K6 cpus are quite power hungry. You can try lowering voltage, but anything more then 0,1-0,2V under the rating could result in system instabilities - unless you are also under-clocking it.

Performance wise they are the same.

kixs wrote:

K6 cpus are quite power hungry.

Really depends on the model.

The "best", and my recommendation, is this 1.6V K6-III+ 400. A German eBay shop was selling tons of these and I know that have this particular chip. It's very energy efficient, but not all boads go that low in terms of voltage. I usually add 0.1V for extra stability with these old machines.

Fabian did a nice write up on it here:

http://www.amoretro.de/2012/07/amd-k6-iii-400 … r-socket-7.html

At 2V, the chip does 550 MHz without any issues.

I do recommend staying away from 600 MHz. AMD themselves never released a 600 MHZ model. They did release an odd 570 or so version, which runs on an odd FSB frequency. So I believer there is an internal limitation of the chip. So just stick with 550 MHz.

The extra gain in performance is not that great though. The K6-III+ 400 already benefits from the higher 100 MHz FSB and the 256 KB Cache, so the extra clock speed of the chip does little in real terms. But fun for benchmarking.

Undervolting is the key to lowering power consumption (and heat). I don't know of any other reason for doing it, but those reasons are good enough as long as the chip stays stable at the speed you want. I spent a lot of time testing undervolt settings on a couple K8s, one a server (to save power) and the other a heat-troubled laptop. I've never messed with it on the K6.
Underclocking is a means to enable undervolting, and overvolting is a means to enable overclocking.

On the K8, I did read about some peculiar scenario where it's been theorized that too much undervolting could actually damage the chip. I think the concern was that if the Vcore was too low, there'd be a chance to end up with a reverse biased voltage somewhere that would be destructive. I have no idea if that issue was real, or just gossip. It's not something I'd worry much about, since CPUs are usually cheap and common. I might change my mind if I actually had one die from this, but until then, not really bothered.

The only experience I have is lowering the core voltage on a Pentium MMX 200 from 2.8v to 2.5v to help reduce heat output. I've done this twice, and stability didn't seem to be affected at all. Of course, as said, that was with Intel chips, so I'm not qualified to speak on undervolting AMD chips.

I'd guess that either the 2.4V is an older version, or maybe the ones that don't hack it at 2.2V are retested at 2.4

The reason for my question is indeed heat output. Not that big a deal anyway.
Just some bizare quriosity that I have regarding CPU's.