EPoX 8KTA3+

Hard Drive Performance

Hard drive performance is something we’ve been looking at a lot lately, trying to find out which ATA/100 controller (ICH2, 686B, HighPoint370, Promise, etc.) actually yields the best performance. In this case, all hard drive tests here performed with the Athlon CPU at 12x100 but the memory settings optimized.

The Sandra Hard Drive test with the 30GB, 5,400RPM, ATA/100 Maxtor Hard drive on the 686B-controlled IDE 1 gave a score of 19406, which is pretty representative of what I've seen from the 686B. The more explicit HD Tach gave the following results:

Look at that CPU utilization: 44 percent!  With the same hard drive on the HighPoint 370A-controlled "UDMA1" port gave a slightly lower Sandra hard drive score of 18628, but this HD Tach readout:

That one deep dip and the lower overall sequential read/write speed notwithstanding, I think the better random access time and read burst speed, and especially the better CPU utilization make this the preferable ATA/100 option here.

RAID

With two 30GB 5,400RPM ATA/100 Quantum hard drives as the master and slave on "UDMA1," I set up a RAID 0 array.  (I am always amazed at how easy this is to do.  Just press Ctrl+H to enter the HPT370 BIOS and follow the on-screen instructions.  It takes about one minute, and you barely even need to read the manual.)  The Sandra score for the array was a whopping 23390, and here's the HD Tach:

There's definitely a performance increase here.  Though it sure isn't in the form of read burst speed, it may just be worth it to shell out for a duplicate hard drive and use the RAID capabilities of this board.

686B Bug

While I was working with the 8KTA3+, the people at this website discovered that the 686B south bridge was causing large files (several GB in size) to be corrupted when copied from one hard drive on an IDE channel controlled by the chip to another. Try as I might, I was not able to duplicate this problem. However, EPoX tech support was very conscientious about getting out a BIOS that incorporated VIA’s offical fix, and in fact were one of the very first companies to do so.