Leader in Industrial Computing Solutions Unveils Latest Series of Servers Optimized for Deep-Learning
Advantech, a leading provider of advanced hardware for global IoT and automation, will exhibit its new line of high-powered scalable NVIDIA GPU servers, the SKY-6000 series, at the 2018 Supercomputing Conference (SC18) in Dallas, TX. The new SKY-6000 series servers sit at the forefront of advancement in utility for deep learning in artificial intelligence applications. While typical servers are only able to support 3 or 4 GPUs, the servers in the SKY-6000 line can support up to 10, yielding a vast increase in power and capability.
All SKY-6000 model variants are validated to the new NVIDIA® Tesla® V100 Tensor Core GPU, the most advanced data center GPU ever built to accelerate AI, HPC and graphics. The servers are backed by Advantech’s singular commitment to value and service, including long-life support, revision control and the ability to customize products, a level of service and support unique in the field.
“The annual Supercomputing Conference is an excellent and unique forum for us to demonstrate the capabilities of the new SKY-6000 servers, which we think represent the very best,” said James Yung, Advantech Product Manager. “We are eager for the chance to show off the full range of SKY-6000 capabilities, optimized with the top processing hardware from our valued partner, NVIDIA, to professionals in the fields of AI and supercomputing.”
The SKY-6100 is a 1U rackmount server that offers dual Intel Xeon scalable processors, up to 512GB DDR4 2666 MHz ECC-REG type memory, IPMI Function remote management support, expandable card decks, and 1200W 1+1 redundant power supply with 80 PLUS Platinum level certification. The SKY-6200, Advantech’s 2U rackmount, includes all of the features of the 6100, but with memory expansion up to 768GB and enhanced card deck options, while the 4U model, the SKY-6400, offers DDR4 REG 2666/2400/2133/1866 MHz DIMM memory up to 384GB in addition to the comprehensive features of the other models. The versions described above support 5, 4 and 4 GPUs, respectively.