[go: up one dir, main page]

CN114741021A - Storage and computing integrated chip - Google Patents

Storage and computing integrated chip Download PDF

Info

Publication number
CN114741021A
CN114741021A CN202210406389.4A CN202210406389A CN114741021A CN 114741021 A CN114741021 A CN 114741021A CN 202210406389 A CN202210406389 A CN 202210406389A CN 114741021 A CN114741021 A CN 114741021A
Authority
CN
China
Prior art keywords
circuit
pulse
input
input signal
integrated
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210406389.4A
Other languages
Chinese (zh)
Inventor
王绍迪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Witinmem Technology Co ltd
Original Assignee
Beijing Witinmem Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Witinmem Technology Co ltd filed Critical Beijing Witinmem Technology Co ltd
Priority to CN202210406389.4A priority Critical patent/CN114741021A/en
Publication of CN114741021A publication Critical patent/CN114741021A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/05Digital input using the sampling of an analogue quantity at regular intervals of time, input from a/d converter or output to d/a converter
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06JHYBRID COMPUTING ARRANGEMENTS
    • G06J1/00Hybrid computing arrangements
    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03MCODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M1/00Analogue/digital conversion; Digital/analogue conversion
    • H03M1/66Digital/analogue converters
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Mathematical Physics (AREA)
  • Automation & Control Theory (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Software Systems (AREA)
  • Analogue/Digital Conversion (AREA)

Abstract

本发明提供一种存算一体芯片,包括:输入信号转换电路模块,用于对输入信号的低N‑bit信号进行幅度调制,并对输入信号的高M‑bit信号进行脉冲调制;存算一体单元阵列,连接所述输入信号转换电路模块,用于对调制后的输入信号进行运算;输出信号转换电路模块,连接所述存算一体单元阵列,用于将运算结果转换为数字输出信号;其中,通过将输入信号拆成2部分,低N‑bit信号用电压幅度表示,高M‑bit信号用脉冲个数或脉冲宽度表示,能够有效减少随着比特数增加导致的DAC复杂度大幅增长的问题,有利于实现芯片的小型化和低功耗。

Figure 202210406389

The invention provides an integrated storage and computing chip, comprising: an input signal conversion circuit module for performing amplitude modulation on the low N-bit signal of the input signal and pulse modulation on the high M-bit signal of the input signal; a cell array, connected to the input signal conversion circuit module, for performing operations on the modulated input signal; an output signal conversion circuit module, connected to the storage-calculation integrated cell array, for converting the operation result into a digital output signal; wherein , by splitting the input signal into two parts, the low N-bit signal is represented by the voltage amplitude, and the high M-bit signal is represented by the number of pulses or pulse width, which can effectively reduce the DAC complexity caused by the increase of the number of bits. It is beneficial to realize the miniaturization and low power consumption of the chip.

Figure 202210406389

Description

存算一体芯片Storage and computing integrated chip

技术领域technical field

本发明涉及半导体集成电路领域,尤其涉及一种存算一体芯片。The invention relates to the field of semiconductor integrated circuits, in particular to an integrated chip for storage and calculation.

背景技术Background technique

为了解决传统冯诺依曼计算体系结构瓶颈,存算一体芯片架构得到广泛关注。存算一体芯片利用浮栅晶体管器件做矩阵乘加运算,通常是基于电平来操作的,即输入信号通过电压或电流幅度来表示,但是,随着比特数增加,电压幅度会超出范围,电压或电流幅度信号进入浮栅晶体管阵列需要DAC,比如,9bit信号需要对应DAC的数量为512个,8bit信号需要对应DAC的数量为256个,随着输入信号bit数增加,需要的DAC数量指数级增长,DAC的面积大,功耗高,不能适应集成化、低成本化的需求,若将输入信号全部通过脉冲表示,运算时延大大延长,不能满足运算速度的需要。In order to solve the bottleneck of the traditional von Neumann computing architecture, the integrated memory-computing chip architecture has received extensive attention. The integrated memory and computing chip uses floating gate transistor devices to perform matrix multiplication and addition operations, usually based on level operation, that is, the input signal is represented by voltage or current amplitude. However, as the number of bits increases, the voltage amplitude will exceed the range, and the voltage Or the current amplitude signal needs a DAC to enter the floating gate transistor array. For example, the number of DACs corresponding to 9bit signals is 512, and the number of DACs corresponding to 8bit signals is 256. As the number of bits of the input signal increases, the number of DACs required is exponential. With the increase of DAC, the area of DAC is large and the power consumption is high, which cannot meet the needs of integration and low cost. If all input signals are represented by pulses, the operation delay will be greatly prolonged, which cannot meet the needs of operation speed.

发明内容SUMMARY OF THE INVENTION

针对现有技术中的问题,本发明提供一种存算一体芯片,能够至少部分地解决现有技术中存在的问题。In view of the problems in the prior art, the present invention provides an integrated chip for storage and computing, which can at least partially solve the problems in the prior art.

为了实现上述目的,本发明采用如下技术方案:In order to achieve the above object, the present invention adopts the following technical solutions:

第一方面,提供一种存算一体芯片,包括:In a first aspect, a memory-computing integrated chip is provided, including:

输入信号转换电路模块,用于对输入信号的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制;The input signal conversion circuit module is used to perform amplitude modulation on the low N-bit signal of the input signal and pulse modulation on the high M-bit signal of the input signal;

存算一体单元阵列,连接所述输入信号转换电路模块,用于对调制后的输入信号进行运算;an integrated unit array of storage and calculation, connected to the input signal conversion circuit module, for performing operations on the modulated input signal;

输出信号转换电路模块,连接所述存算一体单元阵列,用于将运算结果转换为数字输出信号。The output signal conversion circuit module is connected to the integrated storage and calculation unit array, and is used for converting the operation result into a digital output signal.

