Energy-Efficient AI at Any Scale

Modular Framework for Scalable PODs

System Foundation

Rebellions Scalable Design

Rebellions Scalable Design (RSD) is the architectural foundation that empowers the seamless and flexible support for LLMs across a wide range of sizes. From inter-chip connections to multi-card configurations, servers, and even rack-level datacenter deployments, RSD provides a scalable, energy-efficient, and highly capable solution to cater to all requirements.

Key Benefits

Scalability

Scale out from a single chip to multiple racks effortlessly with RSD’s modular architecture.

Flexibility

RSD boasts exceptional flexibility, made possible by our highly configurable chip and the versatile RBLN Compiler. Our cutting-edge technology not only supports a wide range of models but also incorporates the latest advancements like RAG.

Efficiency

Our lightweight, powerful chip includes only the essential components for inference. It incorporates a memory architecture that not only maximizes efficiency but also maintains superior performance.

Card

ATOM™ Card

Desktop Workstation

~4x ATOM™ Card

Server

~16x ATOM™ Card

Rack

~14x Server

Tensor Parallelism

Partitioning workload at the tensor-level, allowing scalable configurations with minimum latency

High Density Server

Maximum performance at minimum footprint

Faster, Easier, Cost-Effective

For Data centers, Cloud service providers, System integrators, Finance sectors, and the Government

Specifications

Server

Rack (configurable)

Card

~16x ATOM™

~224x ATOM™

TFLOPS (FP16)

512

7,168

TOPS (INT8)

2,048

28,672

Memory (GDDR6)

GDDR6 256 GB, 4 TB/s

GDDR6 3.5 TB, 56 TB/s

SRAM

1 GB

14 GB

TDP

2.5 kW

35 kW

Host Interface

PCIe Gen5 x16

Ubuntu Linux, RHEL & Rocky Linux

Cloud Support

Kubernetes, OpenStack

Server

Card

~16x ATOM™

TFLOPS (FP16)

512

TOPS (INT8)

2,048

Memory (GDDR6)

GDDR6 256 GB, 4 TB/s

SRAM

1 GB

TDP

2.5 kW

Host Interface

PCIe Gen5 x16

Ubuntu Linux, RHEL & Rocky Linux

Cloud Support

Kubernetes, OpenStack

Rack (configurable)

Card

224x ATOM™

TFLOPS (FP16)

7,168

TOPS (INT8)

28,672

Memory (GDDR6)

GDDR6 3.5 TB, 56 TB/s

SRAM

14 GB

TDP

35 kW

Host Interface

PCIe Gen5 x16

Ubuntu Linux, RHEL & Rocky Linux

Cloud Support

Kubernetes, OpenStack

Certified Hardware Partnerships

Rebellions powers your AI system through our network of certified hardware partners. Meet Rebellions: stable and ready to deploy.

Key Features

Parallel Computing

By strategically partitioning the computational workload at the tensor level, RSD achieves efficient parallel computing and ensures minimal latency.

Software Stack

Rebellions Software Stack is designed to optimize hardware performance to its fullest potential. RBLN SDK effortlessly integrates with popular industry models, prioritizing seamless integration and user-friendly experiences.

Learn more

Flexible Deployment

A one-stop solution for quick deployment, RSD comes pre-configured and ready to use based on your specific requirements. Whether you need to launch your project as soon as possible or expand your business in the future, RSD's flexibility allows for a swift time-to-market and effortless scalability.

See also...

ATOM™

Cost-Efficient, Powerful AI Acceleration for Small-Sized Data Centers

Explore ATOM™

ATOM™-Max

Boosted Performance for Large-Scale Inference

Explore ATOM™-Max

ATOM™-Lite

Low-Power, Yet Highly Powerful AI Inference at the Edge

Explore ATOM™-Lite

ATOM™ SoC

Inference AI Accelerator for Data Centers

ATOM™ is a fast and power-efficient System-on-Chip for AI inference with remarkably low latency, conceived for deployment in data centers or cloud service providers.
With international recognition from the industry-standard benchmark MLperf™ v3.0, ATOM™ can be scaled to accelerate state-of-the-art AI models of various sizes.

Explore ATOM™ SoC

ION™ SoC

Poised to Revolutionize the High-Frequency Trading Market

In the high-frequency trading (HFT) market, a microsecond can make a huge difference. ION™ was specifically tailored for this finance industry. Within just one year of inception, Rebellions introduced this powerful chip to the market, marking a significant milestone in our pursuit of technological excellence.

Explore ION™ SoC

Career

Join our revolution

Contact

Get in touch with us

System Foundation

ATOM™ Series

REBEL

System Foundation

ATOM™ Series

REBEL

Energy-Efficient AI at Any Scale

System Foundation

Rebellions Scalable Design

Key Benefits

Scalability

Flexibility

Efficiency

Card

Desktop Workstation

Server

Rack

Tensor Parallelism

High Density Server

Faster, Easier, Cost-Effective

Certified Hardware Partnerships

Key Features

Parallel Computing

Software Stack

Flexible Deployment

See also...

Cost-Efficient, Powerful AI Acceleration for Small-Sized Data Centers

Boosted Performance for Large-Scale Inference

Low-Power, Yet Highly Powerful AI Inference at the Edge

Inference AI Accelerator for Data Centers

Poised to Revolutionize the High-Frequency Trading Market

Career

Contact