High-performance computing
Achieve your unique business objectives with enterprise-grade agility, scale, performance, and security with the Microsoft Azure HPC portfolio.
Powerful versatility
Designed for HPC workloads. Whether it’s computational fluid dynamics, finite or infinite element analysis, Monte Carlo simulations, reservoir modeling, or autonomous vehicle development, get the HPC models and tools you need, at the scale you need, from Azure.
Extend your HPC environments to the cloud. Extend your HPC beyond your on-premises boundaries by adding new capabilities, including bursting to the cloud, AI and machine learning, or even full migration.
HPC core application patterns
Based on customer adoption patterns, the Azure HPC portfolio is commonly used across four application patterns. Discover a new generation of powerful, scalable Azure applications and processes designed to empower engineers, scientists, and researchers to innovate beyond traditional HPC scenarios.
Simulation
Simulation for highly efficient research and development message passing interface (MPI) workflows in manufacturing, engineering, life sciences, financial services, and energy.
Computational fluid dynamics (CFD) simulationsDeep learning
Deep learning workloads incorporating Azure AI and machine learning technologies for predictive analytics and insights for smart IoT devices, increased productivity, and deeper collaboration.
Machine learning diagram and overviewRendering
Rendering processes for workloads in media production and engineering design that deliver intended results in a fraction of the time it normally takes.
Rendering architecture scenario diagramsVisualization
Visualization of your workspaces and applications on virtual PCs to securely minimize client PC investments, scale-up on demand, and reduce IT support issues.
Fast networking
Azure is the only public cloud with support for InfiniBand, which provides the interconnect speed and bandwidth necessary to support all MPI types and versions, including Open MPI, MVAPICH2, Platform MPI, Intel MPI, and SparkRDMA.
Powerful infrastructure
Get extraordinary versatility and scalability with high memory virtual machines (VMs), graphic-intensive GPUs, high IOPS storage, or a fully managed, dedicated supercomputing service on Cray.
Seamless orchestration
Avoid the time and expense of re-architecting your applications for the cloud by simply extending the HPC resource environment you already have to Azure. Clone your clusters to the cloud with Azure Cycle and manage your application workflows with Azure Batch.
Azure HPC platform services
Get vast compute resources with a cloud platform built for global scale and outfitted with services specifically geared towards handling HPC workloads.
HPC workflow services
Take advantage of end-to-end application workload management and orchestration services that make Azure a logical extension of your HPC environment or application.
Transformative services
Innovate the next generation of applications with predictive analysis and machine learning model training against large HPC data sets.
Fast, secure networking
Establish private, secure tunnels for hybrid cloud connectivity, and take advantage of Linux RDMA with InfiniBand for MPI workloads within your data center.
Optimized infrastructure and data
Find the right resources at nearly unlimited scale to keep up with your HPC apps, including CPU- and GPU-based virtual machines, automatic scaling, and fast storage.
Azure HPC for industries
Autonomous driving
Design and build autonomous driving vehicles without the hassle of computing bottlenecks from processing big data sets.
Advanced driver assistance systems (ADAS)
Accelerate time to market by using simulations to optimize sensor and algorithm performance.
Crash test simulation
Reduce the costs associated with real crash testing and gain deeper insights from your data output.
Learn how to set up an Azure HPC cluster for automotive engineering simulations by reading the free Run Star-CCM+ in an Azure HPC Cluster white paper.
Download nowRisk compute and analytics (FRTB)
Focus on risk analysis and related regulatory compliance requirements in a flexible, security-enhanced manner.
Trade modulation and simulation
Test the effectiveness of your trading strategy without putting real money at risk.
Risk reporting
Speed up processing and get faster results that are perfectly aligned with rapidly changing market conditions.
Learn more about enabling the financial services risk lifecycle with Azure and R.
Actuarial risk modeling
Run what used to be thousands of hours of analysis in just minutes so that you have more time to review, evaluate, and validate results.
Catastrophic risk modeling
Better predict and mitigate risk from catastrophes and natural disasters using probabilistic or deterministic models.
Risk analytics
Accelerate and streamline your risk processes while adapting to changing practices, demands, and regulations.
Learn more about enabling the financial services risk lifecycle with Azure and R.
Genomic analysis
Get the computational power you need for DNA sequencing and analysis workloads with data management for large data sets.
NONMEM clinical trial simulation
Create population pharmacokinetic and pharmacokinetic-pharmacodynamic modeling workloads.
