公开(公告)号：US20230118303A1

公开(公告)日：2023-04-20

申请号：US18082415

申请日：2022-12-15

申请人： Google LLC

发明人： Jeffrey Adgate Dean ,  Sudip Roy ,  Michael Acheson Isard ,  Aakanksha Chowdhery ,  Brennan Saeta ,  Chandramohan Amyangot Thekkath ,  Daniel William Hurt ,  Hyeontaek Lim ,  Laurent El Shafey ,  Parker Edward Schuh ,  Paul Ronald Barham ,  Ruoming Pang ,  Ryan Sepassi ,  Sanjay Ghemawat ,  Yonghui Wu

IPC分类号： G06F9/48 ,  G06N3/063 ,  G06N3/08

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing machine learning workloads, e.g., computations for training a neural network or computing an inference using a neural network, across multiple hardware accelerators. One of the systems comprises a plurality of accelerator islands, each hardware accelerator island comprising a respective plurality of hardware devices that include a plurality of hardware accelerators and a corresponding host for each of the plurality of hardware accelerators; and a respective scheduler for each of the accelerator islands that is configured to schedule workloads across the plurality of accelerators and corresponding hosts in the accelerator island, wherein the system is configured to: receive data representing a machine learning workload; and assign a respective portion of the machine learning workload to each of the plurality of accelerator islands for scheduling by the respective scheduler for the accelerator island.

公开(公告)号：US11657289B2

公开(公告)日：2023-05-23

申请号：US16840191

申请日：2020-04-03

申请人： Google LLC

发明人： Yanqi Zhou ,  Sudip Roy ,  Amirali Abdolrashidi ,  Daniel Lin-Kit Wong ,  Chao Ma ,  Qiumin Xu ,  Azalia Mirhoseini

IPC分类号： G06N3/04 ,  G06K9/62 ,  G06N3/049

CPC分类号： G06N3/0454 ,  G06K9/6231 ,  G06K9/6262 ,  G06K9/6296 ,  G06N3/049

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing the execution of the operations of a neural network. One of the methods includes obtaining data representing a graph characterizing a plurality of operations of a neural network, wherein each node of the graph characterizes an operation of the neural network and each edge of the graph characterizes data dependency between the operations; processing the data representing the graph using a graph embedding neural network to generate an embedding of the graph; and processing the embedding of the graph using a policy neural network to generate a task output, wherein the task output comprises, for each of the plurality of operations of the neural network, a respective decision for a particular optimization task.

公开(公告)号：US12112198B2

公开(公告)日：2024-10-08

申请号：US18082415

申请日：2022-12-15

申请人： Google LLC

发明人： Jeffrey Adgate Dean ,  Sudip Roy ,  Michael Acheson Isard ,  Aakanksha Chowdhery ,  Brennan Saeta ,  Chandramohan Amyangot Thekkath ,  Daniel William Hurt ,  Hyeontaek Lim ,  Laurent El Shafey ,  Parker Edward Schuh ,  Paul Ronald Barham ,  Ruoming Pang ,  Ryan Sepassi ,  Sanjay Ghemawat ,  Yonghui Wu

IPC分类号： G06F17/10 ,  G06F9/48 ,  G06N3/063 ,  G06N3/08

CPC分类号： G06F9/4881 ,  G06N3/063 ,  G06N3/08

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing machine learning workloads, e.g., computations for training a neural network or computing an inference using a neural network, across multiple hardware accelerators. One of the systems comprises a plurality of accelerator islands, each hardware accelerator island comprising a respective plurality of hardware devices that include a plurality of hardware accelerators and a corresponding host for each of the plurality of hardware accelerators; and a respective scheduler for each of the accelerator islands that is configured to schedule workloads across the plurality of accelerators and corresponding hosts in the accelerator island, wherein the system is configured to: receive data representing a machine learning workload; and assign a respective portion of the machine learning workload to each of the plurality of accelerator islands for scheduling by the respective scheduler for the accelerator island.

公开(公告)号：US20220357985A1

公开(公告)日：2022-11-10

申请号：US17738909

申请日：2022-05-06

申请人： Google LLC

发明人： Jeffrey Adgate Dean ,  Sudip Roy ,  Michael Acheson Isard ,  Aakanksha Chowdhery ,  Brennan Saeta ,  Chandramohan Amyangot Thekkath ,  Daniel William Hurt ,  Hyeontaek Lim ,  Laurent El Shafey ,  Parker Edward Schuh ,  Paul Ronald Barham ,  Ruoming Pang ,  Ryan Sepassi ,  Sanjay Ghemawat ,  Yonghui Wu

IPC分类号： G06F9/48 ,  G06N3/063 ,  G06N3/08

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing machine learning workloads, e.g., computations for training a neural network or computing an inference using a neural network, across multiple hardware accelerators. One of the systems comprises a plurality of accelerator islands, each hardware accelerator island comprising a respective plurality of hardware devices that include a plurality of hardware accelerators and a corresponding host for each of the plurality of hardware accelerators; and a respective scheduler for each of the accelerator islands that is configured to schedule workloads across the plurality of accelerators and corresponding hosts in the accelerator island, wherein the system is configured to: receive data representing a machine learning workload; and assign a respective portion of the machine learning workload to each of the plurality of accelerator islands for scheduling by the respective scheduler for the accelerator island.

