How to balance your HPC hardware with Tyan® and AMD
Tyan's AMD-powered server hardware can balance the demands of compute, storage, memory and IO for specific application areas in HPC, Deep Learning, Big Data and IO-heavy computing.
High-performance computing (HPC) is no longer the preserve of governments and research institutions. Many businesses and organizations are working with ever larger amounts of data, seeking new insights in areas that include materials and molecular systems, financial markets, oil and gas, physics and bioscience. These diverse areas, and their specific workload requirements, make different demands on computing infrastructure.
IO-heavy HPC computing applications, for example, are commonly IO-heavy, requiring fast read/write access to storage. In contrast, Accelerated HPC and Deep Learning workloads benefit from highly parallelized GPU architectures, while Big Data analytics need server designs that prioritize bulk storage. A general purpose server can handle all these tasks. But greater performance can be achieved by customizing server design to maximize the hardware features that improve performance specific to each different application.
For IO-heavy workloads, NVMe provides a robust solution to increasing IOPS performance over previous SATA storage devices. Server designs that favor NVMe, such as the Tyan Transport CX GC79A-B8252, can deliver high performance and low latency, leading to superior responsiveness for a variety of IO-intensive tasks. Such workloads can include medical PACS, genome sequencing and weather forecasting.
The AMD EPYC 7003 processor with AMD 3D V-Cache technology is designed to be the world's highest performance x86 server processor for technical computing. It's a smart solution for the modern data center that helps to balance compute, storage, memory and IO. This not only contributes to lower TCO, but greater energy efficiency and data security.
For Accelerated HPC and Deep Learning (DL) computing, the hardware demands are notably different. DL workloads (e.g. computer vision and NLP) can often take thousands of hours of compute time, so multiple GPUs are often used to improve performance. The Tyan Transport HX TN83-B8251 is an example of a design that's specifically tailored to this workload type, supporting up to 4DW / 8SW GPU cards for parallel processing at scale.
Alongside its extensive GPU support, the HX TN83-B8251 server is a 2U dual socket AMD EPYC 7003 Series processor-based server platform. It includes support for PCIe Gen.4, 16 DDR4 memory slots and fast storage with eight 3.5-inch SATA/NVMe U.2 drive bays.
Built for the most demanding HPC workloads, the Tyan Transport series is a line of world class barebones server systems based on AMD's latest technology, including 3rd Gen AMD EPYC™ processors with AMD 3D V-Cache™. For more information about our newest products, view the Tyan/AMD Product Catalog 2022.
High-performance computing (HPC) is no longer the preserve of governments and research institutions. Many businesses and organizations are working with ever larger amounts of data, seeking new insights in areas that include materials and molecular systems, financial markets, oil and gas, physics and bioscience. These diverse areas, and their specific workload requirements, make different demands on computing infrastructure.
IO-heavy HPC computing applications, for example, are commonly IO-heavy, requiring fast read/write access to storage. In contrast, Accelerated HPC and Deep Learning workloads benefit from highly parallelized GPU architectures, while Big Data analytics need server designs that prioritize bulk storage. A general purpose server can handle all these tasks. But greater performance can be achieved by customizing server design to maximize the hardware features that improve performance specific to each different application.
For IO-heavy workloads, NVMe provides a robust solution to increasing IOPS performance over previous SATA storage devices. Server designs that favor NVMe, such as the Tyan Transport CX GC79A-B8252, can deliver high performance and low latency, leading to superior responsiveness for a variety of IO-intensive tasks. Such workloads can include medical PACS, genome sequencing and weather forecasting.
Powered by AMD EPYC™ 7003 Processors with 3D V-Cache™
The Tyan Transport CX GC79A-B8252 is a 1U dual-socket AMD EPYC 7003 Series processor-based server platform, designed for CSP high-density compute applications. To this end, it features 32 DDR4 DIMM slots, two standard PCIe Gen.4 x16 expansion slots, and an OCP 3.0 LAN mezzanine slot. The GC79A-B8252 design offers 12 2.5-inch drive bays with up to 12 NVMe U.2 support provided.The AMD EPYC 7003 processor with AMD 3D V-Cache technology is designed to be the world's highest performance x86 server processor for technical computing. It's a smart solution for the modern data center that helps to balance compute, storage, memory and IO. This not only contributes to lower TCO, but greater energy efficiency and data security.
For Accelerated HPC and Deep Learning (DL) computing, the hardware demands are notably different. DL workloads (e.g. computer vision and NLP) can often take thousands of hours of compute time, so multiple GPUs are often used to improve performance. The Tyan Transport HX TN83-B8251 is an example of a design that's specifically tailored to this workload type, supporting up to 4DW / 8SW GPU cards for parallel processing at scale.
Alongside its extensive GPU support, the HX TN83-B8251 server is a 2U dual socket AMD EPYC 7003 Series processor-based server platform. It includes support for PCIe Gen.4, 16 DDR4 memory slots and fast storage with eight 3.5-inch SATA/NVMe U.2 drive bays.
Customizing servers for Big Data computing
Finally, and not unsurprisingly, Big Data computing requires a different approach again. This type of computing workload is tasked with analyzing and extracting information from large or complex data sets. This requires platforms that are specifically designed for high-density bulk storage. Enter the Tyan Transport CX TN73-B8037, a 2U multi-node server platform with four front-serviced compute nodes. Each node supports one AMD EPYC 7003 Series processor, four 2.5-inch NVMe/SATA drive bays, eight DDR4 DIMM slots, plus a pair of standard PCIe Gen.4 x16 expansion slots, and two internal NVMe M.2 slots.Built for the most demanding HPC workloads, the Tyan Transport series is a line of world class barebones server systems based on AMD's latest technology, including 3rd Gen AMD EPYC™ processors with AMD 3D V-Cache™. For more information about our newest products, view the Tyan/AMD Product Catalog 2022.