# US 2019/0171448 A1 · Stream processor with low power parallel matrix multiply pipeline

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0171448 A1
- Document ID: US20190171448A1
- Status: Published application
- Kind: A1
- Application number: 15/855,637
- Filed: 2017-12-27
- Published: 2019-06-06
- Earliest claimed: 2017-12-01
- Assignees: Advanced Micro Devices, Inc.
- Applicants: Advanced Micro Devices, Inc.
- Inventors: Jiasheng Chen; Yunxiao Zou; Michael J. Mantor; Allen Rush
- Web page: https://patentgrep.com/patent/US20190171448A1

## CPC

- G06F 9/3867: {using instruction pipelines}
- G06F 7/5443: {Sum of products (for applications thereof, see the relevant places, e.g. G06F17/10, H03H17/00)}
- G06F 9/3001: {Arithmetic instructions}
- G06F 9/30036: {Instructions to perform operations on packed data, e.g. vector, tile or matrix operations}
- G06F 9/30101: {Special purpose registers}
- G06F 9/3013: {according to data content, e.g. floating-point registers, address registers}
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Systems, apparatuses, and methods for implementing a low power parallel matrix multiply pipeline are disclosed. In one embodiment, a system includes at least first and second vector register files coupled to a matrix multiply pipeline. The matrix multiply pipeline comprises a plurality of dot product units. The dot product units are configured to calculate dot or outer products for first and second sets of operands retrieved from the first vector register file. The results of the dot or outer product operations are written back to the second vector register file. The second vector register file provides the results from the previous dot or outer product operations as inputs to subsequent dot or outer product operations. The dot product units receive the results from previous phases of the matrix multiply operation and accumulate these previous dot or outer product results with the current dot or outer product results.

## Claims

### Claim 1

A system comprising:
a first vector register file; and
a first execution pipeline coupled to the first vector register file, wherein the first execution pipeline comprises a plurality of dot product units, and wherein each dot product unit of the plurality of dot product units is configured to:
calculate a plurality of products of elements of a first set of operands and corresponding elements of a second set of operands; and
calculate a sum of an accumulation input and the plurality of products, wherein the sum is an output of the dot product unit.

### Claim 2 (depends on 1)

The system as recited in claim 1, wherein the system is configured to read the first and second sets of operands from the first vector register file and provide the first and second sets of operands to the first execution pipeline.

### Claim 3 (depends on 1)

The system as recited in claim 1, wherein the system further comprises a second vector register file, wherein the system is configured to read a plurality of accumulation inputs from the second vector register file and provide the plurality of accumulation inputs to the first execution pipeline.

### Claim 4 (depends on 3)

The system as recited in claim 3, wherein each dot product unit is further configured to write the output to the second vector register file, wherein the output of a previous dot product operation is the accumulation input which is added to the sum for a current dot product operation.

### Claim 5 (depends on 1)

The system as recited in claim 1, wherein the system further comprises a second execution pipeline, wherein the second execution pipeline is configured to perform operations in parallel with dot product operations being performed by the first execution pipeline.

### Claim 6 (depends on 5)

The system as recited in claim 5, wherein the system is further configured to:
read the first and second sets of operands from the first vector register file and store the first and second sets of operands in the storage elements in a first cycle; and
read a third set of operands from the first vector register file in a second cycle and provide the third set of operands to the second execution pipeline.

### Claim 7 (depends on 1)

The system as recited in claim 1, wherein the first set of operands are rows of a first matrix and wherein the second set of operands are columns of a second matrix, and wherein the plurality of dot product units are configured to multiply the first matrix by the second matrix.

### Claim 8

A method comprising:
calculating a plurality of products of elements of a first set of operands and corresponding elements of a second set of operands; and
calculating a sum of an accumulation input and the plurality of products, wherein the sum is an output of a dot product unit.

### Claim 9 (depends on 8)

The method as recited in claim 8, further comprising reading the first and second sets of operands from a first vector register file and provide the first and second sets of operands to a first execution pipeline.

### Claim 10 (depends on 8)

The method as recited in claim 8, further comprising reading a plurality of accumulation inputs from a second vector register file and provide the plurality of accumulation inputs to a first execution pipeline.

### Claim 11 (depends on 10)

The method as recited in claim 10, further comprising writing the output to the second vector register file, wherein the output of a previous dot product operation is the accumulation input which is added to the sum for a current dot product operation.

### Claim 12 (depends on 8)

The method as recited in claim 8, further comprising performing operations on a second execution pipeline in parallel with dot product operations being performed by a first execution pipeline.

### Claim 13 (depends on 12)

The method as recited in claim 12, further comprising:
reading the first and second sets of operands from a first vector register file and storing the first and second sets of operands in storage elements in a first cycle; and
reading a third set of operands are read from the first vector register file in a second cycle and providing the third set of operands to the second execution pipeline.

### Claim 14 (depends on 8)

The method as recited in claim 8, wherein the first set of operands are rows of a first matrix and wherein the second set of operands are columns of a second matrix, and wherein the method further comprising multiplying the first matrix by the second matrix with a plurality of dot product units.

### Claim 15

An apparatus comprising:
a plurality of vector register files; and
a plurality of execution pipelines coupled to the plurality of vector register files;
wherein the apparatus is configured to:
calculate a plurality of products of elements of a first set of operands and corresponding elements of a second set of operands; and
calculate a sum of an accumulation input and the plurality of products, wherein the sum is an output of the dot product unit.

### Claim 16 (depends on 15)

The apparatus as recited in claim 15, wherein the apparatus is further configured to read the first and second sets of operands from a first vector register file and provide the first and second sets of operands to a first execution pipeline.

### Claim 17 (depends on 16)

The apparatus as recited in claim 16, wherein the apparatus is configured to read a plurality of accumulation inputs from a second vector register file and provide the plurality of accumulation inputs to the first execution pipeline.

### Claim 18 (depends on 17)

The apparatus as recited in claim 17, wherein the apparatus is configured to write the output to the second vector register file, wherein the output of a previous dot product operation is the accumulation input which is added to the sum for a current dot product operation.

### Claim 19 (depends on 15)

The apparatus as recited in claim 15, wherein the apparatus is further configured to perform operations on a second execution pipeline in parallel with dot product operations being performed by a first execution pipeline.

### Claim 20 (depends on 19)

The apparatus as recited in claim 19, wherein the apparatus is further configured to:
read the first and second sets of operands from the first vector register file and store the first and second sets of operands in storage elements in a first cycle; and
read a third set of operands from the first vector register file in a second cycle and provide the third set of operands to the second execution pipeline.
