With sweeping hardware and software innovations, Shasta incorporates next-generation Cray system software to enable modularity and extensibility, a new Cray-designed system interconnect, unparalleled flexibility in processing choice within a system, and a software environment that provides for seamless scalability. The U.S. Department of Energy (DOE) also announced that NERSC, the National Energy Research Scientific Computing Center, has chosen a Cray "Shasta" supercomputer for its NERSC-9 system, named "Perlmutter", in 2020. The programme contract is valued at $146 million, one of the largest in Cray's history, and will feature a Shasta system with Cray ClusterStor storage.
Cray supercomputer systems consistently lead the industry in performance and efficient scaling. Shasta continues this leadership into much larger capability systems, starting from a single cabinet and scaling up to exascale and beyond, to answer more questions in all fields in support of extreme-scale science, innovation and discovery. At a time when demand is rising for single systems to handle converged modelling, simulation, AI and analytics workloads, Shasta's data-centric design allows it to run diverse workloads and workflows all on one system, all at the same time.
Shasta's hardware and software innovations tackle the bottlenecks, manageability, and job completion issues that emerge or are magnified as core counts grow, compute node architectures proliferate, and workflows expand to incorporate AI at scale. Shasta eliminates the distinction between clusters and supercomputers with a single new breakthrough supercomputing system architecture, enabling customers to choose the computational infrastructure that best fits their mission, without tradeoffs. With Shasta you can mix and match processor architectures (X86, Arm, GPUs) in the same system as well as system interconnects from Cray (Slingshot), Intel (Omni-Path) or Mellanox (InfiniBand).
"Shasta will usher in a new era of supercomputing and represents a true game-changer at a time when artificial intelligence and analytics are being brought to bear on increasingly large and complex problems, including classic HPC modeling and simulation challenges, across an ever-broadening set of companies and industries", stated Peter Ungaro, president and CEO of Cray. "It is also very exciting to announce that one of the largest contracts in the history of our company was just signed with NERSC. We are honoured to continue our partnership with NERSC and put Shasta to work in support of their broad mission to enable computational and data science at scale."
"Our scientists gather massive amounts of data from scientific instruments like telescopes and detectors that our supercomputers analyze every day", stated Dr. Sudip Dosanjh, director of the NERSC Center at Lawrence Berkeley National Laboratory. "The Shasta system's ease of use and adaptability to modern workflows and applications will allow us to broaden access to supercomputing and enable a whole new pool of users. The ability to bring this data into the supercomputer will allow us to quickly and efficiently scale and reduce overall time to discovery. We value being able to work closely with Cray to provide our feedback on this next generation system which is so critical to extending our Center's innovation."
"Cray is widely seen as one of only a few HPC vendors worldwide that is capable of aggressive technology innovation at the system architecture level", stated Steve Conway, Hyperion Research senior vice president of research. "Cray's Shasta architecture closely matches the wish list that leading HPC users have for the exascale era, but didn't expect to be available this soon. This is truly a breakthrough achievement."
With Shasta, Cray is also announcing Slingshot, a new high-speed, purpose-built supercomputing interconnect. Slingshot advances Cray's industry leadership in scalable network performance and adds capabilities that broaden Cray's market reach. The Cray-developed Slingshot interconnect will have up to 5x more bandwidth per node and is designed for data-centric computing. Slingshot will feature Ethernet compatibility, advanced adaptive routing, first-of-a-kind congestion control, and sophisticated quality-of-service capabilities.
Support for both IP-routed and remote memory operations will broaden the range of applications beyond traditional modelling and simulation. Quality-of-service and novel congestion management features will limit the impact to critical workloads from system services, I/O traffic, and co-tenant workloads, to increase realized performance and limit performance variation. Reduction in the network diameter from five hops - in the current Cray XC generation - to three will reduce latency and power while improving sustained bandwidth and reliability.
"We listened closely to our customers and dug into the future needs of AI and HPC applications as we designed Shasta", stated Steve Scott, senior vice president and CTO of Cray. "Customers wanted leading-edge, scalable performance, but with lots of flexibility and easy upgradeability over time. I'm happy to say we've nailed this with Shasta. The Shasta infrastructure accommodates a wide variety of processor and network options, allowing customers to run diverse workloads on a single system. And it's got the headroom to accommodate increasingly power-hungry processors and accelerators coming in future years. The Slingshot network tightly binds the compute and storage resources in the system, with groundbreaking congestion control to isolate applications from other network traffic, and Ethernet compatibility for data centre and storage integration. We're immensely excited to bring this new network to market to help accelerate our customers' discoveries."
