Compaq Computer Corporation, Inc.

Advanced Micro Devices Corporation

Other

Advanced Micro Device, Inc.

Compaq Computer Corporation and Advanced Micro Devices, Inc.

John S Thayer

John Gregory Favor

Frederick Daniel Weber

Gary W Thome

Brian E Longhenry

Montgomery C McGraw

Mark W Welker

Paul R Culley

Gary J Piccirillo

Dale J Mayer

Patrick Lee Ferguson

John A Landry

Javier F Izquierdo

Roy E Thoma, Iii

Mustafa A Hamid

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU can be coupled either through a coprocessor bus or a local CPU bus to a conventional processor. The MEU employs vector registers, a vector ALU, and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers can be easily routed and combined at the vector ALU. The vector instructions employ special load/store instructions in combination with numerous operational instructions to carry out concurrent multimedia operations on the aligned operands.

System and method for conditional moving an operand from a source register to destination register

A routable operand and selectable operation processor multimedia extension unit is employed to draw lines in a video system using an efficient, parallel technique. A first series of integral y pixel values and error values are calculated according to Bresenham's line drawing algorithm. Then, subsequent pixels and error values are calculated in parallel based on the previously calculated values.

Line drawing using operand routing and operation selective multimedia extension unit

Load and store instructions which perform unpacking and packing of data bits in separate vector and integer cache storage

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU can be coupled either through a coprocessor bus or a local CPU bus to a conventional processor. The MEU employs vector registers, a vector ALU, and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers can be easily routed and combined at the vector ALU. The MEU may include a plurality of ALUs, registers partitioned into slots, and a decode unit for decoding an instruction specifying operands from any slot from one register and from a fixed slot of another register as well as different operations to be performed by the ALUs on the operands.

Decoding operands for multimedia applications instruction coded with

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU can be coupled either through a coprocessor bus or a local central processing unit (CPU) bus to a conventional processor. The MEU employs vector registers, a vector arithmetic logic unit (ALU), and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers can be easily routed and combined at the vector ALU. The vector instructions employ special load/store instructions in combination with numerous operational instructions to carry out concurrent multimedia operations on the aligned operands.

Microprocessor modified to perform inverse discrete cosine transform

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU may be coupled either through a coprocessor bus or a local CPU bus to a conventional processor. The MEU employs vector registers, a vector ALU, and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers may be easily routed and combined at the vector ALU. The vector instructions employ special load/store instructions in combination with numerous operational instructions to carry out concurrent multimedia operations on the aligned operands. In one embodiment, an arithmetic logic unit may be partitioned into at least two logic portions. A first logic portion may be coupled to receive a first operand from a fixed slot of a first register and a second operand from any slot of a second register. A second logic portion may be coupled to receive a third operand from a fixed slot of the first register and a fourth operand from any slot of the second register. The first logic portion may perform an arithmetic operation dissimilar from the second logic portion.

Computer having multimedia operations executable as two distinct sets of

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU can be coupled either through a coprocessor bus or a local CPU bus to a conventional processor. The MEU employs vector registers, a vector ALU, and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers can be easily routed and combined at the vector ALU. The vector instructions employ special load/store instructions in combination with numerous operational instructions to carry out concurrent multimedia operations on the aligned operands. In one embodiment multiple ALUs may each receive one operand from a fixed source register slot location, where the fixed slot location may be different for each ALU. The operand routing may provide another operand from any source register slot location for another input to each respective ALU.

