Boston Limited’s ANNA Volta embarks as the company’s latest addition to its machine learning armory. Powered by four NVIDIA Tesla V100 GPUs, the compact 1U GPU solution is highly viable for a broad range of industries. GPU technology plays a major role in solving many complex issues, from computer vision and speech recognition to natural language processing to name a few. The ANNA Volta is fully customizable and said to be extremely cost-effective. The dense solution, also available in 4U featuring up to eight V100s, can be supplied with or without a software stack and can also be leveraged in the cloud, depending on user requirements.
Housing four GPUs in the front, no PCI-E switches or bottlenecks are used and all GPUs are connected directly to the PCI-E Gen 3 bus for maximum performance and minimal latency. In addition to the four GPUs, the 1U server provides 4x PCI-E Gen 3 slots for InfiniBand or Omni-Path allowing strong RDMA performance. NVIDIA® Volta™ technology takes advantage of the NVLink interconnect technology in return delivering faster results. Twice the throughput of the previous generation enables more advanced modelling and data parallel approached for strong scaling.
NVIDIA Volta architecture is purpose-built to act as the engine of computers that learn, see, and simulate our world, and as a result, Boston's appliance promises to deliver fast, absolute performance to HPC and deep learning workloads with infinite computing needs. Available immediately, the Boston ANNA Volta is suitable for a variety of use cases, particularly scientific research groups looking for either on-premise or cloud-based solutions for their deep learning algorithms. Need more information? Then visit Boston Limited and NVIDIA.