进一步地,所述输入信号转换电路模块包括:脉冲调制电路、DAC电路以及第一MUX电路;Further, the input signal conversion circuit module includes: a pulse modulation circuit, a DAC circuit and a first MUX circuit;

所述脉冲调制电路以及所述DAC电路的输出端分别连接所述第一MUX电路的两个输入端;所述第一MUX电路的输出端作为所述输入信号转换电路模块的输出端,连接所述存算一体单元阵列的输入端。The output terminals of the pulse modulation circuit and the DAC circuit are respectively connected to the two input terminals of the first MUX circuit; the output terminal of the first MUX circuit is used as the output terminal of the input signal conversion circuit module, and is connected to the output terminal of the first MUX circuit. The input terminal of the integrated unit array of storage and calculation.

进一步地,所述脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。Further, the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit.

进一步地,所述脉冲调制电路为数字时间转换电路或预脉冲截取电路或脉冲计数器。Further, the pulse modulation circuit is a digital time conversion circuit or a pre-pulse interception circuit or a pulse counter.

进一步地,所述输出信号转换电路模块包括:电荷积分放大器电路、ADC电路以及第二MUX电路;Further, the output signal conversion circuit module includes: a charge integration amplifier circuit, an ADC circuit and a second MUX circuit;

所述第二MUX电路的输入端连接所述存算一体单元阵列的输出端;所述第二MUX电路的输出端分别连接所述电荷积分放大器电路以及所述ADC电路的输入端。The input end of the second MUX circuit is connected to the output end of the integrated storage and calculation unit array; the output end of the second MUX circuit is respectively connected to the input end of the charge integration amplifier circuit and the ADC circuit.

第二方面,提供一种存算一体芯片,包括:低N-bit信号运算模块以及高M-bit信号运算模块;In a second aspect, a memory-computing integrated chip is provided, including: a low N-bit signal operation module and a high M-bit signal operation module;

所述低N-bit信号运算模块用于对输入信号的低N-bit信号进行幅度调制后进行存内运算;所述高M-bit信号运算模块用于对输入信号的高M-bit信号进行脉冲调制后进行存内运算。The low-N-bit signal operation module is used to perform in-memory operation after amplitude modulation of the low-N-bit signal of the input signal; the high-M-bit signal operation module is used to perform an in-memory operation on the high-M-bit signal of the input signal. In-memory operation is performed after pulse modulation.

进一步地,所述高M-bit信号运算模块包括:脉冲调制电路、存算一体单元阵列以及输出信号转换电路;所述脉冲调制电路的输出端连接所述存算一体单元阵列的输入端;所述存算一体单元阵列的输出端连接所述输出信号转换电路的输入端。Further, the high M-bit signal operation module includes: a pulse modulation circuit, an integrated storage and calculation unit array and an output signal conversion circuit; the output end of the pulse modulation circuit is connected to the input end of the integrated storage and calculation unit array; The output end of the integrated storage and calculation unit array is connected to the input end of the output signal conversion circuit.

进一步地,所述脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。Further, the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit.

进一步地,所述脉冲调制电路为数字时间转换电路或预脉冲截取电路或脉冲计数器。Further, the pulse modulation circuit is a digital time conversion circuit or a pre-pulse interception circuit or a pulse counter.

进一步地,所述输出信号转换电路采用电荷积分放大器电路。Further, the output signal conversion circuit adopts a charge integration amplifier circuit.

本发明提供的存算一体芯片,包括:输入信号转换电路模块,用于对输入信号的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制;存算一体单元阵列,连接所述输入信号转换电路模块,用于对调制后的输入信号进行运算;输出信号转换电路模块,连接所述存算一体单元阵列,用于将运算结果转换为数字输出信号;其中,通过将输入信号拆成2部分,低N-bit信号用电压幅度表示,高M-bit信号用脉冲个数或脉冲宽度表示,能够有效减少随着比特数增加导致的DAC复杂度大幅增长的问题,有利于实现芯片的小型化和低功耗,同时,减少运算时延,满足高速运算场合需求。The storage-calculation integrated chip provided by the present invention includes: an input signal conversion circuit module for performing amplitude modulation on the low N-bit signal of the input signal and pulse modulation on the high M-bit signal of the input signal; a storage-calculation integrated unit an array, connected to the input signal conversion circuit module, for performing operations on the modulated input signal; an output signal conversion circuit module, connected to the storage-calculation integrated unit array, for converting the operation result into a digital output signal; wherein, By dividing the input signal into two parts, the low N-bit signal is represented by the voltage amplitude, and the high M-bit signal is represented by the number of pulses or pulse width, which can effectively reduce the problem that the complexity of the DAC increases greatly with the increase of the number of bits. , which is conducive to realizing the miniaturization and low power consumption of the chip, and at the same time, it reduces the operation delay and meets the requirements of high-speed operation occasions.

为让本发明的上述和其他目的、特征和优点能更明显易懂,下文特举较佳实施例,并配合所附图式,作详细说明如下。In order to make the above-mentioned and other objects, features and advantages of the present invention more obvious and easy to understand, the preferred embodiments are exemplified below, and are described in detail as follows in conjunction with the accompanying drawings.

