Unlock instant, AI-driven research and patent intelligence for your innovation.

Managing data flow in heterogeneous computing

A data flow, heterogeneous technology, applied in the direction of computing, electrical digital data processing, data processing input/output process, etc., can solve problems such as no data flow

Active Publication Date: 2019-05-21
QUALCOMM INC
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Existing heterogeneous computing runtimes do not have knowledge about data flow between stages

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Managing data flow in heterogeneous computing
  • Managing data flow in heterogeneous computing
  • Managing data flow in heterogeneous computing

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] Various embodiments will be described in detail with reference to the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts. References made to particular examples and implementations are for illustrative purposes and are not intended to limit the scope of the claims.

[0026] Various embodiments may include methods, systems and devices implementing such methods for managing heterogeneous computing by considering multiple declared operations of a program when allocating memory for operations executed across asynchronous processing devices data flow in the device. Apparatus and methods of various embodiments may include mapping data streams for declared operations between heterogeneous processing devices, determining data placement at computation start, allocating memory for data to reduce memory management operations needs, including data synchronization operations, and creating buf...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

Embodiments include computing devices, apparatus, and methods implemented by the apparatus for implementing data flow management on a computing device. Embodiment methods may include initializing a buffer partition of a first memory of a first heterogeneous processing device for an output of execution of a first iteration of a first operation by the first heterogeneous processing device on which afirst iteration of a second operation assigned for execution by a second heterogeneous processing device depends. Embodiment methods may include identifying a memory management operation for transmitting the output by the first heterogeneous processing device from the buffer partition as an input to the second heterogeneous processing device. Embodiment methods may include allocating a second memory for storing data for an iteration executed by a third heterogeneous processing device to minimize a number of memory management operations for the second allocated memory.

Description

Background technique [0001] Heterogeneous computing utilizes various computing elements to accelerate applications. Pipelining computations across multiple stages is a common practice in fields such as computer vision and machine learning. From a data flow perspective, it is common to encounter input / output (I / O) pipelining to transfer computation results from one stage to another, e.g. the output of operation A is used as the input of operation B. At each stage, each operation can be represented by multiple core functions, where each function represents a sequence of computations performed on a particular compute element. As an example, a storage unit may be modified by a central processing unit (CPU) function at a first stage and used as an input to a graphics processing unit (GPU) function at a second stage. Existing heterogeneous computing runtimes have no knowledge about the data flow between stages. As a result, unnecessary data is copied back and forth between memory...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F9/52G06F9/54H04N21/44
CPCG06F9/52G06F9/544G06F12/0811G06F12/084G06F12/0855G06F2212/283G06F2212/314G06F3/0613G06F3/0631G06F3/0656G06F3/0685
Inventor H·赵A·拉曼A·纳塔拉詹
Owner QUALCOMM INC