Intel Corporation

Bob Valentine

Lihu Rappoport

Ronny Ronen

Adi Yoaz

Mattan Erez

Stephan J Jourdan

Alexander D Peleg

Opher D Kahn

Doron Orenstien

Zeev Sperber

Bret L Toll

Christopher J Hughes

Alexander F Heinecke

Benny Eitan

Stephan Jourdan

Raanan Sade

Ron Gabor

Cristina S Anderson

Ido Ouziel

Mark J Buxton

Franck Sala

Amir Leibovitz

Yoav Almog

Mike Espig

An apparatus and method for efficient matrix alignment in a systolic array. For example, one embodiment of a processor comprises: a first set of physical tile registers to store first matrix data in rows or columns; a second set of physical tile registers to store second matrix data in rows or columns; a decoder to decode a matrix instruction identifying a first input matrix, a first offset, a second input matrix, and a second offset; and execution circuitry, responsive to the matrix instruction, to read a subset of rows or columns from the first set of physical tile registers in accordance with the first offset, spanning multiple physical tile registers from the first set if indicated by the first offset to generate a first input matrix and the execution circuitry to read a subset of rows or columns from the second set of physical tile registers in accordance with the second offset, spanning multiple physical tile registers from the second set if indicated by the second offset to generate a second input matrix; and the execution circuitry to perform an arithmetic operation with the first and second input matrices in accordance with an opcode of the matrix instruction.

Method and apparatus for efficient matrix alignment in a systolic array

In one embodiment, logic is provided to receive and execute a mask move instruction to transfer a vector data element including a plurality of packed data elements from a source location to a destination location, subject to mask information for the instruction, such that only portions of the plurality of packed data elements are transferred to the destination location. Other embodiments are described and claimed.

Vector mask memory access instructions to perform individual and sequential memory access operations if an exception occurs during a full width memory access operation

In one embodiment, the present invention includes logic to receive a permute instruction, first and second source operands, and control values, and to perform a permute operation based on an operation between at least two of the control values so that selected portions of the first and second source operands or a predetermined value can be stored into elements of a destination. Multiple permute instructions may be combined to perform efficient table lookups. Other embodiments are described and claimed.

Permute operations with flexible zero control

Methods and apparatus for using micro-op caches in processors are disclosed. A tag match for an instruction pointer retrieves a set of micro-op cache line access tuples having matching tags. The set is stored in a match queue. Line access tuples from the match queue are used to access cache lines in a micro-op cache data array to supply a micro-op queue. On a micro-op cache miss, a macroinstruction translation engine (MITE) decodes macroinstructions to supply the micro-op queue. Instruction pointers are stored in a miss queue for fetching macroinstructions from the MITE. The MITE may be disabled to conserve power when the miss queue is empty-likewise for the micro-op cache data array when the match queue is empty. Synchronization flags in the last micro-op from the micro-op cache on a subsequent micro-op cache miss indicate where micro-ops from the MITE merge with micro-ops from the micro-op cache.

Efficient method and apparatus for employing a micro-op cache in a processor

A memory cache bank prediction unit is provided for use in a processor having a plurality of memory cache banks. The memory cache bank prediction unit has an input port that receives an instruction. The memory cache bank prediction unit also has an evaluation unit, coupled to the input port, that predicts which of the plurality of memory cache banks is associated with the instruction.

Memory cache bank prediction

According to an embodiment of the present invention, a microprocessor includes an expanded logical register set that can be accessed by instructions including legacy opcodes and remapped addressing mode information. The known IA-32 instruction set is limited to accessing eight logical general integer registers. An IA-32 instruction can specify which of the eight logical general integer registers are to be accessed via 3-bit register identifier fields of the addressing mode information of the instruction. Each 3-bit register identifier can specify any of the eight logical general integer registers. An expanded logical register set (e.g., sixteen logical registers, thirty-two logical registers, sixty-four logical registers, etc.) can be accessed by remapping the addressing mode information to include at least four-bit register identifiers without defining new opcodes or defining additional instruction prefixes.

Method and apparatus to support an expanded register set

A cache bank prediction unit is provided for use in a processor having a plurality of cache memory banks. The cache bank prediction unit has an input port that receives an instruction. The cache bank prediction unit also has an evaluation unit, coupled to the input port, that predicts which of the plurality of cache banks is associated with the instruction.

Cache memory bank access prediction

Processors and methods having an expanded logical register set are disclosed. A processor includes may include Intel Architecture-32 (IA-32) instruction set decoding logic and an expanded logical register set. The expanded logical register set may include more than eight logical registers of a first type. An expanded register set decoding logic, coupled to said IA-32 instruction set decoding logic, may determine that an instruction includes an at least four-bit register identifier, the at least four-bit register identifier to specify one logical register of said expanded logical register set.

Growing community of inventors

Qiryat Tivon, Israel

Bob Valentine