H13 GPU System Featured Models
High performance AI/ML and HPC-optimized systems

H13 GPU Systems run on dual socket 4th generation AMD EPYCTM processors and offer double the CPU to GPU throughput with PCI-E 5.0. They can support up to 10 FHFL double-width GPU units including the latest AMD InstinctTM MI200 series and NVDIA H100 GPUs.

Choose Supermicro H13 GPU Systems for the most complex applications and demanding workloads

The 8U 8-GPU system provides unparalleled I/O and thermal capacity supporting the latest AMD Instinct™ MI300X accelerators and NVIDIA HGX H100 GPUs, with 8x 400G networking to enable 1:1 GPU ratio for AI, deep learning at scale.

The 4U and liquid-cooled 2U Quad APU Systems supporting MI300A APUs, which combine CPUs and GPUs, leverages Supermicro’s expertise in multiprocessor system architecture and cooling design finely-tuned to tackle the convergence of AI and HPC

Data Center Optimized for HPC Deployment at Scale
Built for the Future, Available Today
H13 GPU Systems 
H13
AS -4125GS-TNRT2
H13
AS -4125GS-TNRT1
H13
AS -4125GS-TNRT

Featured Product Lines
Services 

Worldwide Global Services

Input/Output  

Up to 10 PCIe 5.0 x16 slots for GPU acceleration.

Drives 

Flexible Storage configuration with 10 hot-swap 2.5” U.2 NVMe drives

Processors 

Dual and Single AMD EPYC™ 9005/9004 Series Processors including CPUs with AMD 3D V-Cache™ technology

Management 

Best-in-class serviceability features with tool-less chassis design

Power

4x 2200W Titanium redundant power supply

Memory

24 x DIMM slots, DDR5-4800 MHz Memory


Form Factor 

4U Rackmount

Product Features

Chemistry Simulation

Research Laboratory/National Laboratory

HPC

Deep Learning

Molecular Dynamics Simulation

Astrophysics Simulation

AI/ML

Optimized for: 

Optimized for AI, Deep Learning, HPC, providing maximum acceleration, flexibility, high-performance and balanced solutions. Supermicro GPU-optimized systems support the latest PCI-E 5.0 accelerators and deliver a multitude of performance gains compared to previous generations. 

The H13 GPU-optimized servers deliver unprecedented acceleration at every scale to power the worlds highest performing data centers for AI, data analytics, and HPC applications.

H13 4U GPU Systems 
Maximum Acceleration for AI/Deep Learning and HPC
H13
AS -8125GS-TNHR
H13
AS -8125GS-TNMR2
Featured Product Lines
Services 

Worldwide Global Services

Input/Output  

8 high-speed 400G networking cards providing direct connectivity to each GPU

Drives 

Up to 16 NVMe and 2 SATA SSD drives

Processors 

Dual AMD EPYC™ 9005/9004 Series Processors including CPUs with AMD 3D V-Cache™ technology

Management 

Best-in-class serviceability features with tool-less chassis design

Power

6x 3000W Titanium redundant power supply

Memory

24 x DIMM slots, DDR5-4800 MHz Memory


Form Factor 

8U Rackmount

Product Features

HPC

Autonomous Driving

Large Language Model (LLM)

Deep Learning

Research Laboratory/ National Lab

Astrophysics Simulation

AI

Optimized for: 

Optimized for AI, LLMs, Deep Learning, and HPC, providing maximum acceleration, flexibility, high-performance and balanced solutions. Supermicro 8U 8-GPU systems support the latest OAM and PCIe 5.0 accelerators to streamline deployment at scale for the largest AI models.

The H13 8U 8-GPU servers deliver unprecedented acceleration at every scale to power the worlds highest performing data centers for AI, Data analytics, and HPC applications.

H13 8U Universal GPU 
Maximum Acceleration for AI/Deep Learning and HPC
H13
AS -2145GH-TNMR
H13
AS -4145GH-TNMR
Featured Product Lines
Services 

Worldwide Global Services

Input/Output  

Dual AIOM supporting 400G networking

Drives 

Flexible storage configuration with 8x 2.5” NVMe drives

(24x SAS/SATA options on 4U)

Processors 

Quad AMD® MI300A Processors

(96 Zen4 cores and 912 GPU cores)

Management 

Best-in-class serviceability features with tool-less chassis design

Power

4x 1600W
Hot-Swappable Titanium Level

Memory

32 HBM3 Stacks (512GB)

Form Factor 

2U Liquid-Cooled

4U Air-Cooled

Product Features

Research Laboratory/ National Lab

CSP (Cloud Solution Provider)

AI/ ML

HPC

Optimized for: 

Targeting accelerated HPC workloads, the 4U air-cooled and 2U liquid-cooled Quad systems integrate 4 AMD Instinct™ MI300A APU accelerators. For flexible customer options, the 2U direct-to-chip liquid-cooled system delivers exceptional TCO with over 51% data center energy cost savings, and 4U air-cooled option providing more storage and IO expansion.

Both systems support dual AIOM and 400G Ethernet to help scale and create a high-density supercomputing cluster with up to 21 2U systems in a 48U rack.

H13 A+ 2U and 4U Quad APU systems
With AMD Instinct™ MI300A Accelerators