附图说明Description of drawings

为了更清楚地说明本申请实施例或现有技术中的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。在附图中:In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are For some embodiments of the present application, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort. In the attached image:

图1示出了本发明实施例中的存算一体芯片的结构框图;Fig. 1 shows the structural block diagram of the integrated chip of storage and calculation in the embodiment of the present invention;

图2示出了本发明实施例中的输入信号转换电路模块的工作原理;Fig. 2 shows the working principle of the input signal conversion circuit module in the embodiment of the present invention;

图3示出了本发明实施例中的存算一体芯片的具体结构;FIG. 3 shows the specific structure of the integrated storage and computing chip in the embodiment of the present invention;

图4示出了本发明实施例中的输入信号转换电路的具体结构;Fig. 4 shows the specific structure of the input signal conversion circuit in the embodiment of the present invention;

图5示出了本发明实施例中的输入信号转换电路的工作原理;Fig. 5 shows the working principle of the input signal conversion circuit in the embodiment of the present invention;

图6示出了本发明实施例中的另一种输入信号转换电路的工作原理;Fig. 6 shows the working principle of another input signal conversion circuit in the embodiment of the present invention;

图7示出了本发明实施例中的输出信号转换电路的具体结构。FIG. 7 shows the specific structure of the output signal conversion circuit in the embodiment of the present invention.

具体实施方式Detailed ways

为了使本技术领域的人员更好地理解本申请方案,下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本申请一部分的实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都应当属于本申请保护的范围。In order to make those skilled in the art better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only The embodiments are part of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the scope of protection of the present application.

以下在实施方式中详细叙述本发明的详细特征以及优点,其内容足以使任何本领域技术人员,了解本发明的技术内容并据以实施,且根据本说明书所揭露的内容、权利要求及图式,任何本领域技术人员可轻易地理解本发明相关的目的及优点。以下的实施例进一步详细说明本发明的观点,但非以任何观点限制本发明的范畴。The detailed features and advantages of the present invention are described in detail in the following embodiments, and the content is sufficient to enable any person skilled in the art to understand the technical content of the present invention and implement it accordingly, and according to the contents disclosed in this specification, claims and drawings , any person skilled in the art can easily understand the related objects and advantages of the present invention. The following examples further illustrate the concept of the present invention in further detail, but are not intended to limit the scope of the present invention in any way.

需要说明的是,在不冲突的情况下,本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

现有存算一体芯片中,存算一体单元阵列(或称flash阵列、闪存单元阵列或浮栅晶体管阵列等)用于存储权重,输入数字信号通过DAC转换成电压/电流幅度,传输到存算一体单元阵列,输出也是检测电压/电流幅度,通过ADC进行输出,导致存算一体单元阵列的前端设置DAC,后端设置ADC,电路面积大,功耗高、成本高。In the existing memory-computing integrated chip, the memory-computing integrated cell array (or called flash array, flash memory cell array or floating gate transistor array, etc.) is used to store weights, and the input digital signal is converted into voltage/current amplitude through DAC, and transmitted to the memory-computing integrated chip. The output of the integrated cell array is also to detect the voltage/current amplitude, which is output through the ADC, resulting in a DAC at the front end of the integrated cell array, and an ADC at the back end, which has a large circuit area, high power consumption and high cost.

本发明实施例提供的存算一体芯片,通过将输入信号拆成2部分,低N-bit信号用电压幅度表示,高M-bit信号用脉冲个数(个数越多,表示输入信号数值越大)或脉冲宽度(宽度越宽,表示输入信号数值越大)表示,能够有效减少随着比特数数增加导致的DAC复杂度大幅增长的问题,有利于实现芯片的小型化和低功耗,同时,减少运算时延,满足高速运算场合需求。In the integrated storage and computing chip provided by the embodiment of the present invention, the input signal is divided into two parts, the low N-bit signal is represented by the voltage amplitude, and the high M-bit signal is represented by the number of pulses (the more the number, the higher the value of the input signal). Larger) or pulse width (the wider the width, the larger the input signal value), which can effectively reduce the problem of the large increase in the DAC complexity caused by the increase of the number of bits, which is conducive to the miniaturization and low power consumption of the chip. At the same time, the operation delay is reduced to meet the needs of high-speed operation occasions.

图1示出了本发明实施例中的存算一体芯片的结构框图;图2示出了本发明实施例中的输入信号转换电路模块的工作原理;结合图1和图2,该脉冲存算一体芯片包括:输入信号转换电路模块、存算一体单元阵列以及输出信号转换电路模块。Fig. 1 shows a structural block diagram of an integrated chip for storing and calculating in an embodiment of the present invention; Fig. 2 shows the working principle of an input signal conversion circuit module in an embodiment of the present invention; The integrated chip includes: an input signal conversion circuit module, an integrated storage and calculation unit array, and an output signal conversion circuit module.

