Running a Parameter Server with Runner
Train a tiny MNIST model with a synchronous Parameter Server pattern using Flame Runner: object cache, remote services, and futures.
Read More →
Serverless Runtime with Flame for Security, Performance and Cost
XFLOPS empowers enterprises to build serverless platforms with Flame, delivering a secure, cost-effective, and high-performance runtime environment. Flame is engineered from decades of expertise in elastic workload management, enabling it to handle the most demanding elastic workloads with exceptional efficiency and scalability, e.g. AI agents, Quant, etc.
Key Features
Elastic
Scale your workloads dynamically based on demand with auto-scaling capabilities and resource optimization.
Security
Session-based authentication and authorization for secure access to your elastic workloads which are running in microVMs, e.g. AI agents, Quant, etc.
Cost Effective
Advanced scheduling algorithms that optimize resource utilization and workload distribution across your infrastructure.
Heterogeneous
Support for various hardware configurations including GPUs, TPUs, and specialized accelerators for elastic workloads.
High Performance
Optimized for maximum throughput and performance, ensuring your elastic workloads run at peak efficiency.
Cloud Native
Designed by Cloud Native architecture, make it possible to be deployed on any cloud platform or on-premise.
Introducing Flame
Flame is our flagship distributed engine for elastic workloads, designed to handle the most demanding elastic workloads with unprecedented efficiency and scalability, e.g. AI agents, Quant, etc.
What is Flame?
Flame is a distributed system designed for elastic workloads, providing a comprehensive suite of mechanisms commonly required by various classes of elastic workloads, including AI/ML, HPC, Big Data, and more. Built upon over a decade and a half of experience running diverse high-performance workloads at scale across multiple systems and platforms, Flame incorporates best-of-breed ideas and practices from the open source community.
Key Capabilities
- Scale: Unlike applications running on a single node, Flame scales workloads across multiple nodes to maximize performance acceleration while ensuring fair resource sharing across multiple tenants and sessions.
- Performance: Elastic workloads typically involve tens of thousands of short tasks. Flame leverages cutting-edge features to improve roundtrip times and throughput in large-scale environments, while intelligently sharing runtime within sessions to minimize startup time.
- Security: Flame utilizes microVM as a runtime for enhanced security, with each runtime environment (executor) dedicated to a single session to prevent data leakage. All Flame components communicate using mTLS for secure inter-component communication.
- Flexibility: Flame defines a comprehensive set of general APIs to support multiple user scenarios. Additionally, Flame supports applications across multiple programming languages through gRPC, including Rust, Go, and Python.
Latest from Our Blog
Stay updated with the latest insights, tutorials, and announcements from the XFLOPS team.
Running Monte Carlo Pi with Flame Runner (Python)
Estimate π with a distributed Monte Carlo workflow and see how Runner manages packaging, services, and task aggregation.
Read More →
Customized Session of OpenAI Agent SDK with Flame
Learn how Flame's CommonData feature enables persistent agent conversations and seamless integration with OpenAI Agent SDK's Session for distributed environments.
Read More →