The process:

Drying and testing

It works!

Three down, many to go!

I straightened pins on some Cyrix 5x86 (sadly only 100 MHz models) recently, and plan to test them soon. It's strangely satisfying to fix someone else's scrap, especially when sourced for a nice price 😀

debs3759 wrote on 2022-01-15, 03:18:

I straightened pins on some Cyrix 5x86 (sadly only 100 MHz models) recently, and plan to test them soon. It's strangely satisfying to fix someone else's scrap, especially when sourced for a nice price 😀

Indeed. Makes me happy to give these long neglected chips a refurbishment and save them from the scrap!

This one is in a particularly bad state of neglect

Before^
After>

Oddly, it actually works despite appearing to have been not only outside, but in direct sunlight for years in I assume the desert.

The solder is in a rather bad state, might be worth a reflow with flux…?🧐🤔

Edit: 🤣, tried touching it up a little and the heat from soldering or the flux caused the top layer to come off the pcb exposing the ground plane which absorbed some solder, I decided to chuck it, best to just let it go. :,( plenty where it came from.

After washing leave PCBs to dry for at least a week before soldering. Alternatively put them in an oven at 50-70°C for a couple of hours.
Moisture seeps into the PCB and will boil when heated to soldering temperatures turning the PCB into popcorn so to say.

That blade of a cutter-knife is a brilliant idea for adjusting bent pins!

Including the ones I had already, I now have 17 233s

Since I have now cleaned up and restored the pins on all of the 233s now, I decided to do some binning.

Interesting data that I found.

I went with a very simple and quick test to check the max posting frequency of each chip. Just as a quick method of finding the best ones.

The goal is only to get far enough in post to start counting memory without a heatsink (takes longer than this to get hot) then the computer gets shut down. If the processor hangs prior to that it failed the test.

Here are the results:

All of these processors obviously work at 233

I started at 100mhz bus 3x

Ten failed to post right off the bat at 300mhz

Next I bumped the bus to 105mhz
Which should be 315mhz but the bios is reporting 333mhz, not sure where the error is there.

Three more processors failed to post at this speed.

Next I bumped it to 110x3 again the board reports 333mhz. One more processor fails to post.

(Down to three out of 17)

Finally I tried 100x3.5 and all remaining failed to post.

Interesting results that seem to show that there is quite a range in result when it comes to overclocking these chips.

I won’t delve in deeper and find the actual max stable overclock for each of these chips but this short little test I’m sure is revealing of something.

None of the 200mhz chips that I have ready would post at 3x110 Didn’t do further testing other than to verify one got close to posting at 300. But I ended the testing.

Nice test. Were all those done at 2.8V or did you already increase voltages?

Doornkaat wrote on 2022-01-18, 06:19:

Nice test. Were all those done at 2.8V or did you already increase voltages?

Sorry, yes 2.8v

Sphere478 wrote on 2022-01-18, 06:52:

Doornkaat wrote on 2022-01-18, 06:19:

Nice test. Were all those done at 2.8V or did you already increase voltages?

Sorry, yes 2.8v

What I found when overclocking CPUs of that era is that some will OC relatively well at stock voltages but hardly profit from increasing voltages while others are mediocre overclockers at stock voltages but become champions when voltages are increased. I recommend testing the seven CPUs that did 300MHz again with increased voltages to find your golden sample. There's a reasonable chance your best overclocker isn't among the three that did best at stock voltage.
Having seventeen of the same CPU is a great sample size for Pentium MMX binning! 😁👌

Doornkaat wrote on 2022-01-18, 07:35:

Sphere478 wrote on 2022-01-18, 06:52:

Doornkaat wrote on 2022-01-18, 06:19:

Nice test. Were all those done at 2.8V or did you already increase voltages?

Sorry, yes 2.8v

What I found when overclocking CPUs of that era is that some will OC relatively well at stock voltages but hardly profit from increasing voltages while others are mediocre overclockers at stock voltages but become champions when voltages are increased. I recommend testing the seven CPUs that did 300MHz again with increased voltages to find your golden sample. There's a reasonable chance your best overclocker isn't among the three that did best at stock voltage.
Having seventeen of the same CPU is a great sample size for Pentium MMX binning! 😁👌

This pile of cpus has allowed me to investigate some multiplier mysteries that I have pondered as of late.

Which pentium 1 cpus were multi locked?

I should probably run the 233s through again on 75x 2x and see if one magically pops up at 300

Wouldn’t that be just amazing to find a smp capable 4x pentium….😲 because tillamooks can’t do smp unfortunately .

Now this is dedication. On Sunday I had to straighten out pins on just one i200 (I used a steak knife). I was thankful when it was over…. I would sooner build my own processor than go through that again.

Sphere478 wrote on 2022-01-18, 09:01:

Wouldn’t that be just amazing to find a smp capable 4x pentium….😲 because tillamooks can’t do smp unfortunately .

It would be! Though I wouldn't count on it because afaik the P55C core simply doesn't have the register for 4x multi.

Meatball wrote on 2022-01-18, 09:36:

Now this is dedication. On Sunday I had to straighten out pins on just one i200 (I used a steak knife). I was thankful when it was over…. I would sooner build my own processor than go through that again.

I only have a few left. I’ve probably done this to about 150 socket 5/7 chips now.

So far this entire pile has been good even the one I tossed because of sun damage 🤣

Doornkaat wrote on 2022-01-18, 09:37:

Sphere478 wrote on 2022-01-18, 09:01:

Wouldn’t that be just amazing to find a smp capable 4x pentium….😲 because tillamooks can’t do smp unfortunately .

It would be! Though I wouldn't count on it because afaik the P55C core simply doesn't have the register for 4x multi.

Indeed, I don’t expect to find one.

I wonder if it may be possible to build some sort of external multiplier interposer?

Loving your work here.

Sphere478 wrote on 2022-01-19, 01:12:

Indeed, I don’t expect to find one.

I wonder if it may be possible to build some sort of external multiplier interposer?

Yes but it works by simply feeding a different FSB to that chip than the motherboards bus speed

Very common technique when installing a fast 286 or 287 to a slow motherboard

rmay635703 wrote on 2022-01-19, 01:20:

Sphere478 wrote on 2022-01-19, 01:12:

Indeed, I don’t expect to find one.

I wonder if it may be possible to build some sort of external multiplier interposer?

Yes but it works by simply feeding a different FSB to that chip than the motherboards bus speed

Very common technique when installing a fast 286 or 287 to a slow motherboard

It needs a buffer or something yeah?

Last one!!

And it’s um,, “special”…

Edit:

Done!!