输入信号转换电路模块用于对输入信号的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制;The input signal conversion circuit module is used to perform amplitude modulation on the low N-bit signal of the input signal and pulse modulation on the high M-bit signal of the input signal;

其中,结合图2,存算一体单元阵列具有多行,每行具有一个输入端,每个输入端用于输入待运算数字信号,例如,第一行输入数字信号X1,对应的输入信号转换电路将该数字信号中的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制,调制后的信号如图所示,而后输入第一行的输入端进行后续运算;Wherein, with reference to FIG. 2 , the integrated storage and computing unit array has multiple rows, each row has an input terminal, and each input terminal is used to input the digital signal to be calculated, for example, the first row inputs the digital signal X 1 , and the corresponding input signal conversion The circuit performs amplitude modulation on the low N-bit signal in the digital signal, and pulse modulation on the high M-bit signal of the input signal. The modulated signal is as shown in the figure, and then input to the input terminal of the first row for subsequent operations ;

举例来说,假设输入数字信号为9比特信号,可以将低8比特用原方案,即利用DAC以电压幅度表示,高1比特用脉冲个数,如果是高1bit的数值是1,则是用1个固定幅度的脉冲。For example, assuming that the input digital signal is a 9-bit signal, the original scheme can be used for the lower 8 bits, that is, the DAC is used to represent the voltage amplitude, and the upper 1 bit is the number of pulses. If the value of the upper 1 bit is 1, it is used 1 fixed-amplitude pulse.

存算一体单元阵列连接所述输入信号转换电路模块,用于对调制后的输入信号进行运算;The integrated unit array of storage and calculation is connected to the input signal conversion circuit module, and is used to perform operation on the modulated input signal;

输出信号转换电路模块连接所述存算一体单元阵列,用于将运算结果转换为数字输出信号。The output signal conversion circuit module is connected to the integrated storage and calculation unit array, and is used for converting the operation result into a digital output signal.

值得说明的是,存算一体单元阵列由多个存储单元排布组成,该存储单元可以为可编程半导体器件。It is worth noting that the memory-computing integrated cell array is composed of a plurality of memory cells arranged, and the memory cells may be programmable semiconductor devices.

具体地,所述可编程半导体器件可为浮栅晶体管、阻变存储器件、相变存储器件或自旋存储器件等。Specifically, the programmable semiconductor device may be a floating gate transistor, a resistive memory device, a phase change memory device, a spin memory device, or the like.

举例来说,该存算一体单元阵列可以是NOR型闪存处理阵列或NAND型闪存处理阵列等。For example, the memory-computing integrated cell array may be a NOR-type flash memory processing array or a NAND-type flash memory processing array.

在一个可选的实施例中,存算一体单元阵列由多个浮栅晶体管阵列排布组成。In an optional embodiment, the memory-computing integrated cell array is composed of a plurality of floating gate transistor arrays.

本领域技术人员可以理解的是,实际应用中,可以使用多个闪存处理阵列中的部分闪存处理阵列进行运算处理,也可以利用全部闪存处理阵列进行运算处理,闪存处理阵列的投入数量依运算需求而定。同理,对于每个闪存处理阵列来说,可以使用其部分器件进行运算处理,也可以利用其全部器件进行运算处理,其器件投入数量依运算需求而定。Those skilled in the art can understand that, in practical applications, part of the flash memory processing arrays in the multiple flash memory processing arrays may be used for computing processing, and all flash processing arrays may be used for computing processing. The input quantity of the flash memory processing arrays depends on the computing requirements Depends. Similarly, for each flash memory processing array, some of its devices can be used for computing processing, and all of its devices can be used for computing processing, and the number of devices invested depends on computing requirements.

在一个可选的实施例中,参见图3,该输入信号转换电路模块包括多个输入信号转换电路,每个输入信号转换电路对应存算一体单元阵列的一行,用于将该行的数字输入信号的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制,调制后的信号输入该行存算一体单元;另外,输出信号转换电路模块包括多个输出信号转换电路,每个输出信号转换电路对应存算一体单元阵列的一列,用于将该列输出的模拟运算结果转换为数字输出信号。In an optional embodiment, referring to FIG. 3 , the input signal conversion circuit module includes a plurality of input signal conversion circuits, and each input signal conversion circuit corresponds to a row of the integrated unit array for storing and calculating the digital input of the row. The low N-bit signal of the signal is amplitude modulated, and the high M-bit signal of the input signal is pulse modulated, and the modulated signal is input into the line storage and calculation integrated unit; in addition, the output signal conversion circuit module includes a plurality of output signal conversion Each output signal conversion circuit corresponds to a column of the integrated unit array, and is used to convert the analog operation result output by the column into a digital output signal.

在一个可选的实施例中,参见图4,该输入信号转换电路模块包括:MUX1脉冲调制电路、DAC电路以及MUX2;In an optional embodiment, referring to FIG. 4 , the input signal conversion circuit module includes: a MUX1 pulse modulation circuit, a DAC circuit, and a MUX2;

MUX1的输入端用于接入该输入信号转换电路的数字输入信号,MUX1的两个输出端作为选择端,分别连接DAC和脉冲信号调制电路;The input terminal of MUX1 is used to access the digital input signal of the input signal conversion circuit, and the two output terminals of MUX1 are used as selection terminals, which are respectively connected to the DAC and the pulse signal modulation circuit;

脉冲调制电路以及DAC电路的输出端分别连接MUX2的两个输入端;MUX2的输出端作为输入信号转换电路的输出端,连接存算一体单元阵列对应行的输入端。The output terminals of the pulse modulation circuit and the DAC circuit are respectively connected to the two input terminals of the MUX2; the output terminal of the MUX2 is used as the output terminal of the input signal conversion circuit, and is connected to the input terminal of the corresponding row of the integrated storage and calculation unit array.

通过时序控制MUX1和MUX2选择性导通DAC或脉冲信号调制电路,以便在输入低N-bit信号时DAC导通进行幅度调制,输入高M-bit信号时脉冲信号调制电路导通进行脉冲调制。The DAC or pulse signal modulation circuit is selectively turned on by timing control MUX1 and MUX2, so that the DAC is turned on for amplitude modulation when a low N-bit signal is input, and the pulse signal modulation circuit is turned on for pulse modulation when a high M-bit signal is input.

