1 # RFC ls004 Shift-And-Add
5 * <https://libre-soc.org/openpower/sv/biginteger/analysis/>
6 * <https://libre-soc.org/openpower/sv/rfc/ls004/>
7 * <https://bugs.libre-soc.org/show_bug.cgi?id=960>
8 * <https://git.openpower.foundation/isa/PowerISA/issues/91>
9 * shift-and-add <https://bugs.libre-soc.org/show_bug.cgi?id=968>
21 **Books and Section affected**:
24 Book I Fixed-Point Shift Instructions 3.3.14.2
25 Appendix E Power ISA sorted by opcode
26 Appendix F Power ISA sorted by version
27 Appendix G Power ISA sorted by Compliancy Subset
28 Appendix H Power ISA sorted by mnemonic
36 shadduw - Shift and Add Unsigned Word
39 **Submitter**: Luke Leighton (Libre-SOC)
41 **Requester**: Libre-SOC
43 **Impact on processor**:
46 Addition of two new GPR-based instructions
49 **Impact on software**:
52 Requires support for new instructions in assembler, debuggers,
59 GPR, Big-manip, Shift, Arithmetic
64 Power ISA is missing LD/ST with shift, which is present in both ARM and x86.
65 Adding more LD/ST is too complex, a compromise is to add shift-and-add.
66 Replaces a pair of explicit instructions in hot-loops.
68 **Notes and Observations**:
70 1. `shadd` and `shadduw` operate on unsigned integers.
71 2. `shadduw` is intended for performing address offsets,
72 as the second operand is constrained to lower 32-bits
74 3. Both are 2-in 1-out instructions.
78 Add the following entries to:
80 * the Appendices of Book I
81 * Instructions of Book I added to Section 3.3.14.2
91 | 0-5 | 6-10 | 11-15 | 16-20 | 21-22 | 23-30 | 31 | Form |
92 |-------|------|-------|-------|-------|-------|----|----------|
93 | PO | RT | RA | RB | sm | XO | Rc | Z23-Form |
97 shift <- sm + 1 # Shift is between 1-4
98 sum[0:63] <- ((RB) << shift) + (RA) # Shift RB, add RA
99 RT <- sum # Result stored in RT
101 When `sm` is zero, the contents of register RB are multiplied by 2,
102 added to the contents of register RA, and the result stored in RT.
104 `sm` is a 2-bit bitfield, and allows multiplication of RB by 2, 4, 8, 16.
106 Operands RA and RB, and the result RT are all 64-bit, unsigned integers.
108 **NEED EXAMPLES (not sure how to embedd sm)!!!**
116 # Shift-and-Add Unsigned Word
120 | 0-5 | 6-10 | 11-15 | 16-20 | 21-22 | 23-30 | 31 | Form |
121 |-------|------|-------|-------|-------|-------|----|----------|
122 | PO | RT | RA | RB | sm | XO | Rc | Z23-Form |
126 shift <- sm + 1 # Shift is between 1-4
127 n <- (RB)[32:63] # Only use lower 32-bits of RB
128 sum[0:63] <- (n << shift) + (RA) # Shift n, add RA
129 RT <- sum # Result stored in RT
131 When `sm` is zero, the lower word contents of register RB are multiplied by 2,
132 added to the contents of register RA, and the result stored in RT.
134 `sm` is a 2-bit bitfield, and allows multiplication of RB by 2, 4, 8, 16.
136 Operands RA and RB, and the result RT are all 64-bit, unsigned integers.
139 The advantage of this instruction is doing address offsets. RA is the base 64-bit
140 address. RB is the offset into data structure limited to 32-bit.
154 Appendix E Power ISA sorted by opcode
155 Appendix F Power ISA sorted by version
156 Appendix G Power ISA sorted by Compliancy Subset
157 Appendix H Power ISA sorted by mnemonic
159 | Form | Book | Page | Version | mnemonic | Description |
160 |------|------|------|---------|----------|-------------|
161 | Z23 | I | # | 3.0B | shadd | Shift-and-Add |
162 | Z23 | I | # | 3.0B | shadduw | Shift-and-Add Unsigned Word |