For Socket A enthusiasts times are, as they say, interesting. Why are they interesting? Diversity. Unlike other platforms, there are a whole raft of motherboar options available to support AMD’s Athlon or Duron processors. Unfortunately, that same diversity that generates more choice in product makes your purchasing decision harder. Well, we are going to have a look at the EPoX 8K7A/8K7A+ and see how it stacks up. This article will use the terms 8K7A and 8K7A+ interchangeably, though the specific motherboard tested was the 8K7A+. This board is functionally identical to the 8K7A except for the fact that the 8K7A+ includes an on-board High Point RAID controller. Since RAID functionality is not part of the review (sorry RAID lovers) it can be (and is) ignored. This is strictly a performance review.
The 8K7A is EPoX’s AMD760/VIA 686B entry into the high performance desktop arena. The AMD760 chipset, though the first available for the Socket A platform, was for a long time the fastest. While other chipsets may (more on this later) have overtaken it, none have done so by much.
The EPoX 8K7A is arguably the fastest of an already fast group. Up against it (so to speak) are the Abit KT7A and the Iwill KA266. Keep in mind when reading this article, that the BIOS settings I use for all contenders are not the factory defaults. To show you how much difference BIOS tweaking can make, I have included scores in a number of tests (towards the end of the article) for a KT7A running with factory “Optimal” settings, with the exception that the memory CAS setting was set to 2 ( the same as the “tweaked” KT7A, and the two DDR boards) and AGP 4X/Fast Writes were enabled. I made this decision to combat “unfair advantage” claims, and it is in my opinion a more apples-to-apples comparison. More on the methodology can be found in the Testing Methods page.
The applications used, their version numbers, drivers and other settings can be found here.
By the same token, stability is paramount. If during testing a system crashed, I tried to analyse why and adjusted settings as necessary, then re-ran the tests. In the final wash up, all systems were able to complete all runs of all tests without a single error. As a result, testing that should have been completed in a week took over a month.
Can I lay claim to having a stable system? This really is a matter of conjecture. What I can say is that with the hardware that I have used, and using the applications in the manner that I use them, the systems as tested are stable. That’s all I can claim.
I could have left all systems at default manufacturer “Optimal” settings, but this poses some interesting questions, such as:
Are the factory settings designed for a) performance, or b) compatibility?
Are the each manufacturers settings a) equally “compatible” or b) some are more aggressive?
If you answered a) to each or either of these questions, stay back after school as you need some extra tutoring. :) Seriously, benchmarking is a far more serious issue than slapping some hardware together and running Quake 3 Arena, 3DMark and declaring a winner.
Manufacturers shoot for broad compatibility for a reason – to reduce support costs. If a customer has a problem (even though it might be easily fixed), the customer’s first reaction is likely to be “Oh,I bought a Brand X motherboard and it was the worst mistake I ever made.” To avoid this, manufacturers will choose compatibility every time, and I can’t blame them. Still, this approach does lock up some of the potential performance of a motherboard. If you are going to buy generic no-name (and no support) components and bang them in together, the manufacturer’s “Optimal” settings may be all that your hardware can tolerate. If, on the other hand, you care about performance, and you research and choose your components carefully, you have more flexibility. For this review, I chose to “hand optimize” each motherboard to remove the effects of manufacturer’s desires for ‘stability at all costs’.
Having said that, the tools used for these tests reflect standard business use along with some gaming. These tests don’t cover multimedia or scientific/engineering use (yet). It is (in my opinion) advisable to understand what the benchmark you are using is actually telling you, and to understand when, and when not, it can be applied. At the moment, I don’t have the expertise in some of the scientific or multimedia tools.
Disclaimer: I know very well the capabilities of the hardware I’ve used for this test. I’ve learned this the hard way (sometimes painfully with lost data and all night re-installs). The tests and the settings I’ve used work for me with the hardware I’ve used. It may not work for you – so be very careful about blindly adopting my BIOS settings for your hardware.
Rule number 1 – backup your data.
Rule number 2 – If you don’t know, don’t do it.
Rule number 3 – backup your data – I can’t stress this enough!
Rule number 4 – Don’t blame me if it goes awry.
Arguably the fastest AMD760 based motherboard is the EPoX 8K7A, and we’re here to find out if this is true – and if it is true, should you sell your house/first-born/significant other in order to posses one. If you have something similar to the other combatants (say an early ALi based motherboard or an SDRAM based Socket A solution), should you have a serious case of motherboard envy?
Representing the “Old Skool” are the Abit KT7A, one of the fastest SDRAM based Socket A motherboards, and even faster (as you will see) in “tweaked” form, and the Iwill KA266, one of the first DDR boards. The Iwill KA266 was never noticeably faster than the tweaked KT7A, but it was not disgraced. Has time wearied the KA266?
Be the first to discuss this article!