HPCBOX Platform
Everything you need for automating and autoscaling your HPC workflows
Why HPCBOX?
Only HPCBOX lets you have fine grained control on your HPC cloud resources and applications from a desktop-centric interface. HPCBOX’s revolutionary intelligent-workflow capability combines cloud infrastructure, applications and managed services to bring you supercomputer technology with the ease of use of a personal computer.
Desktop Experience
Experience the full desktop experience on your HPC cluster and effectively execute all your applications directly on a supercomputer. Whether you are working with a CAE application, Python Notebook for your AI/ML workloads, or developing your application from scratch in .NET, C++ or any other programming language, work effortlessly in a fully interactive graphical desktop environment and increase your productivity.
Native execution of all applications makes it easy to migrate from your local computer to a powerful supercomputer in no time.
AutoScaler
The Intelligent AutoScaler, built into the HPCBOX platform, automatically starts and stops required number of Compute, GPU and CUDA workers suitable for a particular user job. Furthermore, the AutoScaler can automatically identify idle workers and power them off when there are no user jobs waiting to be executed on the HPCBOX cluster.
Combine standard and spot-priced worker nodes to optimize your spend and automatically reschedule computations when spot-priced instances are pulled out of the cluster.
Automation
Design, Test and Optimize your HPC pipelines. Once ready, deploy the automated pipeline for consumption from other systems and by other users. Use the easy to access Web Interface or build your own application, for example on Microsoft Power Platform, to trigger your tested and automated HPC workflow.
Effortlessly pass data between steps in your pipeline by creating Input and Output Links before deployment.
Workflow Editor
Build complex HPC simulation pipelines on a visual drag-and-drop canvas. Choose from 25+ built-in applications — OpenFOAM, ANSYS Fluent, Star-CCM+, LS-DYNA, and more — or create your own custom applications. Connect steps to define execution order, configure each step’s options, and run directly from the editor with real-time status tracking.
- Drag-and-drop canvas with 25+ HPC applications
- Multi-tab editing for concurrent workflows
- Real-time execution status on every step
- Save, load, export, and generate batch scripts
- Auto-layout and dark mode support
Flow Controls
Go beyond linear pipelines. Add decision nodes for conditional branching, split and join nodes for parallel execution, and loop nodes for parameter sweeps — all validated automatically before execution. Model real-world simulation workflows where steps depend on results, run concurrently, or repeat across configurations.
- Decision nodes — branch on success or failure of any step
- Split / Join — run multiple branches in parallel and synchronize
- Loop nodes — iterate over numeric ranges, value lists, or file contents
- Automatic validation — structural checks before save or execute
AI Assistant
Experimental
Create and modify HPC workflows using natural language. Describe your simulation pipeline — the AI assistant identifies the right applications, configures their options, builds the flowchart, validates the structure, and saves a ready-to-run workflow. Works with multiple LLM providers so you can choose the model that fits your organization’s requirements.
- Create complete workflows from a text description
- Modify existing workflows conversationally — “add a decision step after the solver”
- Context-aware — operates on the workflow open in your editor
- Anthropic Claude, OpenAI GPT-4o, Azure OpenAI, and Ollama (local models) supported
- Privacy-first — only workflow metadata is sent, never simulation data
MCP Server
Experimental
Connect your favorite AI development tools to HPCBOX using the open Model Context Protocol (MCP) standard. Use Claude Code, VS Code Copilot, Cursor, or any MCP-compatible agent to list applications, create workflows, and manage your simulation pipelines — all from your existing development environment. Point the agent at an online tutorial and get a ready-to-run workflow in seconds.
- Open standard — works with any MCP-compatible AI client
- Remote access via SSH — agents on your laptop can manage cluster workflows
- 9 tools — list apps, create, update, validate workflows, get guides and examples
- Custom apps auto-discovered — no extra configuration needed
- No data sent to LLMs — the MCP server is local, only the AI client communicates with models