EP-4736055-A1 - DEVICE AND METHOD FOR ACCELERATING PHYSICS-BASED SIMULATIONS USING ARTIFICIAL INTELLIGENCE

Abstract

Method and devices are provided for performing a physics-based simulation. A processing devices comprises memory and a processor. The processor is configured to perform a physics-based simulation by executing a portion of the physics-based simulation, training a neural network model based on results from executing the first portion of the physics-based simulation, performing inference processing based on the results of the training of the neural network model and providing a prediction, based on the inference processing, as an input back to the physics-based simulation.

Inventors

• WHITE, LAURENT S.
• NWANKWO, Darian Osahar
• HORA, Gurpreet Singh

Assignees

• Advanced Micro Devices, Inc.

Dates

Publication Date

20260506

Application Date

20240613

Claims (20)

1. 1 . A processing device comprising: memory; and a processor configured to perform a physics-based simulation by: executing a portion of the physics-based simulation; training a neural network model based on results from executing the portion of the physics-based simulation; performing inference processing based on results of the training of the neural network model; and providing a prediction, based on the inference processing, as an input back to the physics-based simulation.
2. 2. The processing device of claim 1 , wherein the processor is configured to provide the prediction as the input to a next portion of the physics-based simulation.
3. 3. The processing device of claim 1 , wherein the processor is configured to train the neural network model, based on the results from executing the portion of the physics-based simulation, in response to at least one of: an additional portion of the physics-based simulation determined to be executed; and additional training of a neural network determined to be performed.
4. 4. The processing device of claim 3, wherein in response to additional training of the neural network model being determined, the processor is configured to retrain the neural network model using data generated during the portion of the physics-based simulation.
5. 5. The processing device of claim 3, wherein in response to additional training of a neural network model being determined, the processor is configured to train a new neural network model without data generated during the portion of the physics-based simulation.
6. 6. The processing device of claim 1 , wherein the physics-based simulation and the inference processing are performed for a same number of steps.
7. 7. The processing device of claim 6, wherein the same number of steps is performed by the inference processing in less time than the portion of the physicsbased simulation.
8. 8. The processing device of claim 1 , wherein the processor is configured to store part of the results from executing the portion of the physics-based simulation in the memory, the part of the results comprising at least one of: a subsample of the simulated results in space and time; and a portion of state variables.
9. 9. A method of performing a physics-based simulation comprising: executing a portion of the physics-based simulation; training a neural network model based on results from executing the portion of the physics-based simulation; performing inference processing based on results of the training of the neural network model; and providing a prediction, based on the inference processing, as an input back to the physics-based simulation.
10. 10. The method of claim 9, further comprising providing the prediction as the input to a next portion of the physics-based simulation.
11. 11. The method of claim 9, further comprising training the neural network model, based on the results from executing the portion of the physics-based simulation, in response to at least one of: an additional portion of the physics-based simulation determined to be executed; and additional training of a neural network determined to be performed.
12. 12. The method of claim 11 , further comprising, in response to additional training of a neural network model being determined, retraining the neural network model using data generated during the portion of the physics-based simulation.
13. 13. The method of claim 11 , further comprising, in response to additional training of a neural network model being determined, training a new neural network model without data generated during the portion of the physics-based simulation.
14. 14. The method of claim 13, further comprising training the new neural network model in response to an amount of change in simulated physical behavior being greater than a physical behavior change threshold.
15. 15. The method of claim 9, wherein the portion of the physics-based simulation and the inference processing are performed for a same number of steps.
16. 16. The method of claim 15, wherein the same number of steps is performed by the inference processing in less time than the portion of the physicsbased simulation.
17. 17. The method of claim 9, further comprising storing a part of the results from executing the portion of the physics-based simulation, the part of the results comprising at least one of: a subsample of the simulated results in space and time; and a portion of state variables.
18. 18. A method of performing a physics-based simulation comprising: executing a first portion of the physics-based simulation; training a neural network based on results of the first portion of the physics-based simulation to generate a trained neural network model; executing a second portion of the physics-based simulation during a period of time in which the neural network model is trained; performing inference processing based on results of the trained neural network model; and providing a prediction of the inference processing to a third portion of the physics-based simulation when execution of the inference processing completes, wherein one or more predictions, regarding physical processes, are generated from results of the physics-based simulation.
19. 19. The method of claim 18, further comprising providing data, generated by the second portion of the physics-based simulation, to the neural network model while the second portion of the physics-based simulation is executing.
20. 20. The method of claim 18, further comprising using data, generated by the second portion of the physics-based simulation, to validate the trained neural network model after the training of the neural network model is complete.