在一个可选的实施例中,脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。In an optional embodiment, the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit.

在另外一个可选的实施例中,脉冲调制电路为DTC电路(数字时间转换电路,Digital to Time Converter)或预脉冲截取电路或脉冲计数器。In another optional embodiment, the pulse modulation circuit is a DTC circuit (Digital to Time Converter, Digital to Time Converter) or a pre-pulse interception circuit or a pulse counter.

值得说明的是,参见图5,通过MUX1进行选通,低N-bit信号通过DAC后,以电压/电流幅度表示输入大小,高M-bit信号时脉冲信号通过脉冲生成器后,以输入脉冲个数代表输入大小。It is worth noting that, referring to Figure 5, the MUX1 is used for gating. After the low N-bit signal passes through the DAC, the input size is represented by the voltage/current amplitude. When the high M-bit signal passes through the pulse generator, the input pulse signal is used as the input pulse. The number represents the input size.

在一个可选的实施例中,该脉冲生成器可为预脉冲截取电路或脉冲计数器。In an alternative embodiment, the pulse generator may be a pre-pulse truncation circuit or a pulse counter.

在另一个可选的实施例中,脉冲生成器还可以为延迟锁相环DLL或数字-时间转换器(digital to time convertor)。以数字输入信号为101为例,通过延迟锁相环,得到的脉冲输入信号为脉冲有效时间宽度占时钟周期的5/8的脉冲信号。In another optional embodiment, the pulse generator can also be a delay locked loop DLL or a digital to time converter. Taking the digital input signal as 101 as an example, through the delay-locked loop, the obtained pulse input signal is a pulse signal whose effective time width of the pulse accounts for 5/8 of the clock period.

图6示出了本发明实施例中的另一种输入信号转换电路的工作原理;如图6所示,通过MUX1进行选通,低N-bit信号通过DAC后,以电压/电流幅度表示输入大小,高M-bit信号时脉冲信号通过脉冲生成器后,以输入脉冲宽度代表输入大小。FIG. 6 shows the working principle of another input signal conversion circuit in the embodiment of the present invention; as shown in FIG. 6, the MUX1 is used for gating, and after the low N-bit signal passes through the DAC, the input signal is represented by the voltage/current amplitude. size, when the high M-bit signal is used, after the pulse signal passes through the pulse generator, the input pulse width is used to represent the input size.

在一个可选的实施例中,参见图7,该输出信号转换电路包括:MUX3、电荷积分放大器电路、ADC以及MUX4;In an optional embodiment, referring to FIG. 7 , the output signal conversion circuit includes: MUX3, a charge integrating amplifier circuit, an ADC, and a MUX4;

MUX3的输入端连接存算一体单元阵列中对应列的输出端,用于接收该列的运算结果,MUX3的两个输出端作为选择端,分别连接电荷积分放大器的输入端以及ADC的输入端,电荷积分放大器的输出端以及ADC的输出端分别连接MUX4的两个输入端,MUX4的输出端作为该输出信号转换电路的输出端,用于输出将该列的运算结果转换为数字信号的输出信号。The input terminal of MUX3 is connected to the output terminal of the corresponding column in the integrated storage and calculation unit array, which is used to receive the operation result of the column. The two output terminals of MUX3 are used as selection terminals, which are respectively connected to the input terminal of the charge integration amplifier and the input terminal of the ADC. The output end of the charge integration amplifier and the output end of the ADC are respectively connected to the two input ends of the MUX4, and the output end of the MUX4 is used as the output end of the output signal conversion circuit to output the output signal that converts the operation result of the column into a digital signal .

本发明实施例还提供了一种存算一体芯片,包括:低N-bit信号运算模块以及高M-bit信号运算模块;The embodiment of the present invention also provides an integrated storage and computing chip, including: a low N-bit signal operation module and a high M-bit signal operation module;

低N-bit信号运算模块用于对输入信号的低N-bit信号进行幅度调制后进行存内运算;高M-bit信号运算模块用于对输入信号的高M-bit信号进行脉冲调制后进行存内运算。The low-N-bit signal operation module is used to perform in-memory operations after amplitude modulation of the low-N-bit signal of the input signal; the high-M-bit signal operation module is used to pulse-modulate the high-M-bit signal of the input signal. in-memory operations.

即预先将存算一体芯片的存储单元阵列划分为多个子阵列,用于实现不同的运算,低N-bit信号运算模块包括其中的一个或多个子阵列,该一个或多个子阵列用于对低N-bit信号部分进行运算;高M-bit信号运算模块对应其中的一个或多个子阵列,用于对高M-bit信号信号部分进行运算;低N-bit信号运算模块以及高M-bit信号运算模块的运算结果在存算一体芯片的数字域进行处理。That is, the memory cell array of the integrated memory and computing chip is divided into multiple sub-arrays in advance to realize different operations. The low-N-bit signal operation module includes one or more sub-arrays. The one or more sub-arrays are used to The N-bit signal part is operated; the high M-bit signal operation module corresponds to one or more sub-arrays, and is used to operate the high M-bit signal signal part; the low N-bit signal operation module and the high M-bit signal The operation result of the operation module is processed in the digital domain of the integrated storage and calculation chip.

