# US 2023/0058641 A1 · Analog-to-digital conversion with bit skipping functionality

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

- Number: US 2023/0058641 A1
- Document ID: US20230058641A1
- Status: Published application
- Kind: A1
- Application number: 17/406,704
- Filed: 2021-08-19
- Published: 2023-02-23
- Earliest claimed: 2021-08-19
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Monodeep Kar; Ankur Agrawal; Mingu Kang; Kyu-hyoun Kim
- Web page: https://patentgrep.com/patent/US20230058641A1

## CPC

- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- H03M 1/38: sequentially only, e.g. successive approximation type (converting more than one bit per step H03M1/14)
- H03M 1/462: {Details of the control circuitry, e.g. of the successive approximation register}
- H03M 1/466: {using switched capacitors}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H03M: CODING; DECODING; CODE CONVERSION IN GENERAL (using fluidic means F15C4/00; optical analogue/digital converters G02F7/00; coding, decoding or code conversion, specially adapted for particular applications, see the relevant subclasses, e.g. G01D, G01R, G06F, G06T, G09G, G10L, G11B, G11C, H04B, H04L, H04M, H04N; ciphering or deciphering for cryptography or other purposes involving the need for secrecy G09C)

## Abstract

Techniques for performing analog-to-digital conversion are disclosed. For example, a method performs an analog-to-digital conversion of an analog input to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits, the analog-to-digital conversion starting at a given one of the one or more additional bits following the most significant bit.

## Claims

### Claim 1

An apparatus comprising:
at least one processor; and
at least one memory including instruction code;
the at least one memory and the instruction code being configured to, with the at least one processor, cause the apparatus at least to:
perform an analog-to-digital conversion of an analog input to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits, the analog-to-digital conversion starting at a given one of the one or more additional bits following the most significant bit.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the at least one memory and the instruction code are further configured to, with the at least one processor, cause the apparatus to:
detect one or more designated conditions; and
responsive to detecting at least one of the one or more designated conditions, restart the analog-to-digital conversion starting at another one of the set of bits.

### Claim 3 (depends on 2)

The apparatus of claim 2, wherein detecting said at least one of the one or more designated conditions is based at least in part on values of a threshold number of consecutive ones of the one or more additional bits following the given one of the one or more additional bits.

### Claim 4 (depends on 3)

The apparatus of claim 3, wherein the threshold number is selected based at least in part on at least one of: an output probability distribution of a dot-product engine performing a binary multiplication of first and second vectors; and a tolerable error in an output of the analog-to-digital conversion.

### Claim 5 (depends on 2)

The apparatus of claim 2, wherein the given one of the one or more additional bits following the most significant bit comprises a first one of the one or more additional bits following the most significant bit, and wherein restarting the analog-to-digital conversion starting at the other one of the set of bits comprises restarting the analog-to-digital conversion starting at the most significant bit.

### Claim 6 (depends on 2)

The apparatus of claim 2, wherein the given one of the one or more additional bits following the most significant bit (MSB) comprises an (MSB-k)th bit, wherein restarting the analog-to-digital conversion starting at the other one of the set of bits comprises restarting the analog-to-digital conversion starting at an (MSB-k+l)th bit, and wherein l is an integer greater than or equal to one.

### Claim 7 (depends on 6)

The apparatus of claim 6, wherein a value of k is selected based at least in part on at least one of: an output probability distribution of a dot-product engine performing a binary multiplication of first and second vectors; and a tolerable error in an output of the analog-to-digital conversion.

### Claim 8

A method, comprising the step of:
performing an analog-to-digital conversion of an analog input to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits, the analog-to-digital conversion starting at a given one of the one or more additional bits following the most significant bit;
wherein the method is executed by processing circuitry configured to execute instruction code.

### Claim 9 (depends on 8)

The method of claim 8, further comprising the steps of:
detecting one or more designated conditions; and
responsive to detecting at least one of the one or more designated conditions, restarting the analog-to-digital conversion starting at another one of the set of bits;
wherein the given one of the one or more additional bits following the most significant bit (MSB) comprises an (MSB-k)th bit, wherein restarting the analog-to-digital conversion starting at the other one of the set of bits comprises restarting the analog-to-digital conversion starting at an (MSB-k+l)th bit, and wherein l is an integer greater than or equal to one.

### Claim 10

An article of manufacture comprising a non-transitory computer-readable storage medium having embodied therein executable instruction code that when executed by a processor causes the processor to perform the step of:
performing an analog-to-digital conversion of an analog input to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits, the analog-to-digital conversion starting at a given one of the one or more additional bits following the most significant bit.

### Claim 11 (depends on 10)

The article of claim 10, wherein the executable instruction code when executed by the processor further causes the processor to perform steps of:
detecting one or more designated conditions; and
responsive to detecting at least one of the one or more designated conditions, restarting the analog-to-digital conversion starting at another one of the set of bits;
wherein the given one of the one or more additional bits following the most significant bit (MSB) comprises an (MSB-k)th bit, wherein restarting the analog-to-digital conversion starting at the other one of the set of bits comprises restarting the analog-to-digital conversion starting at an (MSB-k+l)th bit, and wherein l is an integer greater than or equal to one.

