# US 2021/0174208 A1 · Search acceleration for artificial intelligence

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

- Number: US 2021/0174208 A1
- Document ID: US20210174208A1
- Status: Published application
- Kind: A1
- Application number: 17/159,982
- Filed: 2021-01-27
- Published: 2021-06-10
- Earliest claimed: 2018-01-31
- Assignees: Pure Storage, Inc.
- Applicants: Pure Storage, Inc.
- Inventors: Fabio Margaglia; Emily Potyraj; Hari Kannan; Cary A. Sandvig
- Web page: https://patentgrep.com/patent/US20210174208A1

## CPC

- G06F 3/061: {Improving I/O performance}
- G06N 3/08: Learning methods
- G06F 3/064: {Management of blocks}
- G06F 3/067: {Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]}
- G06F 16/28: Databases characterised by their database models, e.g. relational or object models
- G06F 16/30: of unstructured textual data (document management systems G06F16/93)
- G06N 5/04: Inference or reasoning models
- G06N 3/063: using electronic means
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

An apparatus for artificial intelligence acceleration is provided. The apparatus includes a storage and compute system having a distributed, redundant key value store for metadata. The storage and compute system having distributed compute resources configurable to access, through a plurality of authorities, data in the solid-state memory, run inference with a deep learning model, generate vectors for the data and store the vectors in the key value store.

## Claims

### Claim 1

An apparatus, comprising:
a storage and compute system having a distributed, redundant key value store for metadata; and
the storage and compute system having distributed compute resources configurable to store metadata for one or more buckets containing data and create views containing links to the data contained in the one or more buckets and stored in memory, the data associated with a similarity predicate based on vectors in the key value store.

### Claim 2 (depends on 1)

The apparatus of claim 1, further comprising:
the storage and compute system having at least one network port, and configurable to receive an object or file, run the inference for the object or the file with the deep learning model, based on the vectors in the key value store, and output a response based on the inference.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein storage capacity of the memory is sufficient to hold all of the data for the inference with the deep learning model throughout the processing of an entire data pipeline.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein the memory is solid state memory.

### Claim 5 (depends on 1)

The apparatus of claim 1, further comprising:
the distributed compute resources configurable to access data in the memory, run inference with a deep learning model, generate vectors for the data and store the vectors in the key value store.

### Claim 6 (depends on 1)

The apparatus of claim 1, further comprising:
the distributed compute resources configurable to compute a vector of an object for a search, perform a hash operation on the vector, search for vectors in the key value store that are within a specified distance of the vector, based on a result of the hash operation, and retrieve one or more full vectors for objects stored in the memory based on a result of the search, wherein the vectors are stored in the key value store using a locality-preserving hash operation.

### Claim 7 (depends on 1)

The apparatus of claim 1, further comprising:
the distributed compute resources configurable to receive tags associated with the data, and generate a list or count of data entities having tags that satisfy a predicate.

### Claim 8

A method, comprising:
generating vectors for data based on performing inference with a deep learning model:
storing the vectors in a key value store;
storing metadata for one or more buckets containing examples associated with unstructured data; and
generating views containing links to the examples that satisfy a similarity predicate, based on a search of the vectors.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
performing the inference for an object with the deep learning model based on vectors in the key value store.

### Claim 10 (depends on 8)

The method of claim 8, further comprising:
outputting a response based on the inference.

### Claim 11 (depends on 8)

The method of claim 8, wherein the system is configurable to write and receive unstructured data into solid state memory of the system.

### Claim 12 (depends on 8)

The method of claim 8, wherein the key value store is a distributed key value store, across storage nodes of the system.

### Claim 13 (depends on 9)

The apparatus of claim 9, further comprising:
search for vectors in the key value store that are within a specified distance of a computed vector associated with a search.

### Claim 14 (depends on 8)

The method of claim 8, wherein the vectors are stored in the key value store using a locality-preserving hash operation.

### Claim 15

A tangible, non-transitory, computer-readable media having instructions thereupon which, when executed by a processor, cause the processor to perform a method comprising:
generating vectors for data based on performing inference with a deep learning model:
storing the vectors in a key value store;
storing metadata for one or more buckets containing examples associated with unstructured data; and
generating views containing links to the examples that satisfy a similarity predicate, based on a search of the vectors.

### Claim 16 (depends on 15)

The computer readable media of claim 15, further comprising:
performing the inference for an object with the deep learning model based on vectors in the key value store.

### Claim 17 (depends on 15)

The computer readable media of claim 15, wherein the method further comprises:
outputting a response based on the inference.

### Claim 18 (depends on 15)

The computer readable media of claim 15, wherein the system is configurable to write and receive unstructured data into solid state memory of the system.

### Claim 19 (depends on 15)

The computer readable media of claim 15, wherein the key value store is a distributed key value store, across storage nodes of the system.

### Claim 20 (depends on 15)

The computer readable media of claim 15, wherein the vectors are stored in the key value store using a locality-preserving hash operation.
