Variable latency pipe for interleaving instruction tags in a microprocessor

A method for issuing instructions in a processor, comprising:

• upon issue of a first issued instruction from an issue queue, associating the first issued instruction with a first instruction tag of a plurality of instruction tags, wherein each one of the plurality of instruction tags comprises a valid bit, an operand type, an instruction tag identifier, and thread information, and wherein the issue queue includes an age array that tracks the relative age of each un-issued instruction in the issue queue, and wherein the issue queue is configured to store a plurality of un-issued instructions, wherein the plurality of un-issued instructions comprises at least three non-identical operand types;
• identifying a predetermined latency of the first issued instruction, based on one or more predetermined latency bits included in the first issued instruction, wherein the predetermined latency bits indicate a number of clock cycles required for execution of the first issued instruction;
• determining a first position of a plurality of positions in a pipe to store the first instruction tag based on the predetermined latency, the plurality of positions being ordered from a head position to a tail position, the pipe storing a plurality of instruction tags each associated with a respective issued instruction, the pipe storing the plurality of instruction tags separately from the plurality of un-issued instructions stored in the issue queue, wherein the pipe stores each of the plurality of instruction tags in a respective position based on the predetermined latency of each respective issued instruction, wherein the position of the instruction tag being determined is based on a predetermined latency of the issued instruction relative to the predetermined latency of each respective issued instruction, and wherein an instruction tag associated with a single-cycle issued instruction bypasses the latency pipe;
• storing the first instruction tag at the determined first position in the pipe;
• identifying a second instruction tag stored at the tail position in the pipe;
• broadcasting one of the plurality of instruction tags from the tail position of the pipe at each clock cycle, wherein at one clock cycle the second instruction tag is broadcast to at least the issue queue, a renaming unit, and a completion unit;
• upon receiving the second instruction tag, issuing a second issued instruction based on information from the age array and information from the second instruction tag such that multi-cycle instructions are prioritized over single-cycle instructions in the event of a conflict, and such that issuing the second issued instruction prioritizes older un-issued instructions in the issue queue whose dependencies correspond to one of the plurality instruction tags that has already been broadcast;
• removing the second instruction tag from the pipe after broadcast of the second instruction tag;
• shifting the plurality of instruction tags towards the tail position of the pipe at each clock cycle;
• upon issue of a third issued instruction from the issue queue, associating the third issued instruction with a third instruction tag;
• identifying a predetermined latency of the third issued instruction, based on one or more predetermined latency bits included in the third instruction, wherein the predetermined latency of the third issued instruction is shorter than the predetermined latency of the first issued instruction, and wherein the predetermined latency bits indicate a number of clock cycles required for execution of the third issued instruction;
• determining a second position of the plurality of positions in the pipe to store the third instruction tag based on the predetermined latency of the third issued instruction, wherein the second position in the pipe is nearer to the tail position than is the first position in the pipe;
• and
• storing the third instruction tag at the determined second position in the pipe.