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

Method for accelerating operations and accelerator apparatus

An accelerator and memory technology, which can be used in instruments, memory systems, biological neural network models, etc., and can solve problems such as overlapping weights and kernels

Pending Publication Date: 2021-01-29
AIMOTIVE KFT
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This arrangement may require storing redundant input data due to overlapping weight kernels

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
  • Method for accelerating operations and accelerator apparatus
  • Method for accelerating operations and accelerator apparatus
  • Method for accelerating operations and accelerator apparatus

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0073] In the following description, reference is made to the accompanying drawings, which show various embodiments by way of illustration. In addition, various embodiments will be described below by referring to several examples. It is to be understood that the embodiments may incorporate changes in design and construction without departing from the scope of the claimed subject matter.

[0074] figure 1 A schematic diagram of the neural network is shown. Neural networks are well-known systems used in artificial intelligence, machine learning, data mining, and other fields. Neural networks can be trained to recognize or classify or otherwise characterize elements in input data. A neural network usually includes at least two interconnected layers, including an input layer 101 and an output layer 103, in figure 1 are respectively denoted as "layer 0" and "layer L-1". The neural network may also include one or more hidden layers 105, in figure 1These are denoted as "Layer 1...

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

The present disclosure relates to a method for accessing and processing data by an accelerator apparatus. The method includes retrieving at least a part of input data to be processed by the accelerator apparatus, segmenting the input data into a plurality of adjacent input tiles, the input tiles having a pre-determined size, storing at least one of the plurality of adjacent input tiles in a data buffer of the accelerator apparatus, accessing data elements of the at least one of the plurality of adjacent input tiles stored in the data buffer by positioning an input window over the data elementsstored in the data buffer, to generate a plurality of input areas, wherein the input window is adjustable according to a set of parameters, and sequentially processing at least one of the plurality of input areas to at least partially process the at least one of the plurality of adjacent input tiles by the accelerator apparatus. The accelerator apparatus and a respective system are disclosed.

Description

technical field [0001] The present invention relates to methods, accelerator devices and systems for accessing and processing data by accelerator devices. In particular, the invention may relate to a method of operating on adjacent input blocks using an adjustable input window. Background technique [0002] Due to recent developments in computing technology, the operation of most equipment, machinery or infrastructure is entirely dependent on computer-based processing of various input data and / or computer-based evaluation of the processing status in order to generate information suitable for controlling equipment, machinery or infrastructure information. Typically, such processing and corresponding decision-making require a large number of complex computing tasks or operations, however, these computing tasks or operations are usually of a similar type. While the described tasks or operations may be performed by a general-purpose processing unit, such as a CPU or core of a ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06N3/04G06N3/063G06F12/0846
CPCG06F12/0846G06F2212/1016G06N3/063Y02D10/00G06N3/045G06F12/0207G06F12/0813
Inventor G·纳吉M·菲赫尔I·H·魏曼
Owner AIMOTIVE KFT