
Clock Controller | Phaselink PLL205-01XC | Winbond W230 | ICS94215AF | Winbond W230 | ICS94215AF |
Product variations | KT7 RAID with Controller Highpoint HPT370 | no | also without AC97-Sound available | no | AK74-SU with Creative ES1373 |
Equipment Table, Part II
Manufacturer | Epox | Gigabyte | Gigabyte | IWill | Soltek |
Product | EP-8KTA | GA-7ZM | GA-7ZX | KV200-R | SL-75KV |
Platform | Socket A | Socket A | Socket A | Socket A | Socket A |
Chipset | VIA KT133 | VIA KT133 | VIA KT133 | VIA KT133 | VIA KT133 |
Form Faktor | ATX | MicroATX | ATX | ATX | ATX |
Memory | |||||
Max. Memory (MByte) | 768 | 1536 | 1536 | 1536 | |
# DIMMS | 3 | 3 | 3 | 3 | 3 |
Slots and Interfaces | |||||
# ISA slots | 1 | 0 | 1 | 0 | 1 |
# PCI slots | 6 | 3 | 5 | 5 | 5 |
AGP 4X | 1 | 1 | 1 | 1 | 0 |
AGP 4X (Pro) | 0 | 0 | 0 | 0 | 1 |
AMR (Audio Modem Riser) | 0 | 1 | 1 | 1 | 0 |
parallel | 1 | 1 | 1 | 1 | 1 |
seriell | 2 | 2 | 2 | 1+1 | 1+1 |
USB ports external | 2 | 2 | 2 | 2 | 2 |
USB slots internal | 1 | 1 | 1 | 1 | 1 |
# IDE connectors (UltraATA/66) | 2 | 2 | 2 | 2 | 2 |
Fan connectors | 3 | 2 | 2 | 3 | 2 |
Integrated (Onboard) Features | |||||
AC97 Sound | VIA VT1611A | AD 1885 | Creative CT5880 (PCI) | CMI8738 PCI | WM9701A |
Audio external: Line Out/In/Mic |
yes/yes/yes | yes/yes/yes | yes/yes/yes | yes/yes/yes | yes/yes/yes |
Audio internal: Line-Ins |
2 | 2 | 2 | 1 | 2 |
Wake on LAN (WOL) | yes | yes | yes | yes | yes |
Wake on Modem (WOM) | no | yes | yes | yes | yes |
Infrared | yes | yes | yes | yes | yes |
Intruder sensor | no | no | no | no | no |
extra UltraATA/100 IDE connectors | no | no | no | 2x RAID (AMI MG80649) | no |
Piezo speaker onboard | no | no | no | no | no |
Frequencies and Settings | |||||
Freq. adjustable with | Jumper or BIOS | DIPs | DIPs | BIOS | Jumper or BIOS |
FSB for DDR system clock | 100 – 133 MHz | 95 – 133 MHz | 95 – 133 MHz | 100 – 166 MHz | 72 – 152 MHz |
Freq. adjustable in 1 MHz steps | 100 – 112 MHz | no | no | 100 – 166 MHz | 72 – 152 MHz |
CPU core voltage adjustable with | DIPs | no | no | no | BIOS |
CPU core voltage | 1.30 – 1.85 V | n/a | n/a | n/a | 1.50 – 1.85 V |
CPU core voltage adjustable in 0.05V steps | yes | n/a | n/a | n/a | yes |
BIOS and Revisions | |||||
BIOS | Award | AMI | AMI (DUAL) | Award | Award |
BIOS version | 14.8.2000 | 17.7.2000 (F9) | 17.7.2000 (F8) | 31.7.2000 | 5.8.2000 |
Board revision | 1.2 | 2.2 | 1.0 | 35100A | Q2 |
Comments | |||||
Clock Controller | Winbond W230 | ICS9248AF-141 | ICS9248AF-141 | ICS94215AF | Phaselink PLL205-11XC |
Product variations | EP-8KTM als MicroATX | no | GA-7ZX-1 without ISA, Single Bios and without CT5880 Audio | no | 75KV-X “talks” through Voice Diagnostic |