• Please review our updated Terms and Rules here

Packard Bell Packmate 386 SX Processor Question

F

FParker

Member
Joined
Feb 6, 2023
Messages
29
Hello,

I have a pretty simple question. Is there a way, if any, to pop this processor out of this socket? It seems as if there would be a way, but I'm not sure. I can get a better sideview of the processor socket tomorrow if this picture does not suffice.20230830_140805.jpg

The reason I ask is because I'd like to replace it with one of those Cyrix "486" 386 processors (or the TI variant of said processor). However, it may not be possible, given the shape of the processor with those nubs (for lack of a better term) jutting out on each corner. I don't know what those would be for, given that I am a bit of a layman on processor design minutiae from the time.

The ultimate reason for the replacement is that I'd like to run Windows 95. I get that this is probably a suboptimal platform, but it's what I have.

I appreciate your help in advance,
Forrest
 
FParker said:
The ultimate reason for the replacement is that I'd like to run Windows 95. I get that this is probably a suboptimal platform, but it's what I have.
Click to expand...
That's not going to work. Even with a Cyrix or TI "486" upgrade chip, the 386SX platform is way, way too slow for Windows 95. Even a fast 386DX/40 MHz will run Windows 95 only very sluggish. And what you have here is about 5-6 times slower. Yes, I'm sure you want to try anyway, but you will only waste time.

The socket itself is just a fancy PLCC socket. So pulling the chip shouldn't be any different. But that's just by the look of it. I never had anything with such a socket. You may have to press down on the squares to unlatch the four sides.
 
Timo W. said:
That's not going to work. Even with a Cyrix or TI "486" upgrade chip, the 386SX platform is way, way too slow for Windows 95. Even a fast 386DX/40 MHz will run Windows 95 only very sluggish. And what you have here is about 5-6 times slower. Yes, I'm sure you want to try anyway, but you will only waste time.

The socket itself is just a fancy PLCC socket. So pulling the chip shouldn't be any different. But that's just by the look of it. I never had anything with such a socket. You may have to press down on the squares to unlatch the four sides.
Click to expand...
Fair enough. Would I, by any chance, be able to run Windows 3.X on this processor? I would think that that would be a bit more manageable than 95.
 
I'm surprised that the business about a "fancy PLCC socket" went without commentary. The NG80386 is a 100 pin QFP. These are usually soldered to the board, rather than socketed. Unlike a PLCC, you don't just pry these out of the socket. If you look carefully, you'll see that all 4 sides of the chip are held by those black retaining clips. You remove or swing those out of the way and the chip should be carefully lifted out. It's rather unusual to see a QFP socket in a consumer item--usually those are reserved for burn-in and testing gear. Be careful not to bend the pins!

L_Intel-NG80386SX-16.jpg
 
Assuming you could remove the CPU, bumping up the clock speed is an entirely different problem.

Since the CPU and bus speed are one and the same, increasing one also increases the other. Since everything on the board is tied to the bus speed, increasing that will also increase everything else. Early 386 boards were generally not very overclocking friendly, and were very much on the ragged edge of stability. Some motherboards also had BIOS code that either checked for overclocking and prevented the board from booting, or had timing critical code that would fall apart when run at faster speeds and cause the board to not boot, be unstable or have erratic behavior.

You can probably get away with bumping it up to 20 MHz, but 25 or 33 MHz most likely wouldn't work. If a clock doubled Cyrix would work in the board, you could get 40, 50 or 66 MHz, but the problem is the 16 bit bus. Both the 386DX and SX have tristate multiplexed buses, because Intel didn't give the CPU enough pins for all of the functions. The SX variant is further kneecapped with a 16 bit data bus and a 24 bit address bus. So not only do transactions over the bus take 2-4x as long, the CPU can only address 16 MB of RAM total.
 
Maybe a TI486SXLC2 might be made to work--clock-doubled with an 8KB cache. However, I haven't compared the pinout with the NG80386. They're pretty hard to come by, so do your homework.
 
This kind of CPU should be soldered directly to the motherboard, unlike earlier BIOS chips.
The black upper cover should be able to be opened from the side or removed its entirety.

Intel-i386-SX-80386SX-16-CPU.jpg
 
Last edited:
FParker said:
The reason I ask is because I'd like to replace it with one of those Cyrix "486" 386 processors (or the TI variant of said processor).
Click to expand...
Wouldn't work anyway. The SX has a 16-bit databus, the 386DX or 486DX has a 32-bit data bus. Those TIs @Chuck mentioned could work, if you manage to get one AND fix it to your board in one way.

But why would you do it? It would be like placing the engine of a ten year newer Ford model A into a Ford model T. You would need a lot of tinkering to make it fit and you won't be sure of the result. Why not obtaining a 486DX machine instead? Much faster and probably supports a bigger HDD.
 
Ruud said:
Wouldn't work anyway. The SX has a 16-bit databus, the 386DX or 486DX has a 32-bit data bus. Those TIs @Chuck mentioned could work, if you manage to get one AND fix it to your board in one way.

But why would you do it? It would be like placing the engine of a ten year newer Ford model A into a Ford model T. You would need a lot of tinkering to make it fit and you won't be sure of the result. Why not obtaining a 486DX machine instead? Much faster and probably supports a bigger HDD.
Click to expand...
I wish to thank everyone for the cornucopia of information re: my questions. I always learn more than anticipated when I come here.

Based on all this info, I will keep this computer stock and use it as it was intended from the factory. I will begin my search for a 486DX or even an early Pentium machine. Another computer never hurt anybody😂. I reckon a run-of-the-mill business oriented machine shouldn't run too much, right?

Thanks,
Forrest.
 
Apparently, the TI parts aren't hard to find---but note that while they have 5V tolerant I/O, they're 3.3V chips. So some jiggering with the power supply would also be required. Not saying that it can't be done--but hardly worth the effort.
 
Chuck(G) said:
Maybe a TI486SXLC2 might be made to work--clock-doubled with an 8KB cache. However, I haven't compared the pinout with the NG80386. They're pretty hard to come by, so do your homework.
Click to expand...

There's an even more rare IBM "486SLC" that has 16 kb of L1 cache. I have a 50 MHz variant, and it performs about as well as a real 486 at 25 MHz.

The problem with using CPUs with cache on them is that they also need the cache coherency circuitry. Without it, there's nothing to evict the cache as it becomes stale, so the system will start to act erratically and eventually crash. Companies that made 386 upgrades would put this on an interposer PCB with the new CPU.
 
You must log in or register to reply here.
Back
Top