AMD Instinct MI300 Accelerators Now Available

AMD Instinct MI300 Accelerators Now Available

AMD announced the availability of the AMD Instinct™ MI300X accelerators – with industry leading memory bandwidth for generative AI and leadership performance for large language model (LLM) training and inferencing – as well as the AMD Instinct™ MI300A accelerated processing unit (APU) – combining the latest AMD CDNA™ 3 architecture and “Zen 4” CPUs to deliver breakthrough performance for HPC and AI workloads. 

“AMD Instinct MI300 Series accelerators are designed with our most advanced technologies, delivering leadership performance, and will be in large scale cloud and enterprise deployments,” said Victor Peng, president, AMD. “By leveraging our leadership hardware, software and open ecosystem approach, cloud providers, OEMs and ODMs are bringing to market technologies that empower enterprises to adopt and deploy AI-powered solutions.” 

Customers leveraging the latest AMD Instinct accelerator portfolio include Microsoft, which recently announced the new Azure ND MI300x v5 Virtual Machine (VM) series, optimized for AI workloads and powered by AMD Instinct MI300X accelerators. Additionally, El Capitan – a supercomputer powered by AMD Instinct MI300A APUs and housed at Lawrence Livermore National Laboratory – is expected to be the second exascale-class supercomputer powered by AMD and expected to deliver more than two exaflops of double precision performance when  fully deployed. Oracle Cloud Infrastructure plans to add AMD Instinct MI300X-based bare metal instances to the company’s high-performance accelerated computing instances for AI. MI300X based instances are planned to support OCI Supercluster with ultrafast RDMA networking.

Several major OEMs also showcased accelerated computing systems, in tandem with the AMD Advancing AI event. Dell showcased the Dell PowerEdge XE9680 server featuring eight AMD Instinct MI300 Series accelerators and the new Dell Validated Design for Generative AI  with AMD ROCm-powered AI frameworks. HPE recently announced the HPE Cray Supercomputing EX255a, the first supercomputing accelerator blade powered by AMD Instinct MI300A APUs, which will become available in early 2024. Lenovo announced its design support for the new AMD Instinct MI300 Series accelerators with planned availability in the first half of 2024. Supermicro announced new additions to its H13 generation of accelerated servers powered by 4th Gen AMD EPYC™ CPUs and AMD Instinct MI300 Series accelerators. 

AMD Instinct MI300X 

AMD Instinct MI300X accelerators are powered by the new AMD CDNA 3 architecture.  When compared to previous generation AMD Instinct MI250X accelerators, MI300X delivers nearly 40% more compute units, 1.5x more memory capacity, 1.7x more peak theoretical memory bandwidth3 as well as support for new math formats such as FP8 and sparsity; all geared towards AI and HPC workloads.  

Today’s LLMs continue to increase in size and complexity, requiring massive amounts of memory and compute. AMD Instinct MI300X accelerators feature a best-in-class 192 GB of HBM3 memory capacity as well as 5.3 TB/s peak memory bandwidth2to deliver the performance needed for increasingly demanding AI workloads. The AMD Instinct Platform is a leadership generative AI platform built on an industry standard OCP design with eight MI300X accelerators to offer an industry leading 1.5TB of HBM3 memory capacity. The AMD Instinct Platform’s industry standard design allows OEM partners to design-in MI300X accelerators into existing AI offerings and simplify deployment and accelerate adoption of AMD Instinct accelerator-based servers. 

Compared to the Nvidia H100 HGX, the AMD Instinct Platform can offer a throughput increase of up to 1.6x when running inference on LLMs like BLOOM 176B4 and is the only option on the market capable of running inference for a 70B parameter model, like Llama2, on a single MI300X accelerator; simplifying enterprise-class LLM deployments and enabling outstanding TCO. 

AMD Instinct MI300A

The AMD Instinct MI300A APUs, the world’s first data center APU for HPC and AI, leverage 3D packaging and the 4th Gen AMD Infinity Architecture to deliver leadership performance on  critical workloads sitting at the convergence of HPC and AI. MI300A APUs combine high performance AMD CDNA 3 GPU cores, the latest AMD “Zen 4” x86-based CPU cores and 128GB of next-generation HBM3 memory, to deliver ~1.9x the performance-per-watt on FP32 HPC and AI workloads, compared to previous gen AMD Instinct MI250X5.  

Energy efficiency is of utmost importance for the HPC and AI communities, however these workloads are extremely data- and resource-intensive. AMD Instinct MI300A APUs benefit from integrating CPU and GPU cores on a single package delivering a highly efficient platform while also providing the compute performance to accelerate training the latest AI models. AMD is setting the pace of innovation in energy efficiency with the company’s 30x25 goal, aiming to deliver a 30x energy efficiency improvement in server processors and accelerators for AI training and HPC from 2020-20256.  

The APU advantage means that AMD Instinct MI300A APUs feature unified memory and  cache resources giving customers an easily programmable GPU platform, highly performant  compute, fast AI training and impressive energy efficiency to power the most demanding HPC  and AI workloads.  

ROCm Software and Ecosystem Partners 

AMD announced the latest AMD ROCm™ 6 open software platform as well as the company’s commitment to contribute state-of-the-art libraries to the open-source community, furthering the company’s vision of open-source AI software development. ROCm 6 software represents a significant leap forward for AMD software tools, increasing AI acceleration performance by ~8x when running on MI300 Series accelerators in Llama 2 text generation compared to previous generation hardware and software. Additionally, ROCm 6 adds support for several new key features for generative AI including FlashAttention, HIPGraph and vLLM, among others. As such, AMD is uniquely positioned to leverage the most broadly used open source AI software models, algorithms and frameworks – such as Hugging Face, PyTorch, TensorFlow and others – driving innovation, simplifying the deployment of AMD AI solutions and unlocking the true potential of generative AI.  

AMD also continues to invest in software capabilities through the acquisitions of Nod.AI and Mipsology as well as through strategic ecosystem partnerships such as Lamini – running LLMs for enterprise customers – and MosaicML – leveraging AMD ROCm to enable LLM training on AMD Instinct accelerators with zero code changes.