AMD Instinct MI300X

AMD Instinct MI300X

Accelerate your generative AI workloads with the AMD Instinct MI300X

AMD Instinct MI300X

AMD designed the MI300X for superior artificial intelligence (AI) and high-performance computing (HPC) performance, featuring up to 192 GB of HBM3 memory with 5.2 TB/s bandwidth. It excels in handling large language models (LLMs) and advanced AI workloads, offering unmatched efficiency and scalability for data centers, making it the ideal choice for next-gen AI solutions.

Key Features

  • Designed to accelerate modern workloads - Push high-performance computing (HPC) applications to new heights
  • Powered by 4th gen Infinity architecture - AMD Infinity Fabric technology delivers excellent I/O efficiency, scaling, and communication within and between industry-standard accelerator module (OAM) device packages
  • Multi-chip architecture - The MI300X uses state-of-the-art die stacking and chiplet technology in a multi-chip architecture

Specifications

Form Factor OAM module
Lithography 5nm FinFET
Active Interposer Dies (AIDs) 6nm FinFET
GPU Compute Units 1216
Matrix Cores 1216
Stream Processors 19,456
Peak Engine Clock 2100 MHz
Memory Capacity Up to 192 GB HBM3
Memory Bandwidth 5.3 TB/s max. peak theoretical
Memory Interace 8192 bits
AMD Infinity Cache™ (last level) 256 MB
Memory Clock Up to 5.2 GT/s
Scale-up Infinity Fabric™ Links 1 PCIe® Gen 5 x16 (128 GB/s)
I/O to host CPU 1 PCIe® Gen 5 x16 (128 GB/s)
Scale-out network bandwidth PCIe Gen 5 x16 (128 GB/s)
RAS features Full-chip ECC memory, page retirement, page avoidance
Maximum TBP 750W
For full product specifications and parts list - download the AMD Instinct MI300X data sheet.