Description

DEVICE AND METHOD FOR ACCELERATING PHYSICS-BASED SIMULATIONS USING ARTIFICIAL INTELLIGENCE CROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims priority to pending U.S. Non-Provisional Patent Application Number 18/344,544, entitled “DEVICE AND METHOD FOR ACCELERATING PHYSICS-BASED SIMULATIONS USING ARTIFICIAL INTELLIGENCE,” filed on June 29, 2023, the entirety of which is hereby incorporated herein by reference. BACKGROUND [0002] Physics-based simulations have been used for decades to make predictions about physical processes and make decisions based on the predictions. For example, physics-based simulations are used for numerical weather forecasting and computational fluid dynamics for aircraft design. These physics-based simulations have benefited from years of improvement in computing algorithms (e.g., algorithms implemented via software) and hardware, making them a standard tool in many scientific fields, alongside lab and field experiments. BRIEF DESCRIPTION OF THE DRAWINGS [0003] A more detailed understanding can be had from the following description, given by way of example in conjunction with the accompanying drawings wherein: [0004] FIG. 1 is a block diagram of an example computing device in which one or more features of the disclosure can be implemented; [0005] FIG 2 is a block diagram of the device, illustrating additional details related to execution of processing tasks on an accelerated processing device, according to an example; [0006] FIG. 3 is a flow diagram illustrating an example method 300 of accelerating a physics-based simulation according to features of the disclosure; and [0007] FIG. 4 illustrates a comparison between the time used to perform the physics-based simulation without using a neural network and the time used to perform the physics-based simulation by combining the physics-based simulation with a neural network, according to an example. DETAILED DESCRIPTION [0008] While physics-based simulations are very useful for making predictions and decisions about physical processes, physics-based simulations often require very large supercomputers with hundreds to thousands of compute nodes, and millions of compute cores. For example, the power required to operate the largest supercomputers can be larger than 10 MW. [0009] While physics-based simulations can be highly accurate, they are also expensive. For example, the grid-based or mesh-based numerical methods used in physics-based simulations require a large number of floating-point operations to be performed and sometimes results in weeks of computations on the largest computers to reach a final state of the simulation. The high cost of physics-based simulations often prevents their use for making predictions and decisions for various physical processes (e.g., designing a new machine, designing new materials or making societal decisions). [0010] In addition, physics-based simulations result in inefficient use of hardware. For example, a recent slowdown in single-core performance improvement has resulted in physics-based simulations being scaled out to many cores, requiring data communication across many nodes. More time is often spent moving data (e.g., between memory and a processor core, or between nodes via a network) than performing work (e.g., performing computations using the data). Even at a node level, grid-based and mesh-based computations are typically memory-bandwidth-limited (i.e. , a low ratio of computations are performed per unit of data moved from memory to registers). [0011] Artificial intelligence is a technology which causes a machine (e.g., computer) to mimic or simulate human intelligence. Machine learning, a subfield of artificial intelligence, enables a machine to learn and improve from experience (e.g., training) to solve complex problems. [0012] “Machine learning networks” are used herein interchangeably with the terms “neural networks,” “machine learning neural networks,” “deep learning neural networks,” and “machine learning models.” [0013] Machine learning neural networks are widely used in a variety of technologies (e.g., image classification) to make predictions or decisions to perform particular tasks (e.g., whether an image includes a certain object). The neural networks typically include multiple layers. At each layer, a filter is applied to the previous layer, and the results of each layer are known as activations or feature maps. The first and last layers in a neural network are known as the input and output layers, respectively, and the layers in between the first and last layers are known as hidden layers. Machine learning models are trained in order to make predictions or decisions to perform a particular task (e.g., whether an image includes a certain object). During training, a model is exposed to different data. At each hidden layer, the model learns essential feature maps from the data. At the output layer, the model makes a prediction and receives feedback regarding t