World Wide Technology (WWT) and Advanced Micro Devices (AMD) have partnered to deliver cutting-edge high-performance computing and AI solutions powered by AMD Instinct GPU accelerators.

AMD's Instinct MI300 and MI350 Series GPUs, built on the CDNA architecture, offer high-bandwidth memory, advanced matrix core technologies and support for a broad range of AI data formats. These accelerators are optimized for:

• Training large language models (LLMs)
• Running inference at scale
• Powering scientific simulations

WWT and AMD deliver differentiated value through validated architectures, hands-on testing in WWT's ATC and AI Proving Ground, and integrated go-to-market strategies. Customers benefit from lower TCO, seamless adoption via AMD's open ROCm™ software stack and access to a robust ecosystem of pre-integrated solutions. Whether on-premises or in the cloud, WWT helps organizations deploy AMD Instinct GPU accelerators to accelerate insights, optimize operations and scale AI infrastructure with confidence.

AMD Instinct MI350 Series

The AMD Instinct™ MI350 Series GPUs represent a leap forward in generative AI (GenAI) and high-performance computing (HPC). Built on the latest CDNA 3 architecture, the MI350 delivers up to 2.4x performance improvement over its predecessor for transformer-based models, and supports FP8 and BF16 precision formats for optimized AI training and inference. 

Key features: 

• 192GB of HBM3 memory, 
• 5.2TB/s memory bandwidth
• Advanced matrix engines designed to accelerate LLMs, recommendation systems and scientific simulations

WWT helps customers evaluate and deploy MI350-based solutions through hands-on environments in the ATC and AI Proving Ground, workload benchmarking and integration services tailored to vertical-specific use cases. Use case deployments include fine-tuning LLMs for financial services, accelerating drug discovery in life sciences, and scaling climate modeling in research institutions. By combining AMD's silicon innovation with WWT's engineering expertise, organizations can confidently adopt MI350 to meet the growing demands of AI and HPC workloads.