An important task in micromagnetics is performing a large ensemble of simulations of a structure with only one or a few changed parameters. An example is computing write error rates in magnetic random-access memories (MRAM), where simulations are needed for different stochastic thermal fields. Such computations, which are done by multiple runs on a single or multiple computing devices. have various numerical overheads related to preprocessing, matrix operations, and excessive data storage. We demonstrate a computational approach for performing a large batch of repetitive micromagnetic simulations on a Graphics Processing Unit (GPU) simultaneously. This batched approach has the following benefits: (i) it utilizes GPUs much more efficiently than the traditional approach of running multiple simulations, including much increased speed and reduced memory consumption, (ii) it hides various numerical overheads, such as preprocessing as well as matrix and time stepping operations, (iii) it makes obtaining results and post-processing much more efficient by allowing computing only required properties, e.g. ensemble averages or switching counts. We extended our FEM based micromagnetic framework FastMag to implement the batched approach in all code components, including pre-processing, matrix operations (Fig. 1a), time stepping with adaptivity to step, order, and decisions of a stopping evens per instance (Fig. 1b). Figure 2 shows major improvements of the computational time and memory consumption for switching simulations of an MRAM cell using three 100-instance batch scenarios on an Nvidia GTX1080 GPU: sequential launching of 100 instances, simulating a 100 x 100 array of sufficiently separated cells (to prevent the strong magnetostatic coupling), batched approach. This work was supported as part of the Quantum Materials for Energy Efficient Neuromorphic Computing, an EFRC funded by the U.S. DOE, Office of Science.

High-performance micromagnetics on GPUs for batched simulations

P3: Micromagnetic Modelling II

technical paper

The 65th Annual Conference on Magnetism and Magnetic Materials (MMM 2020) was held in a virtual form. The COVID-19 pandemic, with its associated uncertainty about international travel and the ability to gather in large groups, meant that the Steering Committee decided to cancel the face-to-face meeting that had been planned in Palm Beach, Florida November 2-6, 2020.

We were reluctant to give up the benefits of meeting face-to-face, and know that shifting to a virtual conference is a significant change for this community. Nevertheless, we are first and foremost concerned about the health and safety of all our Conference participants and their loved ones on returning home after travelling.

The MMM 2020 Steering Committee was working exceptionally hard, along with our sponsors, the IEEE Magnetics Society and AIP Publishing, to make this event serve the needs of our participants, who live in time zones all around the globe, in the best possible way. 

You need a ticket to explore this event.

If you have already registered for MMM 2020, and you have already created an account, then you
just need to login (above).

If you have already registered for MMM 2020, but you have not yet created an account, then you
need to click on &quot;create an account&quot; (above) and then login.

Forgot your password? Go to
https://www.underline.io/log-in

If you have not yet registered for MMM 2020 but would like to do so, go to
https://magnetism.org/register

Still need help? Contact our team!


Xueyang Wang

1

SHORT BIO

Presentations

High-performance micromagnetics on GPUs for batched simulations

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES