Liqid delivers NVIDIA GPU supercomputer rack as it ramps innovation in bare-metal composable infrastructure
13 Nov 2018 Dallas - Liqid, a provider of composable disaggregated infrastructure (CDI) and NVMe solutions and services, is demonstrating a composable GPU supercomputer capable of delivering bare-metal orchestration and provisioning for up to 64 NVIDIA Tesla V100 Tensor Core GPUs, as well as other innovations in composable and solid-state technologies at Supercomputing 2018 (booth # 3841). Through collaboration with a broad spectrum of industry leaders and continued innovation in the composable PCIe fabrics for bare metal resource allocation, the company continues to deliver the resource agility, data performance and hardware disaggregation necessary for the age of Artificial Intelligence (AI).
"Liqid continues to lead innovation in resource utilization and infrastructure disaggregation with our PCIe-based fabric solutions, delivering the ability to right-size physical infrastructure dynamically", stated Sumit Puri, CEO and Cofounder, Liqid. "By combining our infrastructure with powerful NVIDIA GPUs, we have redefined what composability means for the IT industry as a whole. We look forward to continuing to work with NVIDIA and other partners to architect the future of the AI-driven data centre."
At SC18, Liqid is showcasing several new advancements in composable infrastructure technology, including:
- A high-performance, fully composable GPU supercomputer comprising two Dell R740 nodes and 16 32 GB NVIDIA Tesla V100 Tensor Core GPUs, and capable of scaling up to 64 GPUs
- Live demos of the Liqid composable platform highlighting the ability to compose and dynamically reallocate bare-metal of GPU, CPU, NVMe and networking resources for real-world workloads, while providing the user dramatically higher resource utilization
- The Liqid Element solid-state NVMe PCIe fabric solution delivering fully composable ultra-low latency data performance of 1.25 million IOPS for storage-centric tasks
- Liqid's 20-GPU JBOG (just a bunch of GPUs) for large-scale supercomputing installations for applications in AI and deep learning, cloud computing, enterprise and HPC
"With NVIDIA Tesla V100 GPU technology providing the data performance backbone for AI breakthroughs, Liqid is delivering greater GPU scalability and bare-metal GPU orchestration with their composable solutions", stated Paresh Kharya, director of accelerated computing at NVIDIA. "As AI pushes into all industries, solutions providers like Liqid are accelerating adoption of these powerful applications across verticals and reducing the overall data center footprint for data-intensive, high-performance applications."
You can see live demonstrations of Liqid composable solutions at SC18, booth #3841.