Name: AMD 9965 Compute - Small (3 Hosts)
SKU: 10680
Availability: InStock

AMD 9965 Compute – Small (3 Hosts)

AMD 9965 Compute – Small (3 Hosts)
Deploy efficient, high-core compute power with the AMD 9965 Small Cluster—featuring 3 enterprise-grade devices built for virtualized workloads, cloud-native apps, and compute-intensive tasks in compact environments.

Customize your AMD compute setup below to match your performance, memory, and deployment requirements or simply book a call with one of our IT experts.

5 people are viewing this right now

Talk to an expert

Description

AMD 9965 – Turin – CPU-optimized Cluster
Number of cabinets	1	Storage Cabinets	1	Frontend Network Cabinets + Cables + Transceivers	1

Number of nodes/cabinet	3	Storage Nodes	1	TOR swithces/cabinet	2
Number of CPUs/cabinet	6	Storage/Node (TB)	500	Storage aggregate switches/cabinet	2
Number of CPUs/node	2	Storage/cabinet	500	firewalls	2
				Core switches	2
Total number of Nodes	3	Storage switches needed	2	Border leaf switches	2
Total number of CPUs	6	Ports used	8
Total number of cores	1152			Transceivers (estimation)	508

Power requirements/node (per kW)	2	Total power/device (per kW)	2.2
				Total power/switch device (per kW)	0.5
				Total power/firewall device (per kW)	0.3
				Total power for frontend network cabinet (per kW)	2.6

Total power for CPU cabinets (per kW)	8	Total power for Storage cabinets (per kW)	3.2	Cost per cabinet – Cables (USD)	$6,500.00
				estimation	$70,000.00

Total Power for AMD 9965 Cluster (kW)	13.8
				Cost per cabinet – Transceivers – CPU Cabinets (USD)	$100,000.00

Cost per cabinet (USD)	$232,000.00	Cost per cabinet (USD)	$155,000.00	Cost per cabinet (USD)	$125,000.00


Cost for the CPU-only Pod (USD)	$682,000.00

		CAPEX (-minus) Storage	$527,000.00

		CAPEX Storage	$155,000.00
10MW – Cores	11520
10MW – CPUs	60

View More View Less

Additional information

CPU

Vendor AMD
Family Turin
Model 9965
Release Date Oct, 2024
Cores/proc 192
Power draw per proc (per kW) 0.5
Base clock speed (GHz) 2.25
Overclocked speed (GHz) 3.7

Dell
RU's 2
Processors/node 2

HPE (GPU-ready node)
RU's 5
Processor/node 2

Dell (GPU-ready node)
RU's 4
Processor/node 2

Storage

Vendor Pure Storage
Model Flash Array – X (Performance
RU's 3
Storage per device (PB) 1

Switching Layer calculations

TOR/Leaf SWs (2 per cabinet) 2 Ports 48
STOR agg SWs (2 per cabinet) 2 Ports 48

Network uplinks (per node) 4 Net uplinks 12
STOR uplinks (per node) 4 STOR uplinks 12

Total Net uplinks per cabinet 12 Ports used per TOR/Leaf (uplinks from nodes) 6
Total STOR uplinks per cabinet 12 Ports used per STOR agg (uplinks from nodes) 6

Ports used per TOR/Lead (nodes + core) 8
Ports used per STOR agg (nodes + core) 8

Total CORE/SPINE ports used 8 Total available CORE/SPINE ports 96
Total CORE/SPINE STOR ports used 4 Total available CORE/SPINE STOR ports 96

Total BORLF ports used 4