Path: utzoo!attcan!uunet!jarthur!brutus.cs.uiuc.edu!samsung!cs.utexas.edu!asuvax!ncar!ico!ism780c!randvax!ucla-an!stb!gendep!craig Newsgroups: comp.sys.ibm.pc Subject: Re: 640K limit Message-ID: Date: 7 Feb 90 03:38:12 GMT References: <403@marvin.moncam.co.uk> Organization: General Depravity 1 213 447 1543 Lines: 50 emmo@moncam.co.uk (Dave Emmerson) writes: > Which other 2nd/3rd/4th generation processor has this b****y stupid > 'protected' mode? Granted, Motorola does not have this "protected" vs. "real" mode business, but Motorola also is not trying to be compatible back to day one. Many of the most annoying oddities of Intel's processors have to do with their insistence on being compatible with earlier chips; as we all know, the 286/386/486 will run 8086 object code unmodified in "real mode", and the 8086 itself was, while not entirely compatible with the 8080, very close; this is why 8086 segments are 64k long, after all. If Intel had designed the 8086 without the slightest thought of retaining compatibility or similarity to the 8080, the '86 would doubtless have been a very different chip. And Intel might have had more trouble establishing it in the marketplace against the 6809 and 68000 without that ready base of easily-converted 8080 CP/M software. In designing the 68000, Motorola left behind the 6800 entirely, opting for a totally new design that clearly, from day one, was looking ahead to 32-bit processing. After all, the 68000 has only a 24-bit address bus, but all the registers are 32 bits wide, and there are quite a few 32-bit instructions. So basically what it comes down to is that Motorola is working with design ideas that are about 6 years younger than Intel's. But what would happen if, say, the world turned to 64-bit microprocessors? Motorola's 32-bit linear address space couldn't be retained without (you guessed it) something resembling Intel segmentation. The newer Intel processors would be in pretty much the same boat, even in protected mode, for the same reasons. And eventually someone would complain about this "horribly limiting 4 GB segmentation limit" (!), and the chip makers would have to choose between backward compatibility vs. a linear address space, and probably would do just what Intel did with the 286/386... invent a processor with a backward-compatible "real mode" and a new "protected mode" with flat addressing. Look at it this way... at any given time in the last ten years, would it have been worth it to you to throw away all your software to upgrade to the next new processor? I've done it myself (in 1980 I had an Atari 800), but I don't really look forward to doing it again. "Real" vs. "protected" lets you have your cake and eat it too, more or less; it certainly smooths out the transition from old-chip to new-chip, since you effectively have both simultaneously in the same machine.