Well, not unheard of, here are few examples

http://www.chipdb.org/cat-intel-1310.htm

It's probably lower speed part that was "repainted" to be sold as faster speed.

Contemporary fakes usually raised the clock frequency, like from 20 MHz to 25 MHz, or from 25 MHz to 33 MHz.

Etch matches 16MHz version https://www.x86-guide.net/en/cpu/Intel-386DX- … -cpu-no865.html including the double sigma visible on yours between "25" and "85". Clock is located directly under the white "6", maybe you will be able to tell under different angle. It sure does look like S40344 with "44" visible inside lower "8" 😀

Wonder what was the price difference between 16 and 25MHz chip. At the time nobody was selling bare CPUs, upgrading market was born with 486 generation. In November 1990 Computer Shopper cheapest board + 386DX-25 was ~$600. 386DX-16 is nowhere to be seen, must have been a very short production run with Intel switching to SX for low end in 1988.

That's interesting. The original 80386 was a modest 12 MHz model, if I remember correctly.
So 16 MHz was the next faster model after that.
But no idea how it compared to the 20 or 25 models in terms of of popularity/availability.
- I've started with a 40 MHz AMD model first. 😇

I mean, in theory, a standard 80386 can do everything a 386SX can (that's old news, I know).
It has the ability to work with both 16/32-Bit I/O.

And the slower/outdated 80386 models would have been fine for use on existing 80286 mainboard designs, still, just like the 386SX models.
Provided that the price and availability were okay.

Memory performance would have been better with a real 80386, too,
if it had RAM or cache connected to its 32-Bit processor bus.

Hm. I really wonder how a 16 MHz 80386 compared to a 386SX in ~1990..:
Were the 386SX models cheaper than the outdated 16 MHz models of the "full" 80386?
Or were they available in higher volume?

If the "full" 80386 at 16 MHz was available in large quantity, still, it would explain why fakes used it.

Hm. But by 1990, the 16 MHz 80286 models were common, too.
And they were quicker than the 386SX at same clock speed, too.
So the "full" 80386 wasn't the only competitor to the 386SX.
So many questions.

rasz_pl wrote on 2023-10-20, 08:23:

Etch matches 16MHz version https://www.x86-guide.net/en/cpu/Intel-386DX- … -cpu-no865.html including the double sigma visible on yours between "25" and "85". Clock is located directly under the white "6", maybe you will be able to tell under different angle. It sure does look like S40344 with "44" visible inside lower "8" 😀

Wonder what was the price difference between 16 and 25MHz chip. At the time nobody was selling bare CPUs, upgrading market was born with 486 generation. In November 1990 Computer Shopper cheapest board + 386DX-25 was ~$600. 386DX-16 is nowhere to be seen, must have been a very short production run with Intel switching to SX for low end in 1988.

386DX-16 was the standard CPU in most FM Towns machines. So, quite a few exist, but will depend on your region.

^Ah yes, the FM Towns.. I remember it from watching an CPU upgrade video at YouTube.

https://www.youtube.com/watch?v=egs1O7pgIA0

FM Towns 386 is labeled as 16MHz but at that time (late '89 and later) Intel has improved the process to 33MHz (and beyond, but never sold faster 386 chips in order to not undercut their 486 line).

In fact every Towns chips is marked as "IV" which I belive is Intel's "High Speed CHMOS IV Technology" and considering how well my CPUs tolerate 33MHz clock I'm willing to bet these were all 33 or at least 25MHz capable variants but officialy sold as 16MHz (with lower price). There could be some rejects and marginal chips of lower quality but none of these is truly 16MHz. These machines were too late, too rare and too expensive to try and remark their CPUs to 25MHz or higher for profit.

If this fake 386 was made back in the day then it's most likely 386DX-16 or 386DX-20 sold as 25MHz variant. Quite a few of the older chips would work pretty well in 16-bit mode when overclocked, but fail in 32-bit protected mode or when paired with NPU. If it's a new fake then it's most likely a late 386DX-33 or even AMD chip relabeled to attract attention of various collectors.
One can still find early 386-16 and -20 (non-DX) on ebay in reasonable prices, but a -12 variant or non-EE marked would fetch quite a price.

My very small motherboard with the Super 286 stickers had an overclocked 386sx16 underneath all the stickers

Solo761 wrote on 2023-10-20, 07:47:

Well, not unheard of, here are few examples

http://www.chipdb.org/cat-intel-1310.htm

It's probably lower speed part that was "repainted" to be sold as faster speed.

Gotta wonder what the point of remarking a dx25 to a dx2-66 would be.

Go to a flea market once and never return?

My family's first PC had a 33MHz rated AMD 386DX CPU installed on a 25MHz capable motherboard.
Board was a Biostar MB-1325 variant.

This was purchased new from a reputable local shop, in the greater Montreal area in the Automn of 1992.

You guys were right, after the paint came off, there emerged a (bug-free) 80386-16

majestyk wrote on 2023-10-21, 12:26:

You guys were right, after the paint came off, there emerged a (bug-free) 386-16

386 non-DX? Well at least that's worth more money now than 386DX-25. It's B1 step and even EE marked it has quite a few fun bugs.

Deunan wrote on 2023-10-21, 12:30:

majestyk wrote on 2023-10-21, 12:26:

You guys were right, after the paint came off, there emerged a (bug-free) 386-16

386 non-DX? Well at least that's worth more money now than 386DX-25. It's B1 step and even EE marked it has quite a few fun bugs.

Plus it´s 56% overclockable 😀

rmay635703 wrote on 2023-10-21, 00:19:

Gotta wonder what the point of remarking a dx25 to a dx2-66 would be.

Go to a flea market once and never return?

Maybe OCed to 33 to "look" like legit thing since it also used 33 MHz bus? I don't know if BIOS modding was a thing then but if it came in a system it could have been edited to display DX2-66 during boot, and showing 33 MHz bus would be enough to fool someone not deep into this stuff as legit, "It shows 33 in tests because that's how fast it communicates with peripherals, but it works at 66, that's why it's DX2, don't worry about it..."

It even looks like some kind of overdrive CPU underneath all of that paint so that could have been used further to explain why in tests it performs worse than real DX2-66...

Who knows 🤷‍♂️

It's a DX2-50. Early Intel ODPR CPUs were marked to indicate which chip they replace, not what you get in the end.