Apparatus for routing one operand to an arithmetic logic unit from a

A multimedia extension unit (MEU) is provided for performing various multimedia-type operations. The MEU can be coupled either through a coprocessor bus or a local CPU bus to a conventional processor. The MEU employs vector registers, a vector ALU, and an operand routing unit (ORU) to perform a maximum number of the multimedia operations within as few instruction cycles as possible. Complex algorithms are readily performed by arranging operands upon the vector ALU in accordance with the desired algorithm flowgraph. The ORU aligns the operands within partitioned slots or sub-slots of the vector registers using vector instructions unique to the MEU. At the output of the ORU, operand pairs from vector source or destination registers can be easily routed and combined at the vector ALU. The vector instructions employ special load/store instructions in combination with numerous operational instructions to carry out concurrent multimedia operations on the aligned operands. In one embodiment, multiple ALUs may each receive one operand from a fixed source register slot location, where the fixed slot location may be different for each ALU. The operand routing may provide another operand from any source register slot location for another input to each respective ALU.

System and method for routing one operand to arithmetic logic units from

An apparatus and a method perform an N-point Fast Fourier Transform (FFT) on first and second arrays having real and imaginary input values using a processor with a multimedia extension unit (MEU), wherein N is a power of two. The invention repetitively sub-divides the N-point Fourier Transform into N/2-point Fourier Transforms until only a 2-point Fourier Transform remains. Next, it vector processes the 2-point Fourier Transform using the MEU and cumulates the results of the 2-point Fourier Transforms from each of the sub-divided N/2 Fourier Transforms to generate the result of the N-point Fourier Transform.

Enhanced fast fourier transform technique on vector processor with

A routable operand and selectable operation processor multimedia extension unit is employed to motion compensate MPEG video using improved vector processing. A vector processing unit executes an add and divide instruction that adds two vector registers and divides the result in a single instruction. This is implemented through loading a first vector register with a first plurality of elements from a source block. A second vector register is then loaded with a second plurality of elements that are adjacent to the first plurality of elements. The add and divide instruction is then executed on the first and second vector registers, yielding an interpolated source element that is stored in a resultant vector register.

MPEG motion compensation using operand routing and performing add and

System and method for performing an accumulate operation on one or more

A signal represented as a matrix of input values is adaptively filtered using a processor with a multimedia extension unit. A plurality of coefficients are loaded into a first vector register and input values are loaded into a second vector register. Next, for each of the coefficients in the first vector register, (1) a single cycle vector multiply-accumulate operation is performed between the selected coefficient and the input values stored in the second vector register and the result is stored in a third register; (2) a partition-shift on input values in the second vector register is then performed and a new input value is then moved into the second vector register. After each of the four loaded coefficients have been processed, the results are saved and the operation is repeated until all input values in the matrix have been processed.

Enhanced adaptive filtering technique

System and method for conditionally moving an operand from a source

System and method for routing operands within partitions of a source

An apparatus and a method for massaging audio signal perform interpolation, dynamic filtering, and panning on the audio signal represented as a matrix of input values. In the interpolation process, the input values are loaded into first and second vector registers, while fractional coefficients are loaded into a third vector register. Next, the first vector register is subtracted from the second vector register. Additionally, in a single operation, the routine performs a vector multiply operation between the second and third registers and accumulates the result of the vector multiply operation in the second register. The results are saved and the process is repeated until all input values in the matrix have been processed. In the dynamic filtering process, after the data loading step, for each slot in said vector register, the routine performs a multiply operation between the filter coefficient and the slot of the vector register and accumulates the result of the multiply operation in the slot of the second register in a single clock cycle while it retains data of the remaining slots in the vector register in the same clock cycle. The results are saved and the process is repeated until all input values in the matrix have been processed. In the stretching process, after loading data in the appropriate vector register, the routine copies the content of each slot of the vector register into consecutive pair of slots on a second vector register and when the second vector register is full, copies the content of each of the remaining slots in the first vector register into consecutive pairs of slots on a third register. In the panning process, the routine performs a vector multiply operation between the first vector register and a coefficient vector register for each slot in the first vector register. This vector multiply operation is preferably a 32-bit vector multiply operation which is broken down into a low order extended precision multiply accumulate operation and a high order extended precision multiply accumulate operation.

Growing community of inventors

Houston, TX, United States of America

John S Thayer