# Snellius Phase 3 GPU > Supercomputer **Wikidata**: [Q132476834](https://www.wikidata.org/wiki/Q132476834) **Source**: https://4ort.xyz/entity/snellius-phase-3-gpu ## Summary Snellius Phase 3 GPU is a supercomputer located in the Netherlands, ranked 94th on the November 2024 TOP500 list. Built by Lenovo and powered by AMD EPYC 9334 processors, it delivers 13.64 petaflops of sustained performance using NVIDIA H100 SXM5 GPUs. The system became operational in 2024 as part of the Snellius supercomputing infrastructure. ## Key Facts - Ranked 94th on the November 2024 TOP500 list of the world's most powerful supercomputers - Manufactured by Lenovo and installed in the Netherlands - Powered by AMD EPYC 9334 32-core processors running at 2.7 GHz - Equipped with NVIDIA H100 SXM5 GPUs with 94GB memory each - Delivers 13.64 petaflops of sustained (Rmax) performance and 40.18 petaflops peak (Rpeak) performance - Consumes 295.67 kilowatts of power during operation - Contains 52,096 processor cores across its compute nodes - Runs Red Hat Enterprise Linux 9.2 as its operating system - Achieved 116.369 teraflops on the HPCG benchmark test - Features InfiniBand HDR200 networking for high-speed inter-node communication ## FAQs ### Q: What is Snellius Phase 3 GPU's primary purpose? A: Snellius Phase 3 GPU serves as a high-performance computing resource for scientific research and computational workloads, providing researchers in the Netherlands with access to GPU-accelerated processing capabilities for complex simulations and data analysis tasks. ### Q: How powerful is Snellius Phase 3 GPU compared to other supercomputers? A: Snellius Phase 3 GPU ranks 94th globally on the November 2024 TOP500 list, placing it among the top 100 most powerful supercomputers worldwide, with 13.64 petaflops of sustained performance suitable for demanding scientific computations. ### Q: What hardware components make up Snellius Phase 3 GPU? A: The system combines AMD EPYC 9334 32-core processors with NVIDIA H100 SXM5 GPUs, connected via InfiniBand HDR200 networking, running on Lenovo hardware with Red Hat Enterprise Linux 9.2. ## Why It Matters Snellius Phase 3 GPU represents a significant advancement in the Netherlands' supercomputing capabilities, providing researchers with access to cutting-edge GPU-accelerated computing power for tackling complex scientific challenges. As part of the broader Snellius supercomputing infrastructure, this system enables breakthroughs in fields ranging from climate modeling and materials science to artificial intelligence and computational biology. The combination of AMD EPYC processors with NVIDIA H100 GPUs positions it at the forefront of heterogeneous computing, allowing for efficient handling of both traditional CPU workloads and modern GPU-accelerated applications. Its ranking among the world's top 100 supercomputers demonstrates the Netherlands' commitment to maintaining competitive high-performance computing resources, which are essential for national research competitiveness and innovation. The system's energy efficiency, delivering substantial computational power while consuming 295.67 kilowatts, also reflects the growing importance of sustainable supercomputing design in an era of increasing energy costs and environmental concerns. ## Notable For - Ranked among the top 100 supercomputers globally on the November 2024 TOP500 list - Features the latest NVIDIA H100 SXM5 GPUs, representing cutting-edge GPU technology for scientific computing - Combines AMD EPYC 9334 processors with GPU acceleration for heterogeneous computing capabilities - Achieves 13.64 petaflops of sustained performance while maintaining reasonable power consumption at 295.67 kW - Part of the Netherlands' Snellius supercomputing infrastructure, supporting national research initiatives ## Body ### Technical Specifications Snellius Phase 3 GPU operates as a heterogeneous computing system built on Lenovo hardware infrastructure. The system's computational heart consists of AMD EPYC 9334 processors, each featuring 32 cores running at 2.7 GHz base frequency. These CPUs are paired with NVIDIA H100 SXM5 GPUs, which provide the system's primary acceleration capabilities for parallel workloads. ### Performance Metrics The supercomputer achieves 13.64 petaflops of sustained performance (Rmax) and 40.18 petaflops of peak theoretical performance (Rpeak). On the HPCG benchmark, which measures performance on more realistic computational tasks, the system delivers 116.369 teraflops. The system's memory hierarchy and interconnect architecture enable it to handle 1,879,040 concurrent threads (Nmax), supporting massive parallel computations. ### System Architecture The computing nodes are interconnected using InfiniBand HDR200 networking technology, providing high-bandwidth, low-latency communication essential for distributed computing workloads. The system runs Red Hat Enterprise Linux 9.2, offering a stable and well-supported operating environment for scientific applications. With 52,096 processor cores distributed across the compute nodes, Snellius Phase 3 GPU can efficiently partition workloads across multiple processors while maintaining high computational density. ### Energy Efficiency Operating at 295.67 kilowatts, the system demonstrates modern supercomputing's focus on energy efficiency. This power consumption level reflects careful system design balancing computational performance with operational costs and environmental considerations, particularly important for facilities running large-scale computations continuously. ### Research Impact As part of the Snellius supercomputing infrastructure, this system serves the Dutch research community, enabling computational work that would be impossible on conventional computing hardware. The GPU acceleration specifically benefits workloads in machine learning, molecular dynamics, computational fluid dynamics, and other fields where massive parallelism provides significant performance advantages over traditional CPU-only computing approaches. ## References 1. [TOP500](https://www.top500.org/system/180316/)