Hyperion vs. KT266A: VIA 4in1 Driver Comparison

With the release of the new VIA 4in1 Hyperion v4.45 Drivers, there has been some question as to the usefulness of VIA's new drivers on the older KT266A chipset. Since I own this particular chipset of course I am one of those people who are wondering. VIA has their own charts, but they don't tell us much we need to know.

As such I ran a series of tests against four version of VIA's 4in1 drivers: the newest Hyperion 4.45's, the 4.43, 4.42, and 4.35's. If you're familiar with installing and uninstalling the 4in1's you probably know it isn't always an easy task. In fact, it can be extremely difficult, especially when going back to an older driver. The installation may appear to run perfectly, but upon closer inspection the drivers are never actually updated. I noticed this after getting very similar results on three different drivers.

The best tactic is to attempt removal by running the driver installation again, and leaving everything selected. Click next and select the "Uninstall..." radio button at each step. This should uninstall the drivers from your system, but I stress the word should. Most of the time the .inf files remaining in the system directory are detected upon reboot and the old drivers are loaded again before you can help it. If this is the case you might need to remove them manually. I won't go into the details here, but if you can't figure it you are welcome to ask me.

Since the focus of my testing was on the VIA IDE Controller and the CPU to AGP Controller, I selected 3DMark2001 SE and SiSoft Sandra 2001 Drive Bench as my benchmarking tools. These tests alone are by no means a conclusive analysis, but they are still a worthy indicator of some of the differences between driver versions.

The test setup is running an MSI GeForce4 Ti 4600 and two Maxtor 40GB drives in RAID 0 at 7200RPMs. Everything is at stock speeds, I didn't bother overclocking anything so as not to invalidate the tests. The newest VIA RAID Patch v1.02 is used in all the tests. The NVidia Detonator 41.09 WHQL drivers are used as well as RivaTuner to ensure that all tests are run at 1024x768, no AA, no Antisotropic and Mips at 0. All other system configurations and driver version are consistent and identical throughout all of the tests.

I'm not going to linger on the details any more, so I'll just jump in and show you the results from each of these tests across all driver versions.

3DMark2001 SE

SiSoft Sandra 2001 Drive Bench

For the sake of being thorough, you can also find all of the individual benchmarks below:

4in1 v4.35 @ 11109 3DMarks		4in1 v4.42 @ 11086 3DMarks		4in1 v4.43 @ 10976 3DMarks		4in1 Hyperion v4.45 @ 10909 3DMarks
4in1 v4.35 w/ 49596 Drive Index		4in1 v4.42 w/ 42438 Drive Index		4in1 v4.43 w/ 39825 Drive Index		4in1 Hyperion v4.45 w/ 41909 Drive Index

After running all of these tests, it's pretty obvious that all of the VIA 4in1 releases after version 4.35 are hurting performance compared to their younger brethren. My guess is that revisions after 4.35 are unrelated to the KT266A chipset and are probably adding more useless code (useless for older chipsets, anyway).

Your experience may differ, I can't predict how your hardware will react to all of the revisions. I've seen a lot of people commenting on the 4.35 drivers as the 'best overall' driver for the KT266A chipset, but now I can see a real difference on my system. As for the Hyperion 4.45's? Unless you have a new chipset such as KT400 or P4X400 the Hyperions might not be for you.

Subject: UPDATE Posted by: Tekime

Reviewing this article a year later, I have to point out that this was a serious 3DMarking stage for me. You should never consider 3DMark and one Sandra test as a thorough examination of performance; and while I mentioned that briefly I should have been more specific in stating this. Point being... no matter what the test, a few lone benchmarks won't tell you total system performance. What these tests did do, however, was get me a faster 3DMark.