Intel Corporation

James Hadley

Matthew C Merten

Avinash Sodani

Alexandre J Farcy

Morris Marden

Ilhyun Kim

Srikanth T Srinivasan

Robert S Chappell

Jonathan P Douglas

Justin M Deinlein

Daniel J Deleganes

Iredamola Dammy Olopade

Joseph A McMahon

Ravi Rajwar

Sean Mirkes

Vijaykumar Balaram Kadgi

Laura A Knauth

Ammon J Christiansen

Yury N Ilin

Bambang Sutanto

Rahul R Kulkarni

Grace C Lee

Thomas Mullins

Fariborz Tabesh

David Lim

In one embodiment, a reservation station of a processor includes: a plurality of first lanes having a plurality of entries to store information for instructions having in-order dependencies; a variable latency tracking table including a second plurality of entries to store information for instructions having a variable latency; and a scheduler circuit to access a head entry of the plurality of first lanes to schedule, for execution on at least one execution unit, at least one instruction from the head entry of at least one of the plurality of first lanes. Other embodiments are described and claimed.

System, apparatus and method for a hybrid reservation station for a processor

A scheduler implementing a dependency matrix having restricted entries is disclosed. A processing device of the disclosure includes a decode unit to decode an instruction and a scheduler communicably coupled to the decode unit. In one embodiment, the scheduler is configured to receive the decoded instruction, determine that the decoded instruction qualifies for allocation as a restricted reservation station (RS) entry type in a dependency matrix maintained by the scheduler, identify RS entries in the dependency matrix that are free for allocation, allocate one of the identified free RS entries with information of the decoded instruction in the dependency matrix, and update a row of the dependency matrix corresponding to the claimed RS entry with source dependency information of the decoded instruction.

Scheduler implementing dependency matrix having restricted entries

A processor includes a processing unit including a storage module having stored thereon a physical reference list for storing identifications of physical registers that have been referenced by multiple logical registers, and a reclamation module for reclaiming physical registers to a free list based on a count of each of the physical registers on the physical reference list.

Physical reference list for tracking physical register sharing

A method and apparatus for providing fairness in a multi-processing element environment is herein described. Mask elements are utilized to associated portions of a reservation station with each processing element, while still allowing common access to another portion of reservation station entries. Additionally, bias logic biases selection of processing elements in a pipeline away from a processing element associated with a blocking stall to provide fair utilization of the pipeline.

Apparatus including a stall counter to bias processing element selection, and masks to allocate reservation unit entries to one or more processing elements

An apparatus is described having an out-of-order instruction execution pipeline. The out-of-order execution pipeline has a first circuit and a second circuit. The first circuit is to hold a pointer to physical storage space where information is kept that cannot yet be confirmed as being free of potential dependencies on the information. The second circuit is to hold the pointer if the pointer existed in the first circuit when a non speculative region of program code ended and upon retirement of a following speculative overwriter instruction originally coded to overwrite the information.

Copy-on-write buffer for restoring program code from a speculative region to a non-speculative region

Providing thread fairness by biasing selection away from a stalling thread using a stall-cycle counter in a hyper-threaded microprocessor

In one embodiment, the present invention includes a method for determining if an instruction of a first thread dispatched from a first queue associated with the first thread is stalled in a pipestage of a pipeline, and if so, dispatching an instruction of a second thread from a second queue associated with the second thread to the pipeline if the second thread is not stalled. Other embodiments are described and claimed.

Managing multiple threads in a single pipeline

A pipelined instruction decoder for a multithread processor including an instruction decode pipeline, a valid bit pipeline, and a thread identification pipeline in parallel together, with each having the same predetermined number of pipe stages. The instruction decode pipeline to decode instructions associated with a plurality of instruction threads. The valid bit pipeline to associate a valid indicator at each pipe stage with each instruction being decoded in the instruction decode pipeline. The thread identification pipeline to associate a thread-identification at each pipestage with each instruction being decoded in the instruction decode pipeline. The pipelined instruction decoder may further include a pipeline controller to control the clocking of each pipe stage of the instruction decode pipeline, the valid bit pipeline, and the thread identification pipeline. The pipeline controller may invalidate an entire thread of instructions, squeeze out invalid instructions, and/or conserve power by selectively stopping the clocking of pipestages.

Pipelined instruction decoder for multi-threaded processors

A multithread pipelined instruction decoder to clock, clear and stall an instruction decode pipeline of a multi-threaded machine to maximize performance and minimize power. A shadow pipeline shadows the instruction decode pipeline maintaining a the thread-identification and instruction-valid bits for each pipestage of the instruction decoder. The thread-id and valid bits are used to control the clear, clock, and stall of each pipestage of the instruction decoder. Instructions of one thread can be cleared without impacting instructions of another thread in the decode pipeline. In some cases, instructions of one thread can be stalled without impacting instructions of another thread in the decode pipeline. In the present invention, pipestages are clocked only when a valid instruction needs to advance in order to conserve power and to minimize stalling.

Method and apparatus for multi-thread pipelined instruction decoder

Growing community of inventors

Portland, OR, United States of America

James Hadley