"In working with some of the industry's leading deep learning organisations, our customers have begun looking for specialized Cloud configurations requiring higher core per GPU ratios", stated Mike LaPan, vice president, Cirrascale Cloud Services. "We're pleased to offer the AMD EPYC processor to power these deep learning GPU accelerated applications."
Cirrascale Cloud Services offers a dedicated, bare-metal cloud service with the ability for customers to load their very own instances of popular deep learning frameworks, such as TensorFlow, PyTorch, Caffe 2, and others. AMD EPYC powers these solutions with up to 32 cores, 64 threads, 8 memory channels with up to 2TB of memory per socket, and 128 PCIe-3 lanes coupled with the industry's first hardware-embedded x86 server security solution. Use of the new EPYC processor within the Cirrascale Cloud Services platform will drive greater scale and performance. In turn, this will deliver better performing deep learning analytics environments that can collect and analyze larger data sets much faster.
"We're excited to be further extending our EPYC processor footprint to include Cloud service providers such as Cirrascale", stated Scott Aylor, Corporate Vice President and General Manager, AMD Enterprise Solutions. "EPYC delivers superior memory bandwidth and memory capacity per CPU, and is designed to provide customers with the headroom to scale and the performance required to deliver real results with applications in HPC, AI, and deep learning."