9-bit literal to 32-bit value extension

jonty

Here's a question for those who really know the fine details of assembler programming on the Propeller:

When an instruction uses a 9-bit literal value in the s field and the instruction operates on a 32-bit value then is the 9-bit literal always zero extended before it is used by the instruction?

The manual V1.01 explicitly says that ABS and ABSNEG both zero extend. But is this a general rule for all arithmetic and logical operations like ADD, SUB, AND etc?

For example, does ANDN take a 9-bit literal, zero extend it to 32-bits, negate (so that the top 23 bits are now all 1's), then bitwise AND with d?

Thanks in advance
Jonty

Post Edited (jonty) : 12/21/2006 9:24:49 PM GMT

cgracey

jonty said...


When an instruction uses a 9-bit literal value in the s field and the instruction operates on a 32-bit value then is the 9-bit literal always zero extended before it is used by the instruction?


YES

The manual V1.01 explicitly says that ABS and ABSNEG both zero extend. But is this a general rule for all arithmetic and logical operations like ADD, SUB, AND etc?

Just imagine that instead of using an S register, you are using a nine-bit literal that is zero-extended. What the instruction does to the value is constant, regardless of whether it was from a register or a nine-bit zero-extended literal.

For example, does ANDN take a 9-bit literal, zero extend it to 32-bits, negate (so that the top 23 bits are now all 1's), then bitwise AND with d?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

jonty

Thanks very much for the quick reply. Answered my question very nicely.

jonty