AI / ML | AI Server

  |  AGS8600

AGS8600

The Edgecore AGS8600 is a high-performance, scalable, GPU-based server suitable for AI/ML (Artificial Intelligence/Machine Learning) and HPC (High Performance Computing) applications. The server is ideal for training large language models, automation, object classification, and recognition use cases. The system is designed with eight AMD Instinct MI325X processors and dual AMD EPYC 9005/Turin series processors. The AMD MI325X processor integrates 256 GB of HBM3 memory.

Each MI325X offers a 16-lane PCIe Gen5 host interface and seven AMD Infinity Fabric™ links for full connectivity between eight GPUs in a ring. AMD Infinity Fabric technology delivers excellent I/O efficiency, scaling, and communication within and between industry-standard accelerator module (OAM) device packages.

Why Choose the AGS8600 ?

AMD Instinct™ MI325X

Audi

AGS8600 Features

Specifications

Form Factor 8U
Compute Node ▪️CPU: AMD EPYC™ 9005/Turin series processor, 2 sockets
▪️EPYC 9575F 64C 3.3GHz 400W SP5
▪️Memory: 24 x 64GB RDIMM 2 RANK 5600
▪️Operating System: Linux
▪️BIOS: AMI BIOS (32 MB SPI Flash ROM)
▪️Front End Interface: 2 x BCM957608-P2200GQF00 Dual-Port 200GbE QSFP112 PCIe Ethernet NIC
▪️Storage: 2 x 1.92TB U.2 2.5 NVME4 1DWPD (non SED)
GPU AMD MI325X 8 GPUs with Infinity Fabric
Input/Output ▪️USB: 2 x USB3.0 ports
▪️VGA: 1 x D-Sub 15-pin port
▪️RJ-45: 1 x GbE port dedicated for IPMI
▪️Buttons: UID button, power button, reset button
▪️LEDs:
UID led
PWR led
Baseboard management controller
Indicator (over temperature warning indicator/fan and PSU fail)
Scale-Out Interface ▪️RDMA (ROCE v2)
▪️8 x LP slots for switch board (PCIe x16)
▪️8 x BCM957608-P1400GDF00 Single-Port 400G QSFP112-DD PCIe Ethernet NIC
Storage Front: 6 x 7.68TB U.2 2.5 NVME4 1DWPD (non SED)
BMC AST2600
TPM TPM2.0 SPI
PSU 3300 W hot-swap RPSU, 80+ Titanium (4+2 redundancy)
Fans 15 hot-swappable fans
Dimensions W x D x H: 448 x 850 x 351 mm (17.63 x 33.46 x 13.82 inch)
Operating Temperature 10° C ~ 35° C
Expansion Slots 8 half-height + 4 full-height PCIe 5.0 x16 slots
Soſtware ROCm: 6.2.4
RCCL: 2.20.5
PyTorch: 2.3, 2.2, 2.1, 2.0, 1.13
TensorFlow: 2.16.1, 2.15.1, 2.14.1
JAX: 0.4.26
ONNX Runtime: 1.17.3

Downloads

AGS8600 Datasheet R01

Datasheet  2.41MB

    Unlock the computing power of the future!

    Our AI servers help you tackle the most complex challenges.

    Related Solutions

    Related Resource

    2025 Product Catalogue