Molecular dynamics
Increase the accuracy of mathematical formulas used in simulated atom and particle movement analysis.
Computational fluid dynamics (CFD)
Dramatically accelerate your time to results for mechanical design improvements and optimization insights.
Finite element analysis (FEA)
Improve FEA accuracies and element behavior predictability for physical effects such as mechanical stress, fatigue, and fluid flow.
Computational chemistry
Discover possibilities for molecular manufacturing with atomic precision and accurately model/predict properties of new materials.
Learn how to set up an Azure HPC cluster for automotive engineering simulations by reading the free Run Star-CCM+ in an Azure HPC Cluster white paper.
Download nowFast media rendering
Greatly reduce the time required for 3D-to-2D image conversion with video, film, or imagery content from non-linear editing and production systems.
Scalable post-production
Take advantage of the most computationally intensive VFX tools and processes without missing tight production and market deadlines.
Content optimization
Improve consistency and discoverability of new media content with remote content ingestion, in-line transcoding, speech indexing, and metadata tagging.
Learn more about how to use Azure for media rendering workflows.
Reservoir modeling
Produce more informative, upscale grids using single-node jobs, tightly coupled multi-node jobs, or dedicated supercomputing workloads.
Seismic processing and imaging
Quickly process and deliver results from increasingly growing sets of seismic data.
Computational chemistry
Simulate physical properties of streams, use improved sensitivity analysis, and better describe new compounds and materials.
Learn more about how the oil and gas industry is using Azure Batch to accelerate ROI and minimize risk.
Fully managed supercomputing service on Azure
Get a dedicated, fully managed, single tenancy Cray XC series or Cray CS series supercomputer. Take advantage of ClusterStor high-performance storage with your Cray deployment or with Azure H-series VMs. Experience great features like:
- Dedicated, customized, fully managed supercomputers.
- Single tenancy on bare metal for complete control and privacy over your computing environment.
- No data movement. Your data stays on the same Azure network as your RDMA/GPU/FPGA VMs.
- Transformative services such as AI, deep learning, and advanced analytics.
Workflow optimization on Azure
Achieve more with proven processing and operational tools and services for HPC application workflows, including:
Application workload scaling using either HPC-optimized CPU, GPU, or FPGA-based VMs, or with dedicated, single tenant environments for hyperscale class applications.
Learn moreCluster and environment scaling with built-in orchestration and management, saving you the time and cost of re-architecting your applications.
Learn moreOn-demand bursting though API/CLI pathways for democratized access to cloud-scale HPC resources.
Learn moreLow-latency hybrid storage access to allow you to easily manage your work in the cloud across resources.
Learn moreDocumentation, training, and demos
Resources
Training
Learn how to run large-scale parallel and HPC applications efficiently in the cloud with Azure Batch in this free 45-minute learning module from Microsoft Learn.
Get startedLearn what customers are doing with Azure HPC
Simon Blaquière, Reinsurance Actuarial Manager, AXA Global P & C“The ability to do large-scale computing is essential if you’re looking at something like hurricane risk, because it’s a complex event affecting an extremely large geography. With Azure, we have the computing power to model such events and provide more accurate information to our clients. As a result, they can get better coverage.”
Andy Lee, Cofounder and Chief Technology Officer, NeuroInitiative LLC“Our simulation tool, powered by HPC resources in Azure, has the potential to cut today’s 12-to-20-year drug development period in half.”
Mark Everest, IS Development Manager, Renault Sport Formula 1 Team“Before every race or test, we have to compile a report for shipping parts to and from the event so that we can ensure that every single part will get to the track on time. That report used to take someone a week to compile, but we’ve gotten that down to a fraction of that time now—just a few hours.”
Microsoft partnered up with oil and gas giant Schlumberger to launch the company’s first fully commercial SaaS offering, the high-resolution reservoir modeling application INTERSECT, on Azure.
Dave Blommers, Global Head of Technology, Mr. X“With the scalability we now have with Microsoft Avere vFXT for Azure, we can predict demand more accurately than ever before and respond to it much better, especially when dealing with last-minute changes.”
Solution architectures
Media and entertainment rendering architecture
This HPC media rendering solution architecture shows Azure CycleCloud monitoring a Pixar Tractor pipeline manager and orchestrating burst compute node capacity on-demand using Azure low-priority Virtual Machines Scale Sets. An Avere vFXT cache makes data from the existing on-premises filesystem and Azure Blob storage available to compute nodes in Azure.
