Vultr makes AMD Instinct™ MI355X GPU available

Vultr's latest addition to its comprehensive Cloud GPU arsenal is the introduction of the potent AMD Instinct™ MI355X GPUs, providing both Vultr Bare Metal servers and Cloud GPU virtual machines to users. These developments cater to the growing needs for AI training, inference, scientific simulation, and high-performance computing (HPC) workloads, significantly boosting energy efficiency and performance.

The AMD Instinct MI355X GPUs offer a remarkable price-to-performance ratio, empowering new advancements in AI. Constructed on the cutting-edge AMD CDNA™ 4 architecture, the GPUs ensure swift processing times, diminished power usage, and amplified computational throughput. With the inclusion of 288GB of HBM3E memory, they are ideally suited for large datasets and complex models.

An added advantage is the support for FP6 and FP4 precision formats, making processing for AI and HPC workloads more efficient. This combination of high memory capacity and precision formats ensures outstanding performance, requiring fewer GPUs to achieve desired outcomes.

Users can leverage the AMD Instinct MI355X GPUs for unrivalled control using Vultr Bare Metal, accessing the server's entire GPU capacity without virtualisation overhead. Alternatively, the GPUs are available as Vultr Cloud GPU virtual machines, offered in 8-GPU plans delivering near-bare metal performance and seamless deployment in mere seconds.

Additional benefits of these 8-GPU Vultr Cloud GPU plans include:

• Seamless integration with Vultr Cloud Storage, featuring Vultr Block Storage, Vultr File System, and Vultr Object Storage.
• Full access to Vultr Cloud Networking solutions, including redundant networking and Vultr Firewall, alongside multiple Vultr VPCs.
• Redundant, pre-formatted local storage.
• Native access to Vultr snapshots and backups.
• Support for nested virtualisation.
• Enhanced flexibility in software and driver versions.

The new AMD Instinct MI355X GPUs empower high-density, high-throughput deployments with ultra-low latency, seamlessly integrating with Vultr's cloud services. Customers can enjoy Vultr's free GPU Enabled Images containing AMD ROCm™ open software for boosted time-to-value, revolutionising AI, machine learning, and HPC workloads. Execute your provision today with simple deployment through the Vultr control panel or API, accompanied by predictable pricing.