Pretty sure it wasn't the chipset 😀
Sometimes though, the chipset can be complete trash, but the right motherboard/BIOS can make it work okay anyway.
For example, I had an Abit KT7A, with VIA KT133A chipset. No matter what I tried, it was unstable.
Then I exchanged it for an MSI K7T Turbo, which had the same chipset, and pretty much the same features. But that board was 100% stable.
It was slow though. USB was pretty horrible, and trying to use USB for audio devices with DAW was not really possible, because you couldn't get latency under ~30 ms reliably. The same devices would get ~5 ms on a much slower system with Intel chipset with ease.
Likewise, the AGP bus wasn't great. I had an Athlon 1400 CPU in there. But when I tried my GeForce2 GTS card in an Celeron 1000 machine with Intel chipset (I think it was an i815 or something), the AGP bus speed was considerably higher, despite just having a 100 MHz FSB instead of the 266 MHz of the Athlon.
I'm not sure how much of that was a problem with the chipset itself, and how much of that was the result of a bug in the Athlon CPU though.
Namely, the Athlon CPU could only use 4 KB memory pages. Windows prefers using the much larger 4 MB pages for the AGP aperture, because you get better performance that way, less overhead with translations and such.
But yea... my experience with that Athlon system taught me that raw benchmark numbers of CPUs and GPUs don't tell the full story. They didn't warn me about the USB problems, and although the GPU performance in the Athlon wasn't bad, the Celeron indicated that there was still quite some untapped potential left on the table.
The Celeron CPU might not have been that fast, but apparently the chipset worked hard to extract as much performance as possible from all the components.