### Claim 12

A system comprising:
a dot-product summation computation unit configured to perform binary multiplication of first and second vectors;
a successive approximation register analog-to-digital converter unit configured to convert an analog output of the dot-product summation computation unit to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits; and
controller logic operatively coupled to the successive approximation register analog-to-digital converter unit, the controller logic being configured to determine values of the set of bits of the digital output based at least in part on a comparison of an output of the dot-product summation computation unit and a sequence of weighted binary codes;
wherein the sequence of weighted binary codes starts from a given one of the one or more additional bits following the most significant bit.

### Claim 13 (depends on 12)

The system of claim 12, wherein the given one of the one or more additional bits following the most significant bit (MSB) comprises an (MSB-k)th bit, wherein the controller logic is further configured, responsive to detecting one or more designated conditions based at least in part on values of a threshold number of consecutive ones of the one or more additional bits following the given one of the one or more additional bits, to restart determination of the values of the set of bits of the digital output with a new sequence of weighted binary codes that starts from an (MSB-k+l)th bit, and wherein l is an integer greater than or equal to one.

### Claim 14 (depends on 12)

The system of claim 12, wherein the system is implemented as part of an artificial intelligence system.

### Claim 15 (depends on 12)

The system of claim 12, wherein the system is implemented as part of one or more integrated circuits.

### Claim 16

A device comprising:
dot-product summation circuitry configured to output a first signal, the first signal comprising a dot-product sum of binary multiplication of first and second vectors; and
a successive approximation register analog-to-digital converter configured to perform conversion of the first signal output by the dot-product summation circuitry to a digital output comprising a set of bits, the set of bits comprising a most significant bit and one or more additional bits;
wherein the successive approximation register analog-to-digital converter comprises:
successive approximation register controller circuitry configured to output a second signal; and
comparator circuitry comprising a first input node coupled to the dot-product summation circuitry, a second input node coupled to the successive approximation register controller circuitry, and an output node configured to provide an output value based at least in part on a comparison of the first value output by the dot-product summation circuitry and the second value output by the successive approximation register controller circuitry;
wherein the successive approximation register controller circuitry is configured to determine values of the set of bits of the digital output based at least in part on output values of the comparator circuitry produced as a sequence of weighted binary codes is output as the second signal; and
wherein the sequence of weighted binary codes starts from a given one of the one or more additional bits following the most significant bit.

### Claim 17 (depends on 16)

The device of claim 16, wherein the given one of the one or more additional bits following the most significant bit (MSB) comprises an (MSB-k)th bit, and wherein the successive approximation register controller circuitry is further configured, responsive to detecting one or more designated conditions based at least in part on values of a threshold number of consecutive ones of the one or more additional bits following the given one of the one or more additional bits, to restart determination of the values of the set of bits of the digital output based at least in part on output values of the comparator circuitry produced as a new sequence of weighted binary codes are output as the second signal, the new sequence of weighted binary codes starting from an (MSB-k+l)th bit, and wherein l is an integer greater than or equal to one.

### Claim 18 (depends on 16)

The device of claim 16, wherein the successive approximation register controller circuitry is coupled to the comparator circuitry via a plurality of capacitors.

### Claim 19 (depends on 18)

The device of claim 18, wherein the plurality of capacitors have respective different capacitance values.

### Claim 20 (depends on 18)

The device of claim 18, wherein the dot-product summation circuitry comprises a plurality of logical AND gates each receiving as input a first vector element of the first vector and a second vector element of the second vector, the plurality of logical AND gates having respective outputs coupled to first plates of the plurality of capacitors.

### Claim 21 (depends on 18)

The device of claim 18, wherein each of the plurality of capacitors has a same capacitance value.

### Claim 22 (depends on 21)

The device of claim 21, wherein a plurality of switches are coupled between respective ones of the plurality of logical AND gates and the plurality of capacitors, the plurality of switches being configured to switch between a first mode of operation wherein a voltage on a common line coupled to second plates of the plurality of capacitors represents the dot-product sum of the first and second vectors and a second mode of operation wherein the voltage on the common line coupled to the second plates of the plurality of capacitors represents one of the sequence of weighted binary codes.

### Claim 23 (depends on 18)

The device of claim 18, wherein the dot-product summation circuitry comprises an additional plurality of capacitors coupled to a plurality of logical AND gates via a plurality of switches, each of the plurality of logical AND gates receiving as input a first vector element of the first vector and a second vector element of the second vector, the plurality of logical AND gates having respective outputs coupled to first plates of the additional plurality of capacitors.

### Claim 24 (depends on 16)

The device of claim 16, wherein the dot-product summation circuitry comprises a plurality of bit cell processing units, each of the plurality of bit cell processing units:
storing a first vector element value representing one of the elements of the first vector;
receiving as input a second vector element value representing one of the elements of the second vector; and
outputting a value representing binary multiplication of the first vector element value and the second vector element value.

### Claim 25 (depends on 24)

The device of claim 24, wherein each of the plurality of bit cell processing units comprises one of a resistive processing unit and a memory cell.
