<p>The introduction of Microsoft Azure HBv5 instances and Google Cloud’s N4D compute instances mark significant advancements in cloud computing performance, leveraging the capabilities of AMD EPYC processors.</p>
<h3>Microsoft Azure HBv5 Instances</h3>
<p>Microsoft Azure’s HBv5 instances, which have reached general availability, are powered by AMD EPYC 9V64H CPUs paired with HBM3 memory, delivering an impressive 6.7 TB/s memory bandwidth. This is a substantial improvement aimed at enhancing high-performance computing (HPC) workloads. The performance evolution across four generations of Azure’s HB-series virtual machines (from HBv2 to HBv5) between 2019 and 2025 demonstrates considerable enhancements in handling memory-intensive applications such as fluid dynamics and finite element analysis. The HBv5 instances support up to 368 virtual CPUs (vCPUs), a notable increase from the 120-144 vCPU limits seen in previous generations.</p>
<h3>Google Cloud N4D Instances</h3>
<p>Similarly, Google Cloud’s N4D compute instances utilize the AMD EPYC 9B45 Turin processors, emphasizing improved cost-performance ratios for general-purpose workloads. This transition from the previous N2D instances with EPYC Milan processors, skipping the Genoa generation, results in remarkable performance gains. Benchmarks highlight improvements with configurations such as n2d-standard-32 to n4d-standard-32, both offering 16 physical cores, 32 threads, and 128GB of RAM.</p>
<h3>Performance and Value Enhancements</h3>
<p>The general availability of N4D virtual machines, powered by the Turin processors, provides up to 3.5 times better price-performance for web-serving workloads and a 50% increase for general computing tasks over the previous N2D generation. The N4D instances can accommodate up to 96 vCPUs, 768 GB of DDR5 memory, and 50 Gbps networking capacity, supporting Custom Machine Types to optimize resource allocation. These instances complement the C4D series, which caters to demanding, latency-sensitive applications, while the N4D series offers flexibility and cost-efficiency for general-purpose tasks.</p>
<p>Early adopters of N4D have reported performance boosts of 40-75% across various applications, including edge proxies, advertising platforms, and robotics simulations. Both Azure HBv5 and Google Cloud’s N4D illustrate the ongoing evolution and enhancement of virtual machine performance in the cloud, driven by cutting-edge AMD EPYC processor technology.</p>


Collections

Microsoft Azure HBv5 instances and Google Cloud N4D instances now leverage AMD EPYC processors to deliver significant performance improvements. Azure's HBv5 offers 6.7 TB/s memory bandwidth with up to 368 vCPUs for HPC workloads, while Google's N4D provides up to 3.5x better price-performance for web serving and 50% improvement for general computing tasks. Both platforms demonstrate substantial gains over previous generations, with N4D early adopters reporting 40-75% performance boosts across various applications.

Google Cloud's N4D Instances Deliver Enhanced Performance with AMD EPYC Turin Processors