其中,低N-bit信号运算模块中的一个或多个子阵列与高M-bit信号运算模块包括子阵列可以采用不同的子阵列实现,在对芯片面积和功耗要求更好的场合,低N-bit信号运算模块中的一个或多个子阵列与高M-bit信号运算模块包括子阵列可以采用相同的子阵列实现,采用分时复用的方式,控制该子阵列模块执行低N-bit信号运算或高M-bit信号运算。Among them, one or more sub-arrays in the low-N-bit signal operation module and the high-M-bit signal operation module including sub-arrays can be implemented by different sub-arrays. In the case of better requirements on chip area and power consumption, low-N - One or more sub-arrays in the bit signal operation module and the high-M-bit signal operation module including sub-arrays can be implemented by the same sub-array, and the sub-array module is controlled to execute the low-N-bit signal in a time-division multiplexing manner operation or high M-bit signal operation.

值得说明的是,存算一体芯片通常包括数字域和模拟域,模拟域主要是进行模拟运算的存算一体单元阵列以及配套的外围模拟电路,数字域主要包括处理器、控制器等进行数字信号处理的电路模块。It is worth noting that a memory-computing integrated chip usually includes a digital domain and an analog domain. The analog domain is mainly an integrated memory-computing unit array that performs analog operations and supporting peripheral analog circuits. The digital domain mainly includes processors, controllers, etc. for digital signals. Processed circuit modules.

在一个可选的实施例中,高M-bit信号运算模块包括:脉冲调制电路、存算一体单元子阵列以及输出信号转换电路;脉冲调制电路的输出端连接存算一体单元子阵列的输入端;存算一体单元阵列的输出端连接输出信号转换电路的输入端。In an optional embodiment, the high M-bit signal operation module includes: a pulse modulation circuit, an integrated storage-calculation unit sub-array, and an output signal conversion circuit; an output end of the pulse modulation circuit is connected to an input end of the storage-calculation integrated unit sub-array ; The output end of the integrated storage and calculation unit array is connected to the input end of the output signal conversion circuit.

其中,脉冲调制电路的输入端用于接收待运算的数字信号的高M-bit信号,脉冲调制电路用于将高M-bit信号调制为不同脉冲宽度或不同脉冲个数的脉冲信号;在一个可选的实施例中,脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。Among them, the input terminal of the pulse modulation circuit is used to receive the high M-bit signal of the digital signal to be calculated, and the pulse modulation circuit is used to modulate the high M-bit signal into pulse signals of different pulse widths or different numbers of pulses; In an optional embodiment, the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit.

在另一个可选的实施例中,脉冲调制电路为DTC电路或预脉冲截取电路或脉冲计数器。In another optional embodiment, the pulse modulation circuit is a DTC circuit or a pre-pulse truncation circuit or a pulse counter.

在一个可选的实施例中,输出信号转换电路采用电荷积分放大器电路。In an optional embodiment, the output signal conversion circuit adopts a charge integration amplifier circuit.

在另一个可选的实施例中,输出信号转换电路采用脉冲计数器。In another optional embodiment, the output signal conversion circuit adopts a pulse counter.

值得说明的是,在上述各实施例中,低N-bit信号运算模块包括DAC、存算一体单元子阵列、ADC;其中,DAC电路的输入端用于接收待运算的数字信号的低N-bit信号,DAC的输出端连接该存算一体单元子阵列的输入端,用于将低N-bit信号进行幅度调制后的模拟信号传输至该存算一体单元子阵列进行模拟运算;该存算一体单元子阵列的输出端连接ADC的输入端,ADC用于将模拟运算结果转换成数字信号。It should be noted that, in the above embodiments, the low-N-bit signal operation module includes a DAC, an integrated storage-calculation unit sub-array, and an ADC; wherein, the input end of the DAC circuit is used to receive the low-N-bit signal of the digital signal to be calculated. bit signal, the output terminal of the DAC is connected to the input terminal of the integrated storage and calculation unit sub-array, and is used to transmit the analog signal after amplitude modulation of the low N-bit signal to the integrated storage and calculation unit sub-array for analog operation; The output end of the integrated unit sub-array is connected to the input end of the ADC, and the ADC is used to convert the analog operation result into a digital signal.

本发明实施例还提供了一种电子设备,包括上述的存算一体芯片。An embodiment of the present invention also provides an electronic device, including the above-mentioned integrated storage and computing chip.

举例来说,该电子设备可以是:手机、计算机、平板电脑、可穿戴设备等,本发明实施例对此不作限制。For example, the electronic device may be a mobile phone, a computer, a tablet computer, a wearable device, etc., which is not limited in this embodiment of the present invention.

电子设备通过采用存算一体芯片,能够利于小型化和低功耗化。Electronic equipment can be miniaturized and low power consumption by using integrated memory and computing chips.

本发明中应用了具体实施例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处,综上所述,本说明书内容不应理解为对本发明的限制。In the present invention, the principles and implementations of the present invention are described by using specific embodiments, and the descriptions of the above embodiments are only used to help understand the method and the core idea of the present invention; The idea of the invention will have changes in the specific implementation and application scope. To sum up, the content of this specification should not be construed as a limitation to the present invention.

还需要说明的是,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。It should also be noted that the terms "comprising", "comprising" or any other variation thereof are intended to encompass a non-exclusive inclusion such that a process, method, article or device comprising a series of elements includes not only those elements, but also Other elements not expressly listed, or which are inherent to such a process, method, article of manufacture, or apparatus are also included. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article of manufacture, or device that includes the element.