- 1 Operations team uses Azure CycleCloud to configure and launch rendering pipeline cluster.
- 2 Azure CycleCloud orchestrates virtual machine (VM) creation and software configuration for head nodes, license servers, and Avere vFXT Cache.
- 3 Artist submits a render job to the Pixar Tractor pipeline manager.
- 4 Azure CycleCloud detects the change in job queue depth and autostarts render farm nodes in Virtual Machines Scale Sets with location, SKU, and configuration customized by job requirements.
- 5 Render pipeline manager (head nodes) executes render jobs on the new render farm VMs.
- 6 Render jobs pull artifacts from on-premises and Azure Blob storage as needed from NFS-mounted Avere vFXT.
- 7 As each job finishes rendering, resulting artifacts are written back to storage through the Avere vFXT.
- 8 As job queue empties, Azure CycleCloud auto-stops render farm VMs to reduce cost.
Big compute with Azure Batch
Big compute and high performance computing (HPC) workloads are normally compute intensive and can be run in parallel, taking advantage of the scale and flexibility of the cloud. The workloads are often run asynchronously using batch processing, with compute resources required to run the work and job scheduling required to specify the work. Examples of Big Compute and HPC workloads include financial risk Monte Carlo simulations, image rendering, media transcoding, file processing, and engineering or scientific simulations.
This solution implements a cloud-native application with Azure Batch, which provides compute resource allocation and management, application installation, resource auto-scaling, and job scheduling as a platform service. Batch also offers higher level workload accelerators specifically for running R in parallel, AI training, and rendering workloads.
This solution is built on the Azure managed services—Virtual Machines, Storage, and Batch. These services run in a high-availability environment, patched and supported, allowing you to focus on your solution.
- 1 Upload input files and the applications to your Azure Storage account.
- 2 Create a Batch pool of compute nodes, a job to run the workload on the pool, and the tasks in the job.
- 3 Batch downloads input files and applications.
- 4 Batch monitors task execution.
- 5 Batch uploads task output.
- 6 Download output files.
Hybrid risk analysis architecture
This templated risk analysis solution uses Azure HPC compute and GPU virtual machines (VMs) to expand on-premises Tibco GridServer compute to Azure using Azure CycleCloud for auto-scaling integration. The job executes both on-premises and in the cloud by using Avere vFXT fast caching and native NFS access to market data available on-premises.
- 1 Operations team uses Azure CycleCloud to configure and launch risk analysis grid in Azure.
- 2 Azure CycleCloud orchestrates VM creation and software configuration for Tibco Gridserver brokers and HPCCA, in-memory data cache, and Avere vFXT cache.
- 3 Quant (or scheduled batch) submits a risk analysis template workflow to the on-premises Tibco GridServer director. Based on job policies and current on-premises use, the workflow is allowed to burst to Azure to expand on-premises grid capacity.
- 4 The Tibco HPCCA detects the change in queue depth for each Tibco broker and requests additional Tibco engine capacity using the Azure CycleCloud Auto-Scaling API. Azure CycleCloud then autostarts engine nodes in Virtual Machine Scale Sets using the Azure H-series, HB-series, and HC-series VMs to optimize cost and performance and NC-series VMs to provide GPU capacity as required.
- 5 As soon as engine VMs join the Azure Grid, the brokers begin executing tasks to the new nodes.
- 6 Risk jobs pull artifacts from on-premises and Azure Blob storage as needed from NFS mounted Avere vFXT and/or via the fast in-memory cache.
- 7 As each task completes, results are returned to the submitter or driver and data is written back to the in-memory cache, or to NFS storage through the Avere vFXT, as required. Cached data is persisted either on-premises or in Azure Blob storage.
- 8 As task queues drain, the Tibco HPCCA uses the Azure CycleCloud Auto-Scaling API to shrink the compute grid and reduce cost.
HPC infrastructure solutions in the Azure Marketplace
Explore HPC-oriented apps and consulting services in the Azure Marketplace and learn about offerings from Azure HPC partners.
Go to the MarketplaceHPC partners
Get help extending your on-premises HPC workloads to the Azure cloud platform by working with Microsoft integrated systems partners, service providers, and software marketplace partners.
Contact Us
Get started with HPC on Azure. Tell us a little about yourself and an Azure team member will get in touch.