Rebellions
- 6 profiles
- 3 categories
Featured Products
All Products & Services
ATOM™ is a cost-efficient AI accelerator designed to deliver maximum value for small-sized data centers and enterprise applications. Validated by MLPerf benchmark results, ATOM™ is engineered for outstanding hardware efficiency, enabling exceptional performance at a competitive price. Equipped with 16 GB of GDDR6 memory and a 256 GB/s bandwidth for ultra-low latency, ATOM™ provides an unbeatable performance-per-dollar ratio, particularly for small language model (SLM) applications. Its optimized design ensures efficient AI processing for organizations looking to maximize their data center investments.
ATOM™-Lite is a compact and energy-efficient AI accelerator designed for edge inference applications. Operating in a Half Height Half Length (HHHL) form factor, it seamlessly integrates into workstations, edge devices, and laboratory environments. Despite its small size, ATOM™-Lite delivers impressive performance, achieving 32 TFLOPS (FP16) and up to 256 TOPS (INT4), all while consuming a maximum of 65W. This makes it an ideal solution for scenarios where space and power efficiency are critical.
ATOM™-Max is a high-performance AI accelerator developed by Rebellions Inc., designed to meet the demanding needs of large-scale AI inference workloads in data centers and enterprise environments. Leveraging advanced architecture, ATOM™-Max delivers exceptional computational power and efficiency, making it an ideal solution for organizations seeking to enhance their AI capabilities.
REBEL is a high-performance AI accelerator developed by Rebellions, designed to meet the demands of hyperscale workloads. Utilizing advanced HBM3e chiplet technology, REBEL offers exceptional compute density and memory bandwidth, making it ideal for large-scale AI systems. Its energy-efficient design ensures maximum performance while minimizing operational costs, providing a scalable solution that grows with organizational needs. REBEL's architecture supports seamless integration from single cards to multi-rack configurations, accommodating the most demanding AI models and applications.
REBEL-Quad is a high-performance AI accelerator designed to meet the demands of hyperscale workloads. Utilizing advanced HBM3e chiplet-based architecture, it delivers exceptional energy efficiency and computational power, making it ideal for large-scale AI systems. By integrating four dies onto a single IC chip through state-of-the-art chiplet packaging, REBEL-Quad offers scalable solutions from single-card deployments to multi-rack configurations, effectively supporting multi-modal models and enhancing the sustainability of AI data centers. REBEL-Quad addresses the critical need for efficient and scalable AI acceleration in hyperscale environments. By delivering top-tier performance with reduced energy consumption, it enables organizations to deploy state-of-the-art AI services more cost-effectively. Its scalable design ensures adaptability to evolving AI workloads, providing businesses with a competitive edge in the rapidly advancing AI landscape.
