Memory equipped with information retrieval function, method for using same, device, and information processing method

a technology of information retrieval and memory, applied in the field of memory equipped with information retrieval function, method for using same, device, information processing method, etc., can solve the problems of long overall waiting time, large amount of information processing, and difficulty in achieving the effect of increasing the required power consumption and complicating the peripheral circui

Inactive Publication Date: 2017-05-25
INOUE KATSUMI
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Benefits of technology

The invention provides a memory with an information search function that is applicable to big data and overcomes the challenges of conventional information processing. It leverages the concept of content-addressable memory (CAM) to perform parallel information searches while addressing its challenges such as circuit size and high power consumption. This results in a memory element that is more efficient and faster than conventional CPU-based information processing.

Problems solved by technology

However, moving the information data from a hard disk device to a semiconductor memory alone only offers expectation of 1 to 2 orders of speed improvement.
When a CPU performs information processing such as one to sequentially search for specific information on a memory, an amount of information processing becomes extremely large, leading to a long overall waiting time.
This is the bus bottleneck, an inevitable problem of the von Neumann-architecture computers.
Parallel processing (distributed processing) by each CPU is an attempt to solve these challenges, but complicates peripheral circuits to overly enlarge the system.
According to such algorithms, the CPU load is reduced during the search, but pre- and post-processing mandates complex information processing.
These and other inherent problems of the current computers arise from the fact that the CPU performs all the information processing, but from a different point of view, if the memory can find specific information by itself, the information processing discussed above will totally change.
However, content-addressable memories (CAMs) have challenges that they each needs a completely parallel comparison circuit which increases their overall circuit size, and that parallel circuits constituting the comparison circuit consume high current.
For this reason, utilization of the content-addressable memories (CAMs) has been limited to special applications where super fast search is essential as in communication routers.
This reduces the size of operation circuitry compared to providing an operation circuit for each memory, but deteriorates the operation efficiency as well as having a poor chip efficiency and undesirable cost issues.

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
  • Memory equipped with information retrieval function, method for using same, device, and information processing method
  • Memory equipped with information retrieval function, method for using same, device, and information processing method
  • Memory equipped with information retrieval function, method for using same, device, and information processing method

Examples

Experimental program
Comparison scheme
Effect test

example 1

[0073]FIG. 3 shows an example of publication search using a memory provided with an information search function.

[0074]In this example, each of word addresses from 1 to N may be assigned to one vocabulary term such as “information processing,”“information search,”“patent” or “CPU,” and each of vertical lines (columns) of word width n may be assigned to one publication. Also, the word addresses 1-N constitute one record as a whole, and the word addresses 1-N correspond to fields. In other words, if any of the terms including “information processing,”“information search,”“patent” and “CPU,” exist in one publication, “1” will be written into memory cells (fields) corresponding to the existing terms (“0” is omitted and likewise hereafter). Therefore, in this example, N vocabulary terms and n publications (n records) are registered in a database.

[0075]An example of searching for a particular publication from within this stored database will be discussed below.

[0076]For registered database...

example 2

[0084]FIG. 4 shows an example of exact match data search using the memory provided with the information search function.

[0085]In this example, an 8-bit data is assigned to a field, wherein a word address 10 is the most significant bit (MSB) of the field and a word address 17 is the least significant bit (MSB) of the field. Since the data has 8 bits, it may store data in 256 different ways, allowing 256 ways of arbitrary searches from an exact match to a size comparison by properly selecting the eight word addresses from the word address 10 to the word address 17. For example, when performing the exact match search for a data value “10,” which is “00001010” in binary, search may be performed eight times starting with the word address 10 as the most significant bit (MSB) and the word address 17 as the least significant bit (MSB).

[0086]As shown in the lower section of FIG. 4, the operation is sequentially performed from the MSB word address 10 in this example. In this case, “0” and “1”...

example 3

[0087]FIG. 5 shows an example of range data search using the memory provided with the information search function.

[0088]The above described the exact match search for the data value “10,” but when searching for data values equal to or greater than “10,” records with the data values equal to or greater than “16” may be detected by taking a logical sum four times from the MSB of the word address 10 to a word address 13, as shown in the figure.

[0089]Additionally for the lower 4 bits, records with the data values equal to or greater than “10” and less than “16” may be obtained by taking a logical product between a logical sum of word addresses 15 and 16, and a word address 14; and records with the data value equal to or greater than “10” may be obtained by taking a logical sum of the records with the data value equal to or greater than “10” and less than “16” and the previously obtained records with the data value equal to or greater than “16.” Further, by negating records with the data...

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

CPUs are not effective for search processing for information on a memory. Content-addressable memories (CAMs) are effective for information searches, but it is difficult to build a large-capacity memory usable for big data using the CAMs. A large-capacity memory may be turned into an active memory having an information search capability comparable to that of a content-addressable memory (CAM) by incorporating an extremely small, single-bit-based parallel logical operation unit into a common memory. With this memory, a super fast in-memory database capable of fully parallel searches may be realized.

Description

FIELD OF THE INVENTION[0001]The present invention relates to a memory provided with an information search function and the memory's usage, device and information processing method.BACKGROUND OF THE INVENTION[0002]With massive semiconductor memories becoming available at low prices, in-memory database technologies have become a focus of attention as fast-processing techniques for big data. One of the in-memory database technologies is used in data mining, where an enormous amount of data resides on a semiconductor memory (in-memory database) to speed up data accesses and, therefore, searches for information of interest. However, moving the information data from a hard disk device to a semiconductor memory alone only offers expectation of 1 to 2 orders of speed improvement.[0003]It is necessary to clarify challenges of current von Neumann-architecture computers before considering an essence of big data usage.[0004]In current computers, CPU's do all information processing regardless of...

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 & AuthorityApplications(United States)
IPC IPC(8): G06F17/30
CPCG06F17/30982G11C15/00G11C7/1006G11C15/043G06F16/90339G06F16/00G10L15/28G11C15/04
InventorINOUE, KATSUMI
OwnerINOUE KATSUMI