To: CHIPDIR-L@fatcity.com From: Jaap van Ganswijk Subject: 6812 useful Cc: Bcc: X-Attachments: In-Reply-To: References: At 08:47 1998-06-03 -0800, Keith Welker wrote: >By the way in response to Jaap's comments on 'SEMICONDUCTOR >BUSINESS DOWN TURN WORLD WIDE', I think that the 68HC12 >is a very useful processor. The 6812 has clearly been designed fully without the necessary compiler complexity in mind. Therefore (if it has deliberately been designed that way) it was designed for assembler programmers, which is clearly not of this time. Otherwise the processor designers at Motorola are really completely out of touch with reality (and have for example not even read the 1985 book 'Computer Architecture a Quantitive Approach by Hennessy and Paterson' yet, which is THE standard on CPU design. >With it (the B32 version): 16 bit, 32k flash EEPROM, 1k SRAM, >768 byte EEPROM, and up to 64 i/o lines, it is very useful to me. They could also have put that in a 6805, 6809, 6811, 6816, 68332, 88000, Coldfire, PowerPC and whatever part of the chaotic processor landscape they have created over time. The 6812 seems to be a pet project of some aging designers that are completely out of touch with reality and with the other processor design teams within Motorola. Instead of adding compiler unfriendly options like auto incrementing and 8-bit minimum and maximum calculations (!), they should have augmented the 6811 instruction set with fully 16 bits and (rudimentary perhaps) 32 bits instructions. Just a 32 bits accumulator and a word swap instruction combined with 16 bits carry instructions would have been great. And why a carry instruction doesn't propagate the Z-flag is something I have never understood... Clearly a case of never having consulted a good compiler writer. And let's not even start about real orthogonality... (Like adding a 'compare with carry', for example...) Several years before the 6812 was designed I have put a list of these issues on the 6811 mailing list, but the Motorola people seem to be very selective in reading that list. Processor design isn't easy, but the ostrich way of putting your hand in the sand, just isn't the way.