CPU Prices and value.

[Hugo Holden]

The two vintage computers I own are the IBM-5155 and the SOL-20. I thought it would be a good idea to get a spare CPU for them, the 8088 and the 8080 respectively. The more I read about these CPUs and how to program them, the more I come to admire them. Probably though the CPU's in my computers will unlikely require replacing.

I noticed on ebay that some crazy prices were being asked for original white ceramic Intel CPU's, nearly $1000 in some cases. There are a number of very economical epoxy package ones on ebay though.

So I did a little hunting around and for less than $30 each I was able to get the beautiful Japanese made white ceramic CPU's, the 8080 AF and the 8MHz 8088-2, shown in the attached photo. Which, for the complexity and elegance of such a part, I would deem a very fair price. Japanese manufacturing quality is generally outstanding.

Maybe, the original Intel chips might not even be as good. I came across this interesting remark in one of my assembly language books:

"With the 8080/8085 chips manufactured by Intel, the DAA instruction does not make a proper adjustment after subtraction. But with some 8080/8085 chips (nec for example) it does" ... Page 256, Introduction to 8080/8085 Assembly Language Programming, Fernandez & Ashley, Wiley 1981.

These sorts of parts are pretty beautiful and you can stare at them for quite a while, like Gollum's precious ring.

 

[Chuck(G)]

Those NEC 8080A chips might have extra value if they're the ones that have the DAA-after-subtract "feature". I recall that IMSAI bought a bunch of the early NEC 8080A CPUs only to discover the incompatibility with the Intel parts. They ended up using them on their diskette controller board set.

 

[Hugo Holden]

Those NEC 8080A chips might have extra value if they're the ones that have the DAA-after-subtract "feature". I recall that IMSAI bought a bunch of the early NEC 8080A CPUs only to discover the incompatibility with the Intel parts. They ended up using them on their diskette controller board set.

Chuck, presumably this is a one way incompatibility, in that any program that would run on the original Intel 8080 will run on the NEC 8080, but a program written that used the DAA after subtract, that would run on the NEC 8080, won't run on the original Intel IC is that correct ?

So I can expect if I replace the Intel 8080 in my Sol with the NEC chip, there should be no issues....is that correct ?

 

[Chuck(G)]

I think that's probably true--but it's hard to predict what will be in any program--and that's the issue. If you document the behavior of a chip, you can bet that someone, somewhere will depend on that being the absolute truth.

Subtract using DAA on Intel chips is usually gotten around by complement-addition, whereas the NEC chips allow for subtraction. Of course, the Z80 implemented the N flag so that DAA works either way.

I think the behavior was of enough concern that the 8088 made the DAA perform identically to the 8080 implementation and added a DAS operation to handle the subtraction issue.

Probably of more concern is that the boolean (AND OR XOR) on the Intel chip clear carry, but the NEC chip does not. I can see that could create a real problem with some code.