本说明书中的各个实施例均采用递进的方式描述,各个实施例之间相同相似的部分互相参见即可,每个实施例重点说明的都是与其他实施例的不同之处。尤其,对于系统实施例而言,由于其基本相似于方法实施例,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, as for the system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for related parts, please refer to the partial descriptions of the method embodiments.

以上所述仅是本发明的较佳实施例而已,并非对本发明做任何形式上的限制,虽然本发明已以较佳实施例揭露如上,然而并非用以限定本发明,任何本领域技术人员,在不脱离本发明技术方案的范围内,当可利用上述揭示的技术内容做出些许更动或修饰为等同变化的等效实施例,但凡是未脱离本发明技术方案的内容,依据本发明的技术实质对以上实施例所作的任何简单修改、等同变化与修饰,均仍属于本发明技术方案的范围内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention in any form. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art, Within the scope of not departing from the technical solution of the present invention, when the technical content disclosed above can be used to make some changes or modifications to equivalent embodiments with equivalent changes, but any content that does not depart from the technical solution of the present invention, according to the present invention Any simple modifications, equivalent changes and modifications made to the above embodiments still fall within the scope of the technical solutions of the present invention.

Claims (10)

1.一种存算一体芯片,其特征在于,包括:1. An integrated chip for storing and computing, characterized in that, comprising: 输入信号转换电路模块,用于对输入信号的低N-bit信号进行幅度调制,并对输入信号的高M-bit信号进行脉冲调制;The input signal conversion circuit module is used to perform amplitude modulation on the low N-bit signal of the input signal and pulse modulation on the high M-bit signal of the input signal; 存算一体单元阵列,连接所述输入信号转换电路模块,用于对调制后的输入信号进行运算;an integrated unit array of storage and calculation, connected to the input signal conversion circuit module, for performing operations on the modulated input signal; 输出信号转换电路模块,连接所述存算一体单元阵列,用于将运算结果转换为数字输出信号。The output signal conversion circuit module is connected to the integrated storage and calculation unit array, and is used for converting the operation result into a digital output signal. 2.根据权利要求1所述的存算一体芯片,其特征在于,所述输入信号转换电路模块包括:脉冲调制电路、DAC电路以及第一MUX电路;2. The integrated storage and computing chip according to claim 1, wherein the input signal conversion circuit module comprises: a pulse modulation circuit, a DAC circuit and a first MUX circuit; 所述脉冲调制电路以及所述DAC电路的输出端分别连接所述第一MUX电路的两个输入端;所述第一MUX电路的输出端作为所述输入信号转换电路模块的输出端,连接所述存算一体单元阵列的输入端。The output terminals of the pulse modulation circuit and the DAC circuit are respectively connected to the two input terminals of the first MUX circuit; the output terminal of the first MUX circuit is used as the output terminal of the input signal conversion circuit module, and is connected to the output terminal of the first MUX circuit. The input terminal of the integrated unit array of storage and calculation. 3.根据权利要求2所述的存算一体芯片,其特征在于,所述脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。3 . The integrated storage and computing chip according to claim 2 , wherein the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit. 4 . 4.根据权利要求3所述的存算一体芯片,其特征在于,所述脉冲调制电路为数字时间转换电路或预脉冲截取电路或脉冲计数器。4 . The integrated storage and calculation chip according to claim 3 , wherein the pulse modulation circuit is a digital time conversion circuit, a pre-pulse interception circuit, or a pulse counter. 5 . 5.根据权利要求1所述的存算一体芯片,其特征在于,所述输出信号转换电路模块包括:电荷积分放大器电路、ADC电路以及第二MUX电路;5. The integrated storage and computing chip according to claim 1, wherein the output signal conversion circuit module comprises: a charge integration amplifier circuit, an ADC circuit and a second MUX circuit; 所述第二MUX电路的输入端连接所述存算一体单元阵列的输出端;所述第二MUX电路的输出端分别连接所述电荷积分放大器电路以及所述ADC电路的输入端。The input end of the second MUX circuit is connected to the output end of the integrated storage and calculation unit array; the output end of the second MUX circuit is respectively connected to the input end of the charge integration amplifier circuit and the ADC circuit. 6.一种存算一体芯片,其特征在于,包括:低N-bit信号运算模块以及高M-bit信号运算模块;6. An integrated chip for storage and calculation, characterized in that it comprises: a low N-bit signal operation module and a high M-bit signal operation module; 所述低N-bit信号运算模块用于对输入信号的低N-bit信号进行幅度调制后进行存内运算;所述高M-bit信号运算模块用于对输入信号的高M-bit信号进行脉冲调制后进行存内运算。The low-N-bit signal operation module is used to perform in-memory operation after amplitude modulation of the low-N-bit signal of the input signal; the high-M-bit signal operation module is used to perform an in-memory operation on the high-M-bit signal of the input signal. In-memory operation is performed after pulse modulation. 7.根据权利要求6所述的存算一体芯片,其特征在于,所述高M-bit信号运算模块包括:脉冲调制电路、存算一体单元阵列以及输出信号转换电路;所述脉冲调制电路的输出端连接所述存算一体单元阵列的输入端;所述存算一体单元阵列的输出端连接所述输出信号转换电路的输入端。7. The integrated storage and calculation chip according to claim 6, wherein the high M-bit signal operation module comprises: a pulse modulation circuit, an integrated storage and calculation unit array and an output signal conversion circuit; The output terminal is connected to the input terminal of the integrated storage and calculation unit array; the output terminal of the integrated storage and calculation unit array is connected to the input terminal of the output signal conversion circuit. 8.根据权利要求7所述的存算一体芯片,其特征在于,所述脉冲调制电路为脉冲个数调制电路或脉冲宽度调制电路。8 . The integrated storage and computing chip according to claim 7 , wherein the pulse modulation circuit is a pulse number modulation circuit or a pulse width modulation circuit. 9 . 9.根据权利要求7所述的存算一体芯片,其特征在于,所述脉冲调制电路为数字时间转换电路或预脉冲截取电路或脉冲计数器。9 . The integrated storage and calculation chip according to claim 7 , wherein the pulse modulation circuit is a digital time conversion circuit, a pre-pulse interception circuit, or a pulse counter. 10 . 10.根据权利要求7所述的存算一体芯片,其特征在于,所述输出信号转换电路采用电荷积分放大器电路。10 . The integrated storage and calculation chip according to claim 7 , wherein the output signal conversion circuit adopts a charge integration amplifier circuit. 11 .
CN202210406389.4A 2022-04-18 2022-04-18 Storage and computing integrated chip Pending CN114741021A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210406389.4A CN114741021A (en) 2022-04-18 2022-04-18 Storage and computing integrated chip

