How can we know if we have "similar" hardware when you provided zero information on your mainboard

i've seen people try. its not worth the effort. That 20mhz on the chipset, is usually a strict rating. Not that identical chipsets could possibly run at 25mhz, but during testing, they found out it could only go up to 20mhz. So running it at 25 is an exercise in wasting your time.

if the chip worked at 25, they would have sold it as such.

luckybob wrote on 2021-03-06, 00:49:

i've seen people try. its not worth the effort. That 20mhz on the chipset, is usually a strict rating. Not that identical chipsets could possibly run at 25mhz, but during testing, they found out it could only go up to 20mhz. So running it at 25 is an exercise in wasting your time.

if the chip worked at 25, they would have sold it as such.

My VL82CPCAT-20QC runs just fine at 25MHz. I have yet to find any DRAM that can handle 0-wait at that speed, but with 1 wait it's rock solid. VLSI never made a 25MHz marked version of that chipset FWIW

maxtherabbit wrote on 2021-03-06, 00:55:

luckybob wrote on 2021-03-06, 00:49:

i've seen people try. its not worth the effort. That 20mhz on the chipset, is usually a strict rating. Not that identical chipsets could possibly run at 25mhz, but during testing, they found out it could only go up to 20mhz. So running it at 25 is an exercise in wasting your time.

if the chip worked at 25, they would have sold it as such.

My VL82CPCAT-20QC runs just fine at 25MHz. I have yet to find any DRAM that can handle 0-wait at that speed, but with 1 wait it's rock solid. VLSI never made a 25MHz marked version of that chipset FWIW

I agree with luckybob, saw a topic here or vcfed or someplace last year where one was trying diff Xtals and found they could only get up to about 22mhz (on a 20Mhz rated chipset) before instability.
Think you just got lucky Max and found a board able to do it.

Mouldotron wrote on 2021-03-06, 00:01:

Hiya

I recently picked up a motherboard with a 20mhz Harris 286 in the slot. I also own a ton of 25mhz Harris 286 CPU's. I guess I can put a 25mhz chip in a 20mhz board but it'll still run at 20mhz. The other IC's say 20mhz on them. There is a 40mhz crystal as you'd expect, but has anyone with similar hardware tried to change it for a 50mhz crystal in order to use a 25mhz CPU? Just curious.

If you have a lot of Harris 25Mhz you should consider selling a few and use the money towards buying a 25Mhz (from the factory) capable board. Just my opinion...

Thanks for the replies everyone and sorry I've given little info on the mainboard, truth is I don't know what it is, there is no maker/brand on it.

Yes I guess I'll leave this board alone then as it works fine and keep looking for a 25. I've never come across one for sale. Would love one as the 286 is probably my favorite PC architecture, I ran one as my main PC for years in the 90's and did so much with it. Having a 25 would be pretty cool, but I've only got CPUs (ten of them).

Thanks again.

Most Harris 20mhz run fine at 25. Tested a few. Also swapped the oscillator from 40 to 50Mhz and found no problems. Memory runs very hot and during testing I've used 12cm fan to cool the area.

I have a plan to upgrade the oscillators on one other motherboard to go to 25Mhz CPU.

I've got two 16MHz rated chipsets (Headland HT12/A and a VLSI VL82C200-FC) that run perfectly happy at 20MHz. They'll both go even further than that; the HT12/A to 24MHz and the VLSI to 25MHz (but, like maxtherabbit mentions, RAM becomes the limiting factor at that point).

I don't see why you can't try it. 20MHz was fairly top-end for 286 designs; there's not many that are actually rated for 25MHz operation as per their datasheets.

The big issue is whether or not they'll run at 0 wait states at that speed, if they don't, then there's not much point in doing so - my benchmarks show a 25MHz 286 running with 1-wait state is slower (quite a bit, in some aspects) than a 20MHz processor at 0-ws:

https://www.target-earth.net/wiki/doku.php?id … rboard_shootout

Harris originally planned on offering chips up to 33mhz but had too low of yields

Would have been quite interesting having a 286 that ran circles around many 386 machines

rmay635703 wrote on 2021-03-06, 15:35:

Harris originally planned on offering chips up to 33mhz but had too low of yields

Would have been quite interesting having a 286 that ran circles around many 386 machines

+1

The 80286 was a "limited", but serious design.
- Limited in so far, because the 80386 essentially was considered a real mainframe on a chip in these days.
That being said, the 80286 had qualities that laid the foundation for the 80386 and later generations.

