AMD said Google Cloud's new C4D and H4D virtual machines are powered by AMD's latest 5th Gen AMD EPYC processors. These new instances are set to change cloud computing by delivering unparalleled performance, scalability, and efficiency for a wide range of demanding workloads, from data analytics and web serving to high-performance computing and artificial intelligence.
The C4D instances are designed to provide impressive performance, efficiency, and consistency for general-purpose computing workloads and AI inference. Leveraging the advancements of the AMD Zen 5 architecture, C4D instances deliver up to 80% higher throughput per vCPU compared to previous generations. These instances feature a higher core frequency of up to 4.1 GHz and are available in various industry-standard shapes, with sizes ranging from 2 vCPU to 384 vCPU and up to 3TB of DDR5 memory. Additionally, C4D instances support confidential computing, ensuring data security and privacy for sensitive workloads.
On the other hand, the H4D instances are optimized for HPC workloads, featuring AMD EPYC CPUs with Cloud RDMA for efficient scaling of up to tens of thousands of cores. These instances are purpose-built for engineering and scientific applications, offering best-in-class compute capabilities for demanding tasks such as large-scale simulations and complex data analysis. The H4D instances also benefit from the high memory bandwidth and low latency provided by the AMD EPYC architecture, making them ideal for memory-intensive applications.
“Since our launch, 5th Gen AMD EPYC solutions have been widely adopted across our OEM partners, enterprise customers, and now we're excited to bring it to the cloud,” said Dan McNamara, senior vice president and general manager, Server Business, AMD. “Our deep technology partnership with Google Cloud enabled them to rapidly adopt the latest AMD EPYC processors to deliver consistent high performance and cost-efficient instances for their most demanding customers
Both AMD and Intel have focused on improving performance and efficiency in their latest architectures. AMD's Zen 5 architecture offers high throughput and core frequency, while Intel's Lunar Lake uses a hybrid approach with P-Cores and E-Cores to balance performance and power efficiency. Both architectures use a modular chiplet design, allowing for greater flexibility and scalability. AMD's Infinity Architecture and Intel's tile-based design both aim to optimize communication and data transfer within the processor. In terms of memory and I/O, both support DDR5 memory and PCIe 5.0, providing high bandwidth and low latency. Intel adds Thunderbolt 4 and Wi-Fi 7 for enhanced connectivity. When it comes to AI and graphics, Intel's Lunar Lake includes dedicated AI processing (NPU 4) and improved integrated graphics (Xe2-LPG), while AMD focuses on AVX-512 for compute-intensive tasks and Cloud RDMA for HPC workloads. Both AMD and Intel emphasize data security and privacy, making their processors suitable for sensitive workloads. In summary, both AMD's 5th Gen EPYC and Intel's Lunar Lake architectures offer significant advancements in performance, efficiency, and scalability. The choice between them may depend on specific workload requirements and preferences for certain features like AI processing or integrated graphics.
