Web Hosting by Netfirms | Free Domain Names by Netfirms

BIOS
It looks like one of 8KHA+ board and this is not a surprise because they are both based on Award BIOS v6.00 PG.

There are no changes in Magic Health technology. Detailed system monitoring information is displayed on startup screen.

Regarding memory setup, BIOS on Epox boards allows the biggest number of fine-tuning options.

There is an item called "System Performance" for inexperienced users. It allows setting timings automatically from the most reliable mood (and the slowest one) till the turbo mood (at which not every type of memory will work).

Other BIOS sections remained unchanged, e.g. AGP bus setup section.

Overclocking and stability

Virtually all the reviewers noted very high stability of the 8KHA+ board. They were amazed by the fact that Epox had recommended physical removal of a condenser without any negative consequences in order to provide compatibility with some GeForce4 videocards. That is why 8K3A stability arouses very special interest. Will our board demonstrate the same level of stability or not? Remembering high delicacy of Epox boards to the quality of power supply unit, I have assembled my system with the PowerMan 250W unit. While briefly examining the board, I have counted 12 condensers with the capacities of 2200uF each. 3-phase power supply circuit is used.

Some elements are placed from the back side of the board. The same engineers' dodge was used in 8KHA+ and apparently will be used in future Epox models.

This is a regular headache for assemblers, because the board should be installed with great care.

So the 8K3A should not yield to 8KHA+ in stability. And so it does. After I have found a stable FSB clock the board started working without failures.

And now few words about overclocking. My previous configuration was Epox 8KHA+ and Duron 1071 MHz. The memory was working in synchronous mode with Turbo timings.

I have carried out a testing of 8KHA+ with only one memory module of good quality. As the result, FSB clock reached 175MHz. This value is comparable with maximal overclocking of Abit KR7A.

How I was disappointed when the system refused working at either 150Mhz or 145Mhz (timing setup did not improve the situation). Stable work was achieved at 143MHz. "There must be something wrong": the tester thought and started searching in the Web. As the result I have found some information about a particular range of frequencies at which some processors don't work. The range of 140-160MHz and Duron and Athlon (Thunderbird) are referred most frequently. The situation with Athlon XP seems to be very indefinite. Sometimes it works at this clock range, sometimes it doesn't. Also I haven't found any information about Durons with Morgan core.

So to determine the range I unblocked Duron's clock multiplier and started testing.
I've found that from 166MHz the system not only starts but also works stable. The further FSB increase has ended at 180MHz, on higher clocks the system was working with failures.

Finally I note excellent results of overclocking. And now let's take a look at overclocking tools we've got from Epox engineers.
Firstly, no changes were made since 8KHA+ with clock multiplier.

As before, the multiplier can be changed in the range from 6 up to 15 with a step of 0.5.

The next point is FSB clock variation.

With BIOS setup we can set the frequency from 100MHz up to 255MHz with a step of 1MHz. The progress is obvious, the top value for 8KHA+ was 200MHz. Actually, 255MHz is an absolute record among all the SocketA boards. Before this, the maximal possible value was 227MHz which could be reached on Asus A7V266 E. The record was beaten due to ICS94228 clock generator which is used on Epox 8K3A.

The setting of so high frequencies is preferable on 8K3A because of FSB:PCI divider which equals 5:1. If FSB clock is set to 166MHz, the PCI will work on its standard frequency of 33MHz.

There are some more reasons why 166MHz is more preferable. In this case, memory will work synchronously with CPU bus and there will no be delays with signal synchronisation. Also, the user doesn't wait until AMD decides to switch their processors to 166MHz. As the result, bus transfer ratio grows that justifies the use of PC2700 memory.

I told about this in chipset description section.

Stable work at 166MHz implicates unblocking of multiplier and Vcore increase in major cases. We can see a definite progress here. Now the maximal value of Vcore equals 2.2V, which is also a record in some way. I remind that some overclockers had to modify 8KHA+ board with a soldering iron to set Vcore to 2.0V.

So Vcore range is the following: from 1.4V up to 1.85V with a step of 0.025Vand from 1.85V up to 2.2V with a step of 0.05V.

By the way, due to the increased range of Vcore I have succeeded in overclocking my Duron up to 1165MHz (7x166), at this Vcore was set to 1.9V. Furthermore, my processor successfully passed POST at 1204MHz (7x172), which indicates possibilities of further overclocking.

If you take a look at wcpu program screenshots you can see that Epox 8K3A slightly overstates FSB clock on 0.7-1.0MHz.

We can also change memory voltage in the range of 2.5V up to 3.2V with the step of 0.1V.

I have to note a very point: "Watching Dog Timer". If this option is set to "Enable" position, after FSB is set to a value at which your board (or processor) cannot work, the computer reboots and all of the BIOS options are reset to default values. Similar technologies were already used on Abit's and some other manufacturers' boards, but I see this thing on an Epox board in a first time.

Unfortunately, AGP and chipset voltages cannot be increased.

So my first conclusions will be the following. Despite Epox 8K3A outgoes 8KHA+ in some parameters (increased Vcore and FSB range, 1/5 divider), these boards will occupy very different segments of market. This will happen because Epox 8K3A works unstable in the range from 143MHz up to 165MHz (on my computer it doesn't work at all). That's why the basic criterion of choosing a particular board among those two is the type of processor. iiThat is if you firmly decided that your processor will work on FSB clock over 166MHz then you should remember that Duron and Athlon (Thunderbird) can be unblocked easily (in some earlier versions of Thunderbird the multiplier is not fixed). In this case you won't have any problems with reaching 166MHz. The situation is different if you have Athlon XP processors (as we know more productive processors give more outcome with DDR memory). In this case multiplier change is rather laborious operation.

The fact that processor will be overclocked is implied, because otherwise there is no any meaningful difference between Epox 8K3A and 8KHA+.

• Page1 - Introduction
• Page2 - Layout
• Page3 - BIOS And Overclocking
• Page4 - Testing
• Page5 - Conclusions