cpu32/INSTRUCTIONS.md

# RISC combined instruction set

goal: a semi-efficient insatruction set that offers few instructions that offer
the combined functions of conventional risc instuctions.

as such most instructions could be pipelined, I don't know how to feel about that

# Memory Operations
when accessing memory trough an address a consecutive pair of registers is used,
each consecutive pair of registers is given a name, for brevity a pair is 
always referred to as `A`. The possible pairs are:

- A = R1:R0
- B = R3:R2
- C = R5:R4
- D = R7:R6
- E = R9:R8
- F = R11:R10
- G = R13:R12
- H = R15:R14
- I = R17:R16
- J = R19:R18
- K = R21:R20
- L = R23:R22
- M = R25:R24
- N = R27:R26
- O = R29:R28
- P = R31:R30


### LD load value into register
post increment/decrement
syntax: `LD Rd, A+q`
action: `Rd = [A]; A += q`

pre increment/decrement
syntax: `LD Rd, q+A`
action: `A += q; Rd = [A]`

### LDI load immediate into register
syntax: `LDI Rd, imm`
action: `Rd = imm`

### LDIL load long immediate into register
syntax: `LDIL Rd, imm32`
action: `Rd = imm32`

### ST store value from register
post increment/decrement
syntax: `ST A+q, Rd`
action: `[A] = Rd; A += q`

pre increment/decrement
syntax: `ST q+A, Rd`
action: `A += q; [A] = Rd`

### STI store immediate into memory
post increment/decrement
syntax: `STI A+q, imm`
action: `[A] = imm; A += q`

pre increment/decrement
syntax: `STI q+A, imm`
action: `A += q; [A] = imm`

### STIL store long immediate into memory
post increment/decrement
syntax: `STIL A+q, imm32`
action: `[A] = imm32; A += q`

pre increment/decrement
syntax: `STIL q+A, imm32`
action: `A += q; [A] = imm32`


# Arithmetic operations

### ADD
### ADDC
### ADDI
### ADDIC

### SUB Subtract
### SUBC Subtract with carry
### SUBI Subtract immediate
### SBIC Subtract immediate with carry

### MLA Multiply and add
syntax: `MLA A, Ra, Rb, Rc`
action: `A = Rc + (Ra * Rb)`

### MLAI Multiply and add immediate
syntax: `MLAI A, Ra, Rb, imm`
action: `A = imm + (Ra * Rb)`

### MLIA Multiply immediate and add
syntax: `MLIA A, Ra, imm, Rc`
action: `A = Rc + (Ra * imm)`

### MLAC Multiply and add with carry
syntax: `MLAC A, Ra, Rb, Rc`
action: `A = F.C + Rc + (Ra * Rb)`

### MLAIC Multiply and add immediate with carry
syntax: `MLAIC A, Ra, Rb, imm`
action: `A = F.C + imm + (Ra * Rb)`

### MLIAC Multiply immediate and add with carry
syntax: `MLIAC A, Ra, imm, Rc`
action: `A = F.C + Rc + (Ra * imm)`


### SMLA Signed multiply and add
syntax: `SMLA A, Ra, Rb, Rc`
action: `A = Rc + (Ra * Rb)`

### SMLAI Signed multiply and add immediate
syntax: `SMLAI A, Ra, Rb, imm`
action: `A = imm + (Ra * Rb)`

### SMLIA Signed multiply immediate and add
syntax: `SMLIA A, Ra, imm, Rc`
action: `A = Rc + (Ra * imm)`

### SMLAC Signed multiply and add with carry
syntax: `SMLAC A, Ra, Rb, Rc`
action: `A = F.C + Rc + (Ra * Rb)`

### SMLAIC Signed multiply and add immediate with carry
syntax: `SMLAIC A, Ra, Rb, imm`
action: `A = F.C + imm + (Ra * Rb)`

### SMLIAC Signed multiply immediate and add with carry
syntax: `SMLIAC A, Ra, imm, Rc`
action: `A = F.C + Rc + (Ra * imm)`

### MUL, MULS, IMUL, MULI, etc.
can be macro'd with multiply and adds

### AND
### ANDI with immediate
### OR
### ORI with immediate
### XOR
### XORI with immediate

### NEG negate (two's complement)

### ROR Rotate right
### ROL Rotate left
### SHR Shift right
### SHRC Shift right through carry
### SHL Shift left
### SHLC Shift left through carry

# Conditional operations

### JZS Jump on zero set
syntax: `JZS imm32`

### JZC Jump on zero clear
### RJZS Relative jump on zero set
syntax: `RJZS imm`

### RJZC Relative jump on zero clear
### JNS Jump on negative set
### JNC Jump on negative clear
### RJNS Relative jump on negative set 
### RJNC Relative jump on negative clear 
### JCS Jump on carry set
### JCC Jump on carry clear
### RJCS Relative jump on carry set
### RJCC Relative jump on carry clear
### JVS Jump on overflow set
### JVC Jump on overflow clear
### RJVS Relative jump on overflow set
### RJVC Relative jump on overflow clear

# Stack manipulation

### CALL
could be implemented with push/jump
### RCALL
could be implemented with pop/jump

### PUSH
### POP


# Planned to haves

- 1-cycle shift and add
- 1-cycle intger division
- floating point arithmetic
- interrupts
- atomic compare-and-swap
- atomic load-link/store-conditional
- memory management
- memory-mapped registers
- sine and cosine functions, see [cordic][1]

[1]: https://en.wikipedia.org/wiki/CORDIC