Dec. 30, 2018, Taipei, Taiwan

Neousys Technology today announced the latest ruggedized AI inference platform, Nuvo-7164GC series. A compact embedded computer supporting NVIDIA® Tesla® P4/ T4 and Intel® 8th-Gen Core™ i Processor, designed for advanced inference acceleration applications such as voice, video, image and recommendation services.

GPU-aided computers offer significantly more processing power with greater number of parallel processors over conventional CPU computing. With NVIDIA® Tesla® series GPUs, it can deliver up to 40X higher inference capability compared to just CPUs. Nuvo-7164GC supports Tesla® P4 GPU, featuring 5.5 TFLOPS in FP32 and Tesla® T4 GPU, featuring 8.1 TFLOPS in FP32 and 130 TOPs in INT8 for real-time inference based on trained neural network model.

Thanks to Neousys' patented Cassette and air tunnel design on Nuvo-7164GC which guides the intake air to flow through the passive heat sink of NVIDIA® Tesla® P4/ T4, effectively dissipating the heat generated by the GPU. This promising design guarantees system ambient operating temperature up to 60°C with sustained 100% GPU loading.
Nuvo-7164GC also incorporates cutting-edge I/O technologies to boost overall system flexibility, functionality and performance. It has an M.2 NVMe interface that supports disk read/ write speeds over 2000 MB/s and USB 3.1 Gen2/ GbE ports for fast data transfer, such as acquiring HD video data. In addition, Nuvo-7164GC supports MezIO™interface with various IOs such as RS-232/ 422/ 485, isolated DIO, ignition power control, USB3.1 Gen1 and GbE with 802.3at PoE+. "With the combination of a fast CPU and inference accelerator GPU, Nuvo-7164GC is the ideal inference platform for artificial intelligence applications." said Dennis Chen, product manager at Neousys.
- Read More about Neousys Ruggedized AI Inference Platform Supporting NVIDIA® Tesla P4/ T4 and Intel® 8th-Gen Core™ Processor:
https://www.neousys-tech.com/product/product-lines/edge-ai-gpu-computing/nuvo-7164-8th-coffeelake-nvidia-tesla-t4-p4-gpu-embedded-computing