Introducing the new AMD Milan-X processors, improving upon the existing third-genereation EPYC processors with three times more L3 cache targetted to gain significant performance for High Performance Computing workloads. Along with the Milan-X processor, AMD has also introduced a multichip instinct MI200 GPU. This GPU is the first to support 128 GB of HBM2e memory, and delivers a substantial boost for applications critical to the foundation of science.

The AMD Milan-X Processor

The AMD Milan-X processors are an upgrade to the current 3rd Generation Milan-based processors, through its large 768 MB of L3 cache using AMD 3D V-Cache technology, designed for enterprise workloads that spill out of contemporary-sized caches such as structural analysis or Computational Fluid Dynamics (CFD). These processors are designed to accelerate the product development cycle and boost productivity.

Turn raw data into actionable insights faster

Steady multiyear execution delivering innovation and technology

A modern multi-faceted approach to data center security

Gain fast time to value while maximizing IT investments

Processor Technical Specifications

CPU FeaturesAMD Milan-X
CoresUp to 64 Cores
ThreadsUp to 128 Threads
Base Frequency (GHz)Up to 3.05 GHz
Max Boost Frequency (GHz)Up to 3.80 GHz
TDP (W)Up to 280 W
L3 Cache (MB)768 MB
DDR Channels8
Max. DDR FrequencyUp to 3200
Per-Socket Theoretical Memory Bandwidht (GB/s)Up to 204.8
PCIe Gen 4 Lanes128

The AMD MI200 Instinct GPU

The AMD Instinct MI200 accelerators deliver industry-leading performances for data centers at scale. These accelerators are designed to power discoveries at exascale, enabling researchers to tackle the most pressing challenges. Based on the 2nd Gen AMD CDNA architecture, the AMD MI200 accelerators deliver a quantum leap in HPC and AI performance over competitive data center GPUs today. With the AMD Instinct MI200 accelerators and ROCm 5.0 software ecosystem, innovators can tap the power of the world’s most powerful HPC and AI data center GPUs to accelerate their time to science and discovery.

AMD MI200 Instinct series accelerator workloads


Life Sciences






Reinforcement Learning

Image & Object Detection & Classification

Video Detection & Classification

GPU Technical Specifications

GPU FeaturesMI200
Compute UnitsUp to 220CU
Stream Processors14,080
Peak FP64/FP32 VectorUp to 47.9 TFLOPS
Peak FP64/FP32 MatrixUp to 95.7 TFLOPS
Memory Size128GB HBM2e
Memory Interface8,192 Bits
Memory Clock1.6 GHz
Memory BandwidthUp to 3.2 TB/sec sq.
Form FactorOAM
ThermalPassive & Liquid
Max PowerUp to 560 W TDP
Bus InterfacePCIe Gen 4 Support

Download the battlecard now

P4610 whitepaper



# of Cores:  6

# of Threads:  12

Processor Base Frequency:  3.20 GHz

Max Turbo Frequency:  4.60 GHz

Cache: 12 MB SmartCache

Processor Graphics:  Intel® UHD Graphics 630

Graphics Base Frequency:  350 MHz

Graphics Max Dynamic Frequency:  1.20 GHz

Graphics Video Max Memory:  64 GB

4K Support:  Yes, at 60Hz