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210406389.4A CN114741021A (en) 2022-04-18 2022-04-18 Storage and computing integrated chip

Publications (1)

Publication Number Publication Date
CN114741021A true CN114741021A (en) 2022-07-12

Family

ID=82282527

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210406389.4A Pending CN114741021A (en) 2022-04-18 2022-04-18 Storage and computing integrated chip

Country Status (1)

Country Link
CN (1) CN114741021A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050122238A1 (en) * 2003-10-16 2005-06-09 Canon Kabushiki Kaisha Operation circuit and operation control method thereof
CN112148669A (en) * 2020-10-01 2020-12-29 北京知存科技有限公司 Pulse storage and calculation integrated chip and electronic equipment
CN113364462A (en) * 2021-04-27 2021-09-07 北京航空航天大学 Analog storage and calculation integrated multi-bit precision implementation structure
CN113743600A (en) * 2021-08-26 2021-12-03 南方科技大学 Design method of systolic array with integrated storage and computing architecture suitable for multi-precision neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050122238A1 (en) * 2003-10-16 2005-06-09 Canon Kabushiki Kaisha Operation circuit and operation control method thereof
CN112148669A (en) * 2020-10-01 2020-12-29 北京知存科技有限公司 Pulse storage and calculation integrated chip and electronic equipment
CN113364462A (en) * 2021-04-27 2021-09-07 北京航空航天大学 Analog storage and calculation integrated multi-bit precision implementation structure
CN113743600A (en) * 2021-08-26 2021-12-03 南方科技大学 Design method of systolic array with integrated storage and computing architecture suitable for multi-precision neural network

Similar Documents

Publication Publication Date Title
US11842167B2 (en) Switched capacitor vector-matrix multiplier
US8244788B2 (en) Semiconductor integrated circuit device having a plurality of photo detectors and processing elements
CN114707647B (en) Precision lossless calculation integrated device and method suitable for multi-precision neural network
US20220012016A1 (en) Analog multiply-accumulate unit for multibit in-memory cell computing
CN112151095A (en) Memory and computing integrated chip, memory cell array structure
US10795729B2 (en) Data accelerated processing system
WO2021155851A1 (en) Neural network circuit and neural network system
CN105162437B (en) A kind of waveshape generating device and method
CN113346895B (en) Simulation and storage integrated structure based on pulse cut-off circuit
CN111614353A (en) A digital-to-analog conversion circuit and an analog-to-digital conversion circuit multiplexing device in an integrated storage and computing chip
CN113364462B (en) Analog storage and calculation integrated multi-bit precision implementation structure
CN112148669A (en) Pulse storage and calculation integrated chip and electronic equipment
CN115728566B (en) Signal generating device, measurement and control system and quantum computer
CN118072788A (en) Storage and computing integrated circuits, chips and electronic devices
Jeong et al. A ternary neural network computing-in-memory processor with 16T1C bitcell architecture
WO2025039895A1 (en) Storage computing unit, array, macro module, and upper-level macro module
CN111428857A (en) A device and method for convolution operation based on memristor
CN112162947B (en) Output-configurable memory integrated chip and electronic device
CN115309696A (en) Storage and computation integrated chip architecture, packaging method and device
CN114741021A (en) Storage and computing integrated chip
EP4571746A1 (en) Computing-in-memory chip architecture, packaging method, and apparatus
CN117708039A (en) Storage and computing integrated chip architecture, packaging method and device
CN111611196B (en) Integrated memory chip and DAC multiplexing control method thereof
US12386592B2 (en) Memory array structure with dynamic differential-reference based readout scheme for computing-in-memory applications, dynamic differential-reference time-to-digital converter for computing-in-memory applications and computing method thereof
JP4624431B2 (en) A low-power register array for high-speed shift operations.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
CB02 Change of applicant information

Country or region after: China

Address after: Room 213-175, 2nd Floor, Building 1, No. 180 Kecheng Street, Qiaosi Street, Linping District, Hangzhou City, Zhejiang Province, 311100

Applicant after: Hangzhou Zhicun Computing Technology Co.,Ltd.

Address before: 1707, 17th floor, shining building, No. 35, Xueyuan Road, Haidian District, Beijing 100083

Applicant before: BEIJING WITINMEM TECHNOLOGY Co.,Ltd.

Country or region before: China

CB02 Change of applicant information
RJ01 Rejection of invention patent application after publication

Application publication date: 20220712

RJ01 Rejection of invention patent application after publication