IF

Skip over the next instruction unless a condition is true according to the CPU flags.

Opcode	Bytes	Cycles	Form	Satisfied when...
$02	1	1	if zero	ZF=1
$03	1	1	if not zero	ZF=0
$04	1	1	if negative	NF=1
$05	1	1	if not negative	NF=0
$06	1	1	if overflow	OF=1
$07	1	1	if not overflow	OF=0
$08	1	1	if less unsigned	CF=0
$09	1	1	if not less unsigned	CF=1
$0a	1	1	if greater unsigned	CF=1 and ZF=0
$0b	1	1	if not greater unsigned	CF=0 or ZF=1
$0c	1	1	if less	NF≠OF
$0d	1	1	if not less	NF=OF
$0e	1	1	if greater	ZF=0 and NF=OF
$0f	1	1	if not greater	ZF=1 or NF≠OF

Control flag	Output
skip (SF)	If the condition is not satisfied, then SF is 1; otherwise it is 0.

Notes

• If the condition is false, then the subsequent instruction will be skipped by setting the skip flag (SF).

• Skipped instructions only require 1 CPU cycle.

Examples

math::sign()

Here's how the Hybrix compiler implements math::sign(), which returns 1 if i:0 is positive, -1 if it is negative, or returns 0 if i:0 is zero. Since move does not alter the flags, a single compare operation can set the flags for two different if instructions:

    compare i:0, 0
    if greater
    move i:0, 1
    if not greater
    shift right signed i:0, 31

kernel::memset_byte()

if instructions are very often used with jump to implement loops. Here's an example from the Hybrix framework:

module kernel
  . . .
  func memset_byte(target: int, value: byte, num_bytes: int)
    chombit
  end func
  . . .
end module

Here, if jumps out of the loop when the target register reaches target_end:

@kernel.kernel.memset_byte:
    push fp
    move fp, sp
    add sp, 8

    # i:-17 arg_target
    # b:-13 arg_value
    # i:-12 arg_num_bytes
    # i:-8 return ip
    # i:-4 fp

    # i:0 target
    # i:4 target_end

    move i:0, i:-17

    move i:4, i:-17
    add i:4, i:-12

@kernel.kernel.memset_byte.l_2:
    compare i:0, i:4
    if not less
    jump @kernel.kernel.memset_byte.l_3

    store [i:0], b:-13
    add i:0, 1

    jump @kernel.kernel.memset_byte.l_2

@kernel.kernel.memset_byte.l_3:
    add sp, -8
    pop fp
    pop ip
    fill 4