The AMD K5 series doesn't generally overclock well beyond 4 MHz from its rated speed.

Were you able to load Windows 9x and run some programs at lengh?

I have a K5 PR166 as well, but I want to obtain a PR133.. I want to replace my Pentium 133/Voodoo system with a PR133 chip.

feipoa wrote:

The AMD K5 series doesn't generally overclock well beyond 4 MHz from its rated speed.

Were you able to load Windows 9x and run some programs at lengh?

It overclocks just fine if you up the voltage, the regulator capped at a max of 5volts, I didn’t go that far.

The machine in question had windows 98, I installed the os, ran some emulators and surfed the internet with it in that state, seemed rock solid @ 83mhz x 1.5 which on my board read pr200 (however accurate that was)

Ahhh, 125 MHz (83 x 1.5) is fairly easy to achieve on a K5, but not 133 MHz (66 x 2) unless you have a real PR200 chip. Are you able to get 66x2 going?

feipoa wrote:

Ahhh, 125 MHz (83 x 1.5) is fairly easy to achieve on a K5, but not 133 MHz (66 x 2) unless you have a real PR200 chip. Are you able to get 66x2 going?

i had opposite experience. my pr166 worked at 66*2 and completed all benchmarks but crashes at 83*1.5.

I've never heard of a PR166 running reliably at 133 MHz. At what voltage were you running?

oerk wrote:

Does anyone know if the 5k86 really is faster/more efficient than the SSA/5 stepping? Since the later 5k86 uses a higher PR rating than it's clockspeed, while the earlier one has a PR rating exactly at it's clockspeed.

I quote myself from cpuworld:

So, here's my little comparison. It's no comprehensive test, just a few haphazard results from games and demos I often test with […]
Show full quote

So, here's my little comparison. It's no comprehensive test, just a few haphazard results from games and demos I often test with, all running under DOS 7.1.

AI Bench [basically a rotating textured torus in true color]: 894 vs 871 points, a measly 2.6 percent lead for the PR133.

Doom: 73.2 vs 86.0 fps - a more healthy 17% difference.

Duke Nukem 3D (320x200): 64 vs 76 fps. Almost 19%.

Duke Nukem 3D (640x480): 26 vs 29 fps, mere 11.5%, probably heavier load from frame buffer copies.

Quake 320x200, nosound: 19.8 vs 24.6 fps, 24.2% difference in probably the most FPU intensive benchmark.

The board was a FIC PT-2003 (i430 FX chipset, 512kB PB COAST cache), the VGA an Asus V3000 PCI (Riva 128).

On the overclocking debate - my PR166 (the one I play with, not the one on display) was pretty happy at 2*66MHz@3.5V, but I never tried it at 1.5x83 - maybe later. Can't say it was rock stable as I - for example - didn't even try to boot Windows on it, I just did DOS benchmarks and watched a few demos.

feipoa wrote:

I've never heard of a PR166 running reliably at 133 MHz. At what voltage were you running?

His experience may be a misconception, if I set my board to 2x I don’t get a real 2x in the chip
I have to set 3x to get 2x.

Perhaps he had the board set to 2x or 2.5x or whatever you set a pr166 at but obviously he wouldn’t actually be running at the multiplier silkscreened on the mobo.

1.5x and 2x provide the same multiplier on my chip
2.5 and 3 are different

It's been a long time since I've run these K5 chips for testing, but something in the back of my head is hinting that these chips are multiplier locked? There's 1.5x, 1.75x, and 2.0x for the PR100/PR133, PR166, and PR200, respectively.

whatever it is, you can have a look at my benchmark results in this thread:
Re: my socket7 cpu benchmark results