In the following, $xxx represents a memory location and %xxx represents a register.
dst represents a destination, src represents a source, lft represents the left-hand argument of a binary operation, rgt represents the right-hand argument of a binary operation, arg represents the only argument of a unary operation,
l $addr, %reg
st %reg, $addr
add %lft, %rgt, %dst
sub %lft, %rgt, %dst
mul %lft, %rgt, %dst
div %lft, %rgt, %dst
rem %lft, %rgt, %dst
sqrt %arg, %dst
In single-precision mode, there are 16 registers. In double-precision mode, there are 8 registers. In this mode, the register selector's top bit is ignored, but it should be set to zero.
Register selectors are 4 bits (a nibble) in length. Memory location selectors are 8 bits (a byte) in length.
add: 0001 llll rrrr dddd (%lft + %rgt -> %dst)
sub: 0010 llll rrrr dddd (%lft - %rgt -> %dst)
mul: 0011 llll rrrr dddd (%lft * %rgt -> %dst)
div: 0100 llll rrrr dddd (%lft ÷ %rgt -> %dst)
rem: 0101 llll rrrr dddd (%lft % %rgt -> %dst)
l: 0110 dddd ssss ssss ($src -> %dst)
st: 0111 ssss dddd dddd (%src -> $dst)
sqrt: 0000 0001 aaaa dddd (sqrt %arg -> %dst)
lz: 0000 0000 0001 dddd (0 -> %dst)
l1: 0000 0000 0010 dddd (1 -> %dst)
lp: 0000 0000 0011 dddd (pi -> %dst)
le: 0000 0000 0011 dddd (e -> %dst)
Short-form instructions:
1aaa aaaa 1bbb bbbb - two short-form instructions, a and b.
0*** **** 1*** **** - does not encode any short instructions
1*** **** 0*** **** - does not encode any short instructions
0*** **** 0*** **** - does not encode any short instructions
what it is