Vultr has announced early access availability of the AMD Instinct MI355X GPU, scheduled for delivery in Q3 2024. This latest addition supports demanding artificial intelligence workloads by utilizing AMD’s fourth-generation AMD CDNA architecture, providing significant performance and energy efficiency improvements.
The AMD Instinct MI355X GPUs include 288 GB of HBM3E memory, 8 TB/s of memory bandwidth, and expanded datatype support (FP6 and FP4), suitable for AI training and inference workloads and high-performance computing (HPC) tasks. Vultr notes that facilities leveraging AMD Instinct MI355X GPUs will benefit from direct liquid cooling (DLC) options that enhance thermal management and hardware density at scale, thereby optimizing performance per rack.
Additionally, AMD’s ROCm software platform enhances GPU capabilities, further optimizing AI inference, AI training, and compatibility with relevant AI frameworks. According to AMD, integration of MI355X GPUs and ROCm software provides tailored solutions meeting diverse modern AI and HPC requirements.
“AMD is the trusted AI solutions provider of choice, enabling customers to tackle the most ambitious AI initiatives, from building large-scale AI cloud deployments to accelerating AI-powered scientific discovery,” said Negin Oliver, Corporate Vice President of Business Development, Data Center GPU Business, AMD. “AMD Instinct MI350 series GPUs paired with AMD ROCm software provide the performance, flexibility, and security needed to deliver tailored AI solutions that meet the diverse demands of the modern AI landscape.”
Alongside MI355X, Vultr’s portfolio includes AMD EPYC CPUs (9004 Series, 7003 Series), Instinct MI325X and MI300X GPUs, and the latest AMD EPYC 4005 Series CPUs, providing integrated solutions offering full-stack support for accelerated compute workloads.
Source: Vultr
