Leave your address - as soon as the price of the product goes down, you'll know about it right away
Processor AMD EPYC 9965 2.25 GHz 384 MB L3 Socket SP5
Processor AMD EPYC 9965 2.25 GHz 384 MB L3 Socket SP5
In cart {{totalProductsCount}} item(s)
Your cart is empty
Processor AMD EPYC 9965 2.25 GHz 384 MB L3 Socket SP5
Leave your address - as soon as the price of the product goes down, you'll know about it right away
Processor AMD EPYC 9965 2.25 GHz 384 MB L3 Socket SP5
Processor AMD EPYC 9965 2.25 GHz 384 MB L3 Socket SP5
The AMD EPYC 9965 is a 5th‑generation server processor operating at a base frequency of 2.25 GHz with up to 3.7 GHz boost and a large 384 MB L3 cache. Built for data center, cloud and AI workloads, this Socket SP5 CPU features 192 “Zen 5” cores and supports dodeca-channel DDR5 memory with a maximum memory bandwidth of 576 GB/s, making it suitable for high-density virtualization, CPU-only AI inference, and GPU-accelerated systems as a host processor.
The EPYC 9965 delivers high integer and AI inference performance per socket while reducing rack count, power consumption and total cost of ownership compared with previous generations and some competitor platforms. With up to 192 cores and large L3 cache, it provides increased vCPU density for cloud environments and improved per‑server throughput for many AI tasks. Certain configurations are optimized to host multiple GPUs to maximize system-level performance for large models.
Install the EPYC 9965 into a compatible SP5 motherboard following the server vendor’s installation and thermal guidelines. Populate supported DDR5 DIMMs across the dodeca-channel memory channels to achieve optimal memory bandwidth. Configure the system firmware and operating system for large core counts and tune BIOS settings for workload type (virtualization, AI inference, or GPU host). When used as a host CPU for GPU systems, ensure proper PCIe lane allocation and cooling for both CPU and accelerator cards.
For cloud and virtualization, allocate one vCPU per physical core to maximize utilization and density; for AI inference, evaluate CPU-only performance for models up to ~13B parameters before adding GPU resources, and when deploying GPUs use an EPYC host CPU configuration tested for your target model to achieve expected throughput and efficiency.
{{highlightedFeature.featureTitle}}:
{{getProductFeatureValueById(product, highlightedFeature.featureId)}}
| Warehouse | Stock | Delivery to the store | Store Price | |
| WH94 | 3 | 22.06.2026 Best delivery | 8,963.99 € | |
| WH354 | 8 | 01.07.2026 | 10,357.99 € |
|
8,963.99 €
Stock:
3
Delivery Date:
22.06.2026
Best delivery
Warehouse: WH94
|
|
10,357.99 €
Stock:
8
Delivery Date:
01.07.2026
Warehouse: WH354
|
|
9,158.99 €
Stock:
3
Delivery Date:
08.07.2026
Warehouse: WH109
|
|
9,958.49 €
Stock:
8
Delivery Date:
22.06.2026
Best delivery
Warehouse: WH15
|
|
10,083.49 €
Stock:
13
Delivery Date:
22.06.2026
Best delivery
Warehouse: EU9
|