- The 286 featured an integrated Memory Management Unit (MMU)
- It had virtual memory (1GB)
- Was multi-tasking capable
- Had support for exception handling, maskable interrupts (?)
- It introduced a priority scheme (ring scheme)
- Memory-protection based on segmentation
- Variable segment size (64KB max)
- Address calculations (such as base+index) were handled by a dedicated unit, not via ALU

In other words, it was a professional unit.
If it wasn't for the IBM PC and the desire to multi-task MS-DOS applications, it would have had a much better reputation.

What many people fail to understand, is, that the 286 was released in 1982 - just one year after the IBM Model 5150.
In that time frame, the IBM was totally irrelevant.
There were other MS-DOS Compatibles around, also.
It wasn't until a few years later that PC DOS had any real value.

So in simple words, development of the 286 had started earlier, maybe in 1979-1981.
At this point, the IBM PC wasn't even finished, either. It wasn't even clear what OS the PC should run.
CP/M was an option at the time. 😀

if you want a fast 286, get a 386sx 😀 I have a couple of harris 25mhz cpus but had some issues getting them going with oscillator swaps. I suspect the board layout was too noisy for 25mhz with its support chips and such and how it ranc traces.

I have benchmarked a 286 vs a 386sx, both at 16MHz, both on VLSI200 series chipset. The 286 was about 100 dhrystones faster

I do have several 386's 😀

Always felt like the SX was a bit of a cheap-out, the DX being the proper 386 and the SX was their first foray into making budget versions (eg Celeron).

I used to have two virtually identical systems, both with AMD 40mhz, one being a SX and the other a DX. The latter ran circles around the former...

-

Apologies in advance, I don't mean to hijack this thread but my question, although separate, seems somewhat related and maybe a bit overkill to create an entirely new thread for.

I have a 286/16 board where the CPU is socketed in a PLCC socket. I also have a 25Mhz 286 that I'd like to use on this board instead. I know the oscillator needs to be replaced (from 32Mhz to 50Mhz) to support the faster CPU, however, if I replace the 16Mhz vs with the 25Mhz, the system should still work I assume, it would just run at 16Mhz? Reason, I just want to test is the 25Mhz chip is actually working.

Yes, replacing the 16MHz 286 with a Harris 25MHz chip, but leaving everything else as-is should work absolutely fine. All 286 PLCC processors have exactly the same core and exactly the same pin-out and voltage requirements.

In my experience Intersil CPUs tend to overclock better than Harris ones.
They are the same design really, but wonder if Intersils were produced on improved fabrication process ...

pan069 wrote on 2021-03-12, 22:42:

Apologies in advance, I don't mean to hijack this thread but my question, although separate, seems somewhat related and maybe a bit overkill to create an entirely new thread for.

I have a 286/16 board where the CPU is socketed in a PLCC socket. I also have a 25Mhz 286 that I'd like to use on this board instead. I know the oscillator needs to be replaced (from 32Mhz to 50Mhz) to support the faster CPU, however, if I replace the 16Mhz vs with the 25Mhz, the system should still work I assume, it would just run at 16Mhz? Reason, I just want to test is the 25Mhz chip is actually working.

megatron-uk wrote on 2021-03-12, 22:50:

Yes, replacing the 16MHz 286 with a Harris 25MHz chip, but leaving everything else as-is should work absolutely fine. All 286 PLCC processors have exactly the same core and exactly the same pin-out and voltage requirements.

Yes exactly !

pan069: If you do not have a PLCC chip puller be very careful as you can crack the socket or ruin the cpu, I have used a dental type tool to gently lift em out before getting a real puller but even it could do damage if not slowly worked one side then the other and back and forth....

@megatron-uk @Horun Thanks guys. I actually have a few of chip these pullers, never fully understood how they were to be used but it makes sense now! 😀