Supercomputing 2024

We're going to SC24!

Boston Limited can’t wait for this year’s Supercomputing 2024, where we will be showcasing cutting-edge technology solutions designed to address the growing demands of AI, machine learning, high-performance computing (HPC) and sustainability. At our booth, we’ll feature a range of advanced hardware and cooling solutions from our partners, each designed to empower organisations with enhanced performance and energy efficiency.

Explore the future of computing with these key technologies:

Grace Superchip Server (ARS-221GL-NR)

The Grace Superchip Server is a 2U system built for high-performance computing and AI workloads. Powered by NVIDIA’s Grace CPU Superchip, it supports up to two PCIe-based GPUs, delivering the power necessary for deep learning, large language models (LLMs) and complex analytics. Whether it's data science, simulation or HPC applications, the Grace Superchip Server offers optimal performance with impressive energy efficiency.

Benefits:

Achieve 10x faster performance for massive data processing tasks.
Supports up to two NVIDIA GPUs, perfect for GPU-accelerated workloads.
144-core energy-efficient NVIDIA Grace CPUs, reducing power consumption while enhancing computational output.

ZutaCore AMD Hyper (AS-1125HS-TNR)

ZutaCore® is leading the transformation to an eco-safe data industry with HyperCool, a two-phase liquid cooling solution for the data centre. It uses a direct-to-chip method — one of the most effective forms of chip cooling — to apply waterless heat transfer fluid directly to the chips to extract and disperse heat. Since no water is used, equipment is fully protected from leaks, corrosion, and other water-related threats.

Benefits:

Cools the hottest processors enabling the highest sustained performance with ZERO throttling, server densification, and reduces power usage.
HyperCool supports up to and beyond 2800 watts per processor in a 1U server form factor.
Reduces Energy Consumption - Saves up to 80% power consumption compared to air-based cooling.

Boston Igloo AI+

A high-performance, scalable storage solution designed to support AI workloads without compromising on accelerator investment. Leveraging ultra-low latency and NVMe technology, it delivers exceptional throughput to meet the needs of AI model training and data processing. Striking the optimal balance of cost-efficiency and performance, the Boston Igloo AI+ enables multiple GPU based servers (such as NVIDIA DGX) to access shared storage, ensuring faster model development without the need for extensive storage investment. With the ability to scale as AI compute needs grow, it’s ideal for organisations seeking to expand their AI infrastructure while optimising data access and processing.

Benefits:

Delivers up to 80GB/s+ storage throughput in the 2U form factor and 40GB/s+ in the 1U form factor.
Scalable architecture with up to 737TB of raw capacity.
Built to support evolving AI needs, the Igloo AI+ adapts to your organisation’s growth.

AmpereOne CPU

AmpereOne® is the next generation of Cloud Native Processors from Ampere, extending the family of high-performance processors to new industry leading core counts. The AmpereOne family establishes new levels of performance and VM density for computing solutions at scale.

AmpereOne compliments the existing Ampere® Altra® products and adheres to the same Cloud Native principles. Ampere is more. More cores, more memory bandwidth, more memory capacity, more IO bandwidth, and more Cloud Native features.

Benefits:

High Performance and Predictable with up to 192 cores.
Elastic and Scalable.
Power Efficient and Sustainable.

Gaudi 2 Accelerator Card

The Gaudi 2 Accelerator Card offers an innovative solution for AI training, delivering significant speed-ups for deep learning models. Designed to reduce time-to-train for large models, this card is perfect for businesses working on AI and deep learning at scale.

Benefits:

50% faster time-to-train for large models like GPT-3.
Supports multi-GPU configurations for enhanced parallel training.
Increased efficiency in both training and inference, reducing power consumption and cost.

Castrol Flight Case with Liquid Cooling Demo Unit

Our Castrol Flight Case features an immersive liquid cooling demo unit, highlighting the latest in sustainable cooling for high-performance systems. This technology is a collaborative innovation between Castrol and Submer, offering a significant reduction in energy consumption and operational costs for data centres.

Benefits:

Energy-efficient cooling for high-performance computing systems.
Dramatically reduces operational costs and carbon footprint.
Ideal for organisations seeking sustainable solutions for their data centres.

By offering a variety of advanced systems that target different computing needs, Boston is committed to empowering organisations with the latest in high-performance, energy-efficient technology solutions. Visit us at Supercomputing 2024 to learn more about how these innovations can transform your workflows.