EPoX 8KHA+ Socket A Motherboard (KT266A) review


Performance & Benchmarking

I will start with measuring relative performance with Sisoft Sandra, this will give you a good idea about how fast the memory performance actually is, and what gains you get by increasing it to the next speed which is also rumoured to have a 1/5 divider, i.e. when you run your computer at 100mhz and it officially supports that speed you are using a 100/3 divider, this is the speed of the PCI bus, i.e. 33mhz, when you move up to the next step your mainboard automatically selects the 1/4 divider (if it has one), and thus you can run your PCI bus at 133/4 which yet again yields 33mhz.

The thing that happens when you move higher up though is that if your mainboard doesn't have yet another speed divider you are going to overclock all PCI peripherals in your computer, this includes Soundcards, LAN, Harddrives CD/DVD ROMs etc, you get the picture...

And while it at first glance might look like a good idea to overclock the "whole" computer infact it isn't, for one, chances that one of your peripherals doesn't support such a high bus speed and thus you will not be able to overclock any more, until you remove the device which is your limiting factor that is... but devices can also take damage from this; for example I got a CD-ROM drive to die once...

So what we want is another divider which kicks in at the next logical speed which is 166mhz, this divided by 5 namely yields yet gain 33mhz, but if that divider doesn’t exist you are going to overclock your PCI bus to 41,5mhz and at this speed very few devices are capable to run at...

I had a problem here with the board, when the FSB was set to 160mhz or above the board would not “soft restart”, (restart when you choose to do so in Windows), forcing me to do a “hard restart” (pressing the reset button on the case). I don’t know what the cause of this problem might be, I just got a blank screen when the computer was supposed to post…

The highest stable speed I reached was 175Mhz, I think this was the limit of the mainboard because up to this speed I could run the RAM at “maximum performance”, but then it just starts to hang when posting, setting the RAM to “most stable” doesn’t help anything… Though it could have been the graphics card too.

Test System


  • CPU AMD Athlon XP 1600 Mhz (133mhz FSB)
    CPUfx’s Z3 watercooler

  • Motherboard Epox 8KHA+
    BIOS “8khi1b02“ date 11/02/2001

  • Memory 1x 256MB DDR-SDRAM, CL2, 133 Mhz TwinMOS

  • Graphics card   3Dfx Voodoo 5 5500 AGP

  • Harddrive   Maxtor Diamond Max Plus 60, 40GB

  • Enermax 431w PSU

Drivers and Software

  • VIA 4 in 1 version 4.35 FINAL, Georges "VIA Latency Patch" v0.14

  • Graphic card driver 3dfx official WHQL 1.04.00

  • DirectX version 8.1

  • Operating System Windows 2000 Professional

  •                            Fixes: "Largepageminimum.reg", Service Pack 2

  • Desktop resolution 1024x768, 16bpp colour, 85hz refresh rate


  • Sisoft Sandra Professional 2001 version 11.8.53, 32bit

  • Quake 3 Arena version 1.30
                       Command line = +set s_initsound 0
    Graphic detail set to “normal”
    Benchmark using “four”

  • Return to Castle Wolfenstein demo 1
    Command line = +set s_initsound 0
    Graphic detail set to “normal”
    Benchmark using “atdemo8”  

  • 3DMark 2001

  • WinRAR version 2.90
                       Compress a 238MB file (d2exp.mpq)

Epox has confirmed that the 8KHA+ infact doesn’t have a /5 divider, this means that the tests I have run above are also run at a PCI clock speed of 41,5Mhz (166/4=41,5Mhz)

CPU host clock set for 100 to 119MHz: PCI divider is 1/3.

CPU host clock set for 120 to 200MHz: PCI divider is 1/4.

As you can see the memory bandwidth is substantially increased, which leads us to believe that a mainboard will infact be developed around this bus speed... KT333 anyone? ;-)

Ok, now lets move on to the real world tests and see if our theoretical bandwidth increases show of in the real world also...


Go to next page !

Get weekly updates on new
articles, news and contests
in your mail!