INSTRUCTION DESCRIPTIONS

MOVEP

Move Peripheral Data

MOVEP

ister SSH is specified as a destination operand, the system stack pointer (SP) is prein- cremented by 1 before SSH is written. This allows the system stack to be efficiently extended using software stack pointer operations.

When a 56-bit accumulator (A or B) is specified as a source operand S, the accumulator value is optionally shifted according to the scaling mode bits S0 and S1 in the system status register (SR). If the data out of the shifter indicates that the accumulator extension register is in use and the data is to be moved into a 24-bit destination, the value stored in the destination is limited to a maximum positive or negative saturation constant to mini- mize truncation error. If a 24-bit source operand is to be moved into a 16-bit destination register D, the 8 MS bits of the 24-bit source operand are discarded, and the 16 LS bits are stored in the 16-bit destination register. Limiting does not occur if an individual 24-bit accumulator register (A1, A0, B1, or B0) is specified as a source operand instead of the full 56-bit accumulator (A or B). This limiting feature allows block floating-point opera- tions to be performed with error detection since the L bit in the condition code register is latched.

When a 56-bit accumulator (A or B) is specified as a destination operand D, any 24-bit source data to be moved into that accumulator is automatically extended to 56 bits by sign extending the MS bit of the source operand (bit 23) and appending the source oper- and with 24 LS zeros. Whenever a 16-bit source operand S is to be moved into a 24-bit destination, the 16-bit source is loaded into the LS 16 bits of the destination operand, and the remaining 8 MS bits of the destination are zeroed. Note that for 24-bit source operands both the automatic sign-extension and zeroing features may be disabled by specifying the destination register to be one of the individual 24-bit accumulator registers (A1 or B1).

Note: Unlike other MOVE-type instructions, if an AGU register (Mn, Nn, or Rn) is directly changed with MOVEP, the new contents will be available for use during the immediately following instruction. There is no instruction cycle pipeline delay associated with

MOVEP.

Restrictions: The following restrictions represent very unusual operations, which proba- bly would never be used but are listed only for completeness.

A MOVEP instruction used within a DO loop which specifies SSH as the source oper- and or LA, LC, SR, SP, SSH, or SSL as the destination operand cannot begin at the address LA–2, LA–1, or LA within that DO loop.

MOTOROLA

INSTRUCTION SET DETAILS

A - 221

Page 490
Image 490
Motorola DSP56000, 24-Bit Digital Signal Processor manual Movep