Nobody likes 180 though, it doesn't sound like a real speed, how about we rename it to "166+ Turbo"

BitWrangler wrote on 2022-02-10, 15:47:

Nobody likes 180 though, it doesn't sound like a real speed, how about we rename it to "166+ Turbo"

Even Intel sold 180MHz PPro and Pentium MMX Overdrive.

pshipkov wrote on 2022-02-10, 15:25:

Ok, you really missed the important information :) 5v to cpu will affect the cpu lifespan. Live fast, die young. […]
Show full quote

Ok, you really missed the important information 😀
5v to cpu will affect the cpu lifespan.
Live fast, die young.

I didn't explore this area enough, but feel motivated now.
Will give the 3 200mhz capable cpus here a spin on air cooling and see what comes out of it.
But if you slap a low voltage Peltier on that magical cpu that may be enough for complete stability while still avoiding the water condensation.
At minimum you may be able to run 180mhz fully stable on air cooling with 4v or even 3.3v ?

Let us know if any work better on 5v!

BitWrangler wrote on 2022-02-10, 15:47:

Nobody likes 180 though, it doesn't sound like a real speed, how about we rename it to "166+ Turbo"

Lol

Ok, tried briefly this morning 3x60MHz and 3x66MHz on air cooling.

--- 200MHz

One ADW CPU can POST at 3x66 + 4V, but no BOOT.
One ADZ CPU can BOOT. Needs 4V for that.
Both don't light up at 3V/5V.
No clock divider. Running 1:1.
System is not stable.
Looks like i will need a trimmer bolted on the motherboard to fine-tune stuff and eventually get things to a better place.
I am pretty sure that slapping a low-voltage Peltier on the ADZ CPU will get it over the finish line.
Will try that later today.

Why i didn't explore this early on ?
Looking back at the time when i was messing around with that stuff - the only ADZ processor that handles 200MHz well enough was throwing tantrums at 3x60 with anything but 3V. I guess this confused me and i decided that the CPU is flaky. But fresh look today gives me a different perspective - if the chip likes 200 at 4V and 180 at 3V only - so be it. That's how things are sometimes.

--- 180MHz

One ADW can do it at 3V.
One ADZ can do it at 3V.
One ADZ can do it at 4V.
No clock divider. Running 1:1.
System felt pretty stable, but eventually failed my favorite combo-grind-test: 3D rendering + VC6 code compilation + IE4 opening www.vogons.com + GlQuake 1 in Win95.
Low-voltage Peltier takes care of business for good.

---

I am decently hopeful now that Chadti99's post about air cooling will prompt me to do some more tweaking and eventually move that 180MHz PC to 200MHz.
Let's see ...

---

3x60 (180) vs 4x40 (160) - that is +20MHz to FSB.
3x66 (200) vs 3x60 (180) - only +6MHz to FSB.
The friction from 180 to 200 is much higher than 160 to 180.
So 180MHz is kind of the pivot point for 486 hardware overclocking.
But i get BitWrangler's joke 😉

Okay I know I've been all over the place providing results but here are my results with an air cooled AMD5x86 at 200Mhz, seemingly 100% stable, at 5 volts. I'd like to perform the full set of tests on the LuckyStar LS-D board I have, and will at some point, as it beats the Biostar in most tests. I really like that the Biostar offers attachment points on the CPU socket for me to secure an oversized cooler and mount vertical in a tower case. A large cooler and thermal paste is required for stability. I'll also add that I've tried this board at 66x3 and it ends up being slower as it requires the slowest cache timings for stability. At 66Mhz, neither 1:1 or 1:2/3 CPU:PCI ratio are stable on this board or any other board I've tried with this CPU on air cooling.

