Yes, ES=Engineering Sample

Some additional info here https://www.pcjs.org/blog/2015/02/23/

Thank you! So based on that I guess it is a step B0 engineering sample. (It's weird they used the letter "O" and not a zero.)

That's great news then. Not only is it one of the early buggy 386s, it's an engineering sample too. 😀

LChackr wrote on 2020-05-30, 19:46:

Not only is it one of the early buggy 386s, it's an engineering sample too. 😀

I remember the time I suffered through an extremely long win95 install for evaluation on a 4MB PS/2 model 70 only to get

Setup Error B1: Setup has detected an 80386 processor that is not compatible with this version of Windows.
Before you can run this version of Windows, you need to upgrade your processor.
Contact your computer manufacturer for more information.

https://www.pcjs.org/blog/2015/10/27/

Never saw a BO processor though. Where did you find this CPU?

Not only is it one of the early buggy 386s

Every non-SX 386 from Intel without the double sigma marking should be seen as buggy and safe for 16-bit software only.

derSammler wrote on 2020-05-31, 14:30:

Not only is it one of the early buggy 386s

Every non-SX 386 from Intel without the double sigma marking should be seen as buggy and safe for 16-bit software only.

Out of curiosity, can software detect these buggy CPUs somehow?

It must have been pretty embarrassing for intel to admit that their brand new (and very expensive) 32-bit CPU could only run 16-bit code. Although everybody likes to talk about FDIV, this was a far bigger F@#% up in my opinion.

myrkraverk wrote on 2020-06-01, 13:09:

derSammler wrote on 2020-05-31, 14:30:

Not only is it one of the early buggy 386s

Every non-SX 386 from Intel without the double sigma marking should be seen as buggy and safe for 16-bit software only.

Out of curiosity, can software detect these buggy CPUs somehow?

I see the above PCjs blog has all the software tests to identify these chips; I had not read it until now.

Anonymous Coward wrote on 2020-06-01, 13:43:

It must have been pretty embarrassing for intel to admit that their brand new (and very expensive) 32-bit CPU could only run 16-bit code. Although everybody likes to talk about FDIV, this was a far bigger F@#% up in my opinion.

It wasn't the only significant bug in 386, although most other ones could be worked around. After the FDIV though Intel finally figured out that adding some SRAM cells to microcode path and a mechanism to upload data there would let them patch the CPU rather than having to replace it, and usually with minimal perfomance cost since the instructions most likely to have bugs in them are the complex ones - and these are slow to execute anyway.

AFAIK Intel kept making 386DX in PGA package well into 2000 and they had never bothered to fix the POPA like they did for SX. Or perhaps the "fix" in SX case is just different internal timings due to narrower bus.

Also no double sigma means it has 32bit multiply bug.

Cheers,

Never saw a BO processor though. Where did you find this CPU?

I found it on eBay. Someone just listed it as a 386 CPU. I immediately recognized that it was old, and interesting.

You should find the 12MHz version.

You should post it on the CPU World forum.

http://www.cpu-world.com/forum/

pentiumspeed wrote on 2020-06-01, 19:01:

Also no double sigma means it has 32bit multiply bug.

Cheers,

That is not correct. The double sigma means it was tested and doesn't have the bug. If it was tested and does have the bug, it will say "16-bit software only". If it is the stepping that could be buggy but doesn't have either marking, it could be OK or it could be buggy. Later steppings had no marking as the bug had been fixed. This CPU is an early stepping that probably does have the bug.

Ok, what is "IV" stands for?

Next question related to this bugs found in 386DX processors, what about AMD Am386 processors, did AMD fix many bugs when they cloned Intel's 386DX?

Cheers,

IV should stand for "CMOS IV", meaning it's more advanced than "CMOS III". I can't remember what exactly all the differences are between them, but wikipedia should explain it in detail.

How can I tell on intel 386 revision by looking at it, For example D0 is what?

Thanks,

Katmai500 wrote on 2020-08-26, 16:46:

You should post it on the CPU World forum.

http://www.cpu-world.com/forum/

Agree ! You may have one of the only ones that are still in the hands of someone to Document it (recyclers do not care !)

Agree ! You may have one of the only ones that are still in the hands of someone to Document it (recyclers do not care !)

Sure. I'll post it over there tomorrow. I am really happy that I was able to snag it before a recycler or anyone else did. I want to drop it into a 386 board and see which bugs it does have. 😀

It seems like it has had an interesting life. There are a few hand-made ink markings on the ceramic that have been mostly erased. One of them reads "DAISY".

You should find the 12MHz version.

I WISH!

If it was tested and does have the bug, it will say "16-bit software only".

I also really want to find one of those chips that is officially marked like that.

This CPU is an early stepping that probably does have the bug.

Yes. The officially released B0 step 386s had a number of bugs. (It wasn't until the D-step that the 386s were solid.) So I do suspect that since this is an engineering sample of a B0 step then it should have the 32-multiply bug, among a list of others. Of course, I think that's a good thing. It's part of the history of the evolution of the 386.

LChackr wrote on 2020-08-27, 03:05:

Agree ! You may have one of the only ones that are still in the hands of someone to Document it (recyclers do not care !)

Sure. I'll post it over there tomorrow. I am really happy that I was able to snag it before a recycler or anyone else did. I want to drop it into a 386 board and see which bugs it does have. 😀

Where you even able to get it at a reasonable price, then?