HPE
HPE S0K89C NVIDIA L4 24GB server accessory
PCIe Gen5 accelerator with 24 GB memory and tensor cores for AI workloads
- FeaturesPCIe Express Gen5 provides increased bandwidth and improves
- PCIe Express Gen5 increases bandwidth for faster CPU-memory transfers
- Fourth-generation tensor cores accelerate AI workloads dramatically
- Faster GPU memory boosts overall accelerator performance
- Third-generation NVLink doubles direct GPU-to-GPU bandwidth
- NVIDIA confidential computing protects data with hardware-based security
AI Accelerator for Server Workloads
This server accessory is an NVIDIA L4 accelerator with 24 GB of memory, built for HPE systems. It connects through a PCIe Gen5 slot to deliver high throughput for compute-heavy tasks. The card targets datacentre deployments that need dedicated AI processing power.
Suited for Inference and Training Tasks
Teams running machine-learning inference, model training or high-performance rendering will benefit from the fourth-generation tensor cores and faster GPU memory. Buyers who need a simple display output adapter or a general-purpose compute card without AI focus should look at something else.
PCIe Gen5 Interface Determines Compatibility
The accelerator requires a PCIe Gen5 slot to operate at its specified bandwidth. Servers limited to earlier PCIe generations will not provide the full data-transfer speed between CPU memory and the card. Confirm the platform supports this interface before purchase.
Highlights
- PCIe Express Gen5 increases bandwidth for faster CPU-memory transfers
- Fourth-generation tensor cores accelerate AI workloads dramatically
- Faster GPU memory boosts overall accelerator performance
- Third-generation NVLink doubles direct GPU-to-GPU bandwidth
- NVIDIA confidential computing protects data with hardware-based security
Specifications
| Brand | HPE |
|---|---|
| Model | S0K89C |
| Features | PCIe Express Gen5 provides increased bandwidth and improves data-transfer speeds from CPU memory. Fourth-generation tensor cores for dramatic AI speedups. Faster GPU memory to boost performance. Third generation NVLink doubles the GPU-GPU direct bandwidth. Third-generation RT cores for speeding up rendering workloads. NVIDIA confidential computing is a built-in security feature of Hopper-based accelerators. |
| Shipping weight | 0.91 kg |
| Package size | 178 × 127 × 76 mm |
Questions about this item
What slot does the HPE S0K89C need to run the NVIDIA L4 24GB?
The HPE S0K89C installs in a PCIe Gen5 slot to connect the NVIDIA L4 24GB GPU to the server.
Does the HPE S0K89C support NVLink with 24GB GPUs already in the server?
Third-generation NVLink on the HPE S0K89C doubles GPU-GPU direct bandwidth when paired with compatible 24GB GPUs.
How can I confirm the HPE S0K89C tensor cores are active after installation?
The HPE S0K89C shows fourth-generation tensor core activity in NVIDIA monitoring tools once the driver loads.
New to this? Buying server hardware with crypto — what to check, what it costs to pay, and what is never asked for.
Also in this aisle
People compared these
Same shelf, same checkout — eight coins and a 30-minute rate lock.
LaCie STHT8000800 8TB RAID subsystem USB 3.1 Type C RAID 0 1
- USB 3.1 Type C
- 0, 1
SanDisk Professional SDPH34H-024T-NBAAB 24TB RAID enclosure
- Thunderbolt 3 and U…