Here is an unedited almost 2 hour long bench and torture session, with me rambling a bit in my native KY twang.

https://youtu.be/6gqgnFn3QZ8. *Still processing High Def

1) Biostar 8433UUD 3.1, 256k 12NS Cache, fastest timings, 50Mhz FSB, 1:1 CPU:PCI
2) AMD 5x86 P75 133 @ 200Mhz (50x4), 5volt, air cooled
3) 32MB EDO 60NS
4) SD2IDE
5) Matrox G200
6) ESS1868
7) no NIC
8 ) DOS6.22 & Windows95
9) Onboard PS/2 mouse port

Continued

Continued

Neat, now the UUD at 200 MHz (4x50) matches that of the LSD ta 200 MHz (3x66), that is, they both score 21.8 fps in Quake. Does it run Windows in this configuration? If so, request your trophy from CPU Galaxy.

feipoa wrote on 2022-03-16, 14:31:

Neat, now the UUD at 200 MHz (4x50) matches that of the LSD ta 200 MHz (3x66), that is, they both score 21.8 fps in Quake. Does it run Windows in this configuration? If so, request your trophy from CPU Galaxy.

Yes it installs and runs Windows no problem, you’ll see that it runs Windows in the long boring video if it ever finishes processing. I wasn’t paying attention and recorded in 4k30.

I think it’s likely I could hit 21.9, at stable Windows settings, if I were to test the LS-D board at 66x3 in DOS6.22. It turns out DOS6.22 provides slightly faster results than 7. All my previous results have been in 7 unfortuneatly.

*I see that pshipkov maxed out at 21.8 on the LSD at stable settings, so 21.8 might be it.

Awesome. Love it.
Magical CPU in action.

We can get LSD to hit past 22 fps in Q1 on a narrow overclock + maxed out bios params, but fully stable system is much more impressive.

About the socket.
You can swap the "lid" only - basically the moving part. It is easy and safe. Small attention needs to be paid to the attachment points. This way you can put different lid you like.

pshipkov wrote on 2022-03-17, 00:40:

Awesome. Love it. Magical CPU in action. […]
Show full quote

Awesome. Love it.
Magical CPU in action.

We can get LSD to hit past 22 fps in Q1 on a narrow overclock + maxed out bios params, but fully stable system is much more impressive.

About the socket.
You can swap the "lid" only - basically the moving part. It is easy and safe. Small attention needs to be paid to the attachment points. This way you can put different lid you like.

Oh wow, I didn’t realize this but it kind of makes sense, any chance you can demonstrate the lid swap with a video or some pics?

I’m thinking this still might not work as the socket clips are part of the immovable bottom half.

You could also just replace the whole socket with a desoldering gun.

feipoa wrote on 2022-03-17, 01:11:

You could also just replace the whole socket with a desoldering gun.

Yes, def another option I might explore.

Used picture i found online. Scribbled on it.

1. lift the handle to loosen the lid.
2. use a tool with sharp/pointy end to gently pull the plastic pieces of the lid holding onto the "teeth" of the base.
3. pull. done.

lol, those clips are fun! Very delicate and very brittle. Guaranteed that you break at least 1 of the 4.

feipoa wrote on 2022-03-17, 01:11:

You could also just replace the whole socket with a desoldering gun.

How difficult do you think this would be with the right tools and patience? Frustration level seems high 🤣.

From my experience: with a cool head and a fresh scalpel, there's a 50% chance you break only 1 clip. Practice on a dead board. Check your box of dead boards to see if you have a spare top. Brand A top sleeves don't necessarily fit brand B's.

EDIT: I've taken off quite a few of these because previous owners like to jamm in CPUs with bent pins. Those bent pins mess up the sockets inside and they need to be bent back into place without the top sleeve on.

Chadti99 wrote on 2022-03-18, 11:29:

feipoa wrote on 2022-03-17, 01:11:

You could also just replace the whole socket with a desoldering gun.

How difficult do you think this would be with the right tools and patience? Frustration level seems high 🤣.

I've tried it with a soldering iron once and absolutely fucked it up, it is definitely not a walk in the park.

feipoa wrote on 2022-03-18, 11:45:

From my experience: with a cool head and a fresh scalpel, there's a 50% chance you break only 1 clip. Practice on a dead board. Check your box of dead boards to see if you have a spare top. Brand A top sleeves don't necessarily fit brand B's.

EDIT: I've taken off quite a few of these because previous owners like to jamm in CPUs with bent pins. Those bent pins mess up the sockets inside and they need to be bent back into place without the top sleeve on.

Not to go too far off topic but how did you diagnose bent socket pins or was it a hunch from experience? Did you use a post diagnosis card?