pinesal wrote on 2022-02-10, 21:36:
I was ghosted again by the ebay seller so I am just going to buy at the asking price. I'd be willing to sell one of the spares to someone for $25 shipped once I get them. (connected USA only)
Some mentioned benchmarking all three CPUs and keeping the best one. How would I go about doing that? What software do I need and should I be expecting to scrape and reapply thermal paste 3 time? 🤣
Thanks.
I think they meant binning.
And yes, you should.
But you may encounter the limits of the board before the cpu.
Turn off mobo cache, loosen mem timings
And set cpu multi to say 3x
Find your max mobo fsb frequency. Do a actual benchmark burn in to find this
Keep the bios settings
Next
Set cpu to 2x= 6x and fsb to 83mhz
And lower the voltage, see if all run at your lowest setable voltage. Which unfortunately may be of no help because many boards only go down to the actual rated voltage of the chip. 2.0v
Anyway, after setting the lowest voltage,
You are now ready for the real test,
Start raising the fsb until one doesn’t post and write the setting that happened at. Hopefully it happened before your max mobo frequency
So with this result you will know it was the cpu that failed to post at x frequency and nothing else.
This is useful because it is quick. You will be able to see which of the three will post at the highest clock and choose that one to keep.
You can find stable frequency later which will take more time.
Suffice to say the one that posts the highest should have the highest stable clock also. Which of course will be less than the postable clock
RaiderOfLostVoodoo wrote on 2022-02-11, 00:39:Yes, all K6 CPUs have an open multiplier.
That why the boards+CPUs are so expensive compared to Pentium II. Back in the day ever […]
Show full quote
pentiumspeed wrote on 2022-02-11, 00:12:
does the K6-2E plus have adjustable multiplier via jumpers? To run on 66MHz and 75MHz fsb? Around 400-500MHz all I wanted and less too.
Yes, all K6 CPUs have an open multiplier.
That why the boards+CPUs are so expensive compared to Pentium II. Back in the day everybody wanted Pentium II because they offer a bit more performance. Now everyone wants K6 for the flexibility. 🤣
The plus variants (K6-2+ and K6-III+) can even change their multiplier during operation via software, because they're mobile CPUs.
cyclone3d wrote on 2022-02-11, 00:25:
What he means is:
Your manual will probably state, that 5.5x is the highest multiplier. But for the latest K6 CPUs the 2x multi gets interpreted as 6x.
pinesal wrote on 2022-02-10, 21:36:
Some mentioned benchmarking all three CPUs and keeping the best one. How would I go about doing that? What software do I need
Benchmarking is the wrong word. The correct word is binning.
You'll need Prime95. But an older version. The newer versions don't work with old CPUs.
You play around with FSB+multi+voltage to see what's possible. Then test for stability with Prime95.
pinesal wrote on 2022-02-10, 21:36:
should I be expecting to scrape and reapply thermal paste 3 time? 🤣
No, you don't need to do that. Socket 7 CPUs don't run that hot. The low voltage variants (like my K6-III+ 400ATZ who runs at 1.6V) don't even need a cooler.
Thermal paste doesn't hurt, but for testing it's a waste of ressources imho. Do the testing without paste and once you've decided which one you gonna use, you can apply paste.
Yeah 🤣 it’s crazy but people forget that there actually were a few socket 7 chips that you could legit run without a cooler at certain very conservative settings while others, forget it!!!
But most of the hotter chips will “survive” without paste. Which is wild, but totally a thing.
Now I’m curious… hottest socket 7 chip?
Coolest I’m pretty sure is either a (underclocked/undervolted)k62+ Or tillamook