Shasta lets customers fully realize Cray's longtime vision of adapting supercomputing systems to workloads using optimized processing and networking. This becomes additionally valuable as customers are increasingly concerned about choosing the optimal architecture as their workloads rapidly evolve. With Shasta, Cray can incorporate any processor choice - or a heterogeneous mix - with a single management and application development infrastructure. Customers can flex from single to multi-socket processor nodes, GPUs, FPGAs and other processing options that will emerge, such as AI specialized accelerators. Customers can make late-binding decisions on compute technology and not sacrifice capability, because Shasta's design allows tailoring of system density and injection bandwidth to optimize price and performance.
"We are tremendously proud that AMD EPYC processors are part of the next-generation Cray Shasta platform at NERSC", stated Forrest Norrod, senior vice president and general manager, AMD Datacenter and Embedded Systems Group. "The Shasta system represents a new paradigm for supercomputing that is going to take the industry into the exascale era. The Cray Shasta architecture supports both AMD EPYC CPUs and accelerators, including AMD Radeon Instinct GPUs. That combination of CPU and GPU power offers an incredible level of performance for many HPC applications."
"Intel and Cray have a long-standing strategic partnership to deliver supercomputing systems. We are excited to work with Cray to advance supercomputing capability with their new Shasta platform optimized for Intel Xeon Scalable processors and other Intel HPC products. Intel's suite of HPC products provides the building blocks for a balanced design that delivers the performance and scalability needed to tackle the extreme challenges of both HPC and AI on a common infrastructure", stated Trish Damkroger, vice president and general manager, Intel's Data Center Group.
"Marvell and Cray have a long-standing relationship of collaboration to bring Arm-based processors to the high-performance computing environment", stated Raghib Hussain, executive vice president and chief strategy officer, Marvell Semiconductor Inc. "The Cray XC50 supercomputer, now commercially available, integrates Marvell ThunderX2 processors based on 64-bit Armv8-A architecture to bring best-in-class SoC compute performance and memory bandwidth. We congratulate Cray on the introduction of its exascale-class Shasta system and look forward to continuing to work with Cray to usher in the next era of Arm-based supercomputing."
"We are happy to support the new Shasta platform with our smart InfiniBand interconnect solutions, enabling our mutual customers to achieve leading performance and scalability", stated Gilad Shainer, vice president of marketing at Mellanox Technologies. "The combination of InfiniBand's acceleration engines and Cray's software will drive the next generation of HPC and AI applications and workloads."
"The Shasta-based system selected for NERSC's new Perlmutter supercomputer incorporates the most-advanced NVIDIA GPU acceleration platform available, providing the U.S. Department of Energy a clear path to exascale computing", stated Ian Buck, vice president and general manager of accelerated computing at NVIDIA. "Whether used for traditional scientific computing or new, emerging AI and machine learning workloads, Perlmutter will offer the scientists that use it an incredibly advanced acceleration platform, an essential tool to help them unlock mysteries and advance science."
Designed from the ground up to support a decade or more of advancements in computational processing, Shasta eliminates the need for frequent, expensive upgrades and will drive exceptionally low total cost of ownership over the lifetime of the system. Shasta packaging comes in two options: a 19" air- or liquid-cooled, standard datacenter rack and a high-density, liquid-cooled rack designed to hold 64 compute blades with multiple processors per blade. Both options can scale to well over 100 cabinets.
As processor wattage increases over time to boost computational performance, Shasta's flexible cooling design eliminates the need to do forklift upgrades of system infrastructure to accommodate higher-power processors. Cray designed Shasta to support processors exceeding 500 watts with highly efficient cooling, resulting in less waste and lower costs while meeting critical processor density requirements. Shasta systems are also designed to meet warm-water-cooling datacenter standards like W3 and W4 requirements throughout the world.
Shasta systems are expected to be commercially available in Q4 of 2019.