公开(公告)号：US20230176840A1

公开(公告)日：2023-06-08

申请号：US17921933

申请日：2021-06-07

申请人： Google LLC

发明人： Yanqi Zhou ,  Sudip Roy ,  Amirali Abdolrashidi ,  Daniel Lin-Kit Wong ,  Chao Ma ,  Qiumin Xu ,  Hanxiao Liu ,  Phitchaya Mangpo Phothilimthana ,  Shen Wang ,  Anna Darling Goldie ,  Azalia Mirhoseini ,  James Laudon

IPC分类号： G06F8/41

CPC分类号： G06F8/443

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for compiler optimizations using a compiler optimization network. One of the methods includes receiving an input program, wherein the input program defines a graph of operation modules, wherein each node in the graph is a respective operation module, and each edge between nodes in the graph represents one operation module receiving the output generated by another operation module. The input program is processed by a compiler optimization network comprising a graph-embedding network that is configured to encode operation features and operation dependencies of the operation modules of the input program into a graph embedding representation and a policy network that is configured to generate an optimization action for each of one or more nodes encoded in the graph embedding representation. The compiler optimization network generates an output optimization plan comprising one or more optimization actions for the input program.

公开(公告)号：US20230306266A1

公开(公告)日：2023-09-28

申请号：US18321691

申请日：2023-05-22

申请人： Google LLC

发明人： Yanqi Zhou ,  Sudip Roy ,  Amirali Abdolrashidi ,  Daniel Lin-Kit Wong ,  Chao Ma ,  Qiumin Xu ,  Azalia Mirhoseini

IPC分类号： G06N3/084 ,  G06N3/049 ,  G06F18/20 ,  G06F18/21 ,  G06F18/2115 ,  G06N3/045

CPC分类号： G06N3/084 ,  G06N3/049 ,  G06F18/29 ,  G06F18/217 ,  G06F18/2115 ,  G06N3/045

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing the execution of the operations of a neural network. One of the methods includes obtaining data representing a graph characterizing a plurality of operations of a neural network, wherein each node of the graph characterizes an operation of the neural network and each edge of the graph characterizes data dependency between the operations; processing the data representing the graph using a graph embedding neural network to generate an embedding of the graph; and processing the embedding of the graph using a policy neural network to generate a task output, wherein the task output comprises, for each of the plurality of operations of the neural network, a respective decision for a particular optimization task.

公开(公告)号：US11556381B2

公开(公告)日：2023-01-17

申请号：US17738909

申请日：2022-05-06

申请人： Google LLC

发明人： Jeffrey Adgate Dean ,  Sudip Roy ,  Michael Acheson Isard ,  Aakanksha Chowdhery ,  Brennan Saeta ,  Chandramohan Amyangot Thekkath ,  Daniel William Hurt ,  Hyeontaek Lim ,  Laurent El Shafey ,  Parker Edward Schuh ,  Paul Ronald Barham ,  Ruoming Pang ,  Ryan Sepassi ,  Sanjay Ghemawat ,  Yonghui Wu

IPC分类号： G06F15/82 ,  G06F9/48 ,  G06N3/08 ,  G06N3/063

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing machine learning workloads, e.g., computations for training a neural network or computing an inference using a neural network, across multiple hardware accelerators. One of the systems comprises a plurality of accelerator islands, each hardware accelerator island comprising a respective plurality of hardware devices that include a plurality of hardware accelerators and a corresponding host for each of the plurality of hardware accelerators; and a respective scheduler for each of the accelerator islands that is configured to schedule workloads across the plurality of accelerators and corresponding hosts in the accelerator island, wherein the system is configured to: receive data representing a machine learning workload; and assign a respective portion of the machine learning workload to each of the plurality of accelerator islands for scheduling by the respective scheduler for the accelerator island.

公开(公告)号：US20210248445A1

公开(公告)日：2021-08-12

申请号：US16840191

申请日：2020-04-03

申请人： Google LLC

发明人： Yanqi Zhou ,  Sudip Roy ,  Amirali Abdolrashidi ,  Daniel Lin-Kit Wong ,  Chao Ma ,  Qiumin Xu ,  Azalia Mirhoseini

IPC分类号： G06N3/04 ,  G06K9/62

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing the execution of the operations of a neural network. One of the methods includes obtaining data representing a graph characterizing a plurality of operations of a neural network, wherein each node of the graph characterizes an operation of the neural network and each edge of the graph characterizes data dependency between the operations; processing the data representing the graph using a graph embedding neural network to generate an embedding of the graph; and processing the embedding of the graph using a policy neural network to generate a task output, wherein the task output comprises, for each of the plurality of operations of the neural network, a respective decision for a particular optimization task.

公开(公告)号：US10417439B2

公开(公告)日：2019-09-17

申请号：US15480971

申请日：2017-04-06

申请人： Google LLC

发明人： Philip Korn ,  Steven Euijong Whang ,  Natalya Fridman Noy ,  Sudip Roy ,  Neoklis Polyzotis ,  Alon Yitzchak Halevy ,  Christopher Olston

IPC分类号： G06F17/30 ,  G06F21/62 ,  G06F16/21 ,  G06F16/215

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating a catalog for multiple datasets, the method comprising accessing multiple extant data sets, the extant data sets including data sets that are independently generated and structurally dissimilar; organizing the data sets into collections, each data set in each collection belonging to the collection based on collection data associated with the data set; for each collection of data sets: determining, from a subset of the data sets that belong to the collection, metadata that describe the data sets that belong to the collection, wherein the metadata does not include the collection data, and attributing, to other data sets in the collection, the metadata determined from the subset of data sets; and generating, from the collections of data sets and the determined metadata, a catalog for the multiple datasets.