Note: The MODE indication on each node shows the node’s state from the previous cycle, not the current cycle. For example, all nodes show IDLE on the first cycle after startup.
JMP/JEZ/JNZ/JGZ/JLZ <label>JRO <literal>MOV <literal|ACC|NIL> <ACC|NIL>ADD <literal>SUB <literal>NEGSWPSAVNOPThese instructions always execute in one RUN cycle.
JRO <port>MOV <port> <ACC|NIL>ADD <port>SUB <port>These instructions block in the READ state until a value is available on port, then execute in one RUN cycle.
MOV <literal|ACC|NIL> <port>This instruction places a value on port, blocks in the WRTE state for at least one cycle, then executes in one RUN cycle when the value is consumed by another node.
MOV <port1> <port2>This instruction blocks until a value is available on port1, blocks in the WRTE state for at least one cycle, then executes in one RUN cycle when the value is consumed by the node on the other end of port2.
Writes take two cycles because values are written to ports, not to other nodes. The destination node cannot read the value until after the source node has written to the port, and the source node cannot continue execution until the value is read and the port is empty again.
Note: MOV <port1> <port1> is perfectly valid and also takes two cycles.
Stack Memory Nodes are the exception to this rule. A Stack Memory Node can execute one write and an unlimited number of reads (in that order) per cycle.
JRO <0|NIL><label>: JMP <label>JRO ACC<label>: JEZ/JNZ/JGZ/JLZ <label>Exercise: What happens when you JRO UP/DOWN/LEFT/RIGHT and the input value is zero? How might that be useful?