NVIDIA L40S 48GB GDDR6 PCIe x16 FHFL Accelerator GPU HPE P68953-B21 P67869-001
In-store Pickup
Local Pickup in Dallas, TX Available
- Specifications
- Description
| Manufacturer Part Number | |
| Manufacturer | |
| Condition | |
| Standard Warranty | 90 Day Replacement Warranty |
| Item Category | |
| Model | |
| Form Factor | |
| Interface | |
| External Interfaces | |
| Storage Capacity | |
| Transfer Rate | |
| Network Management Type | |
| Processor Type | |
| Clock Speed | |
| Number of Cores | |
| Cache | |
| Socket Type | |
| TDP | |
| Compatible Port | |
| Number of Ports | |
| CPU Type | |
| HDD Form Factor | |
| HDD Interface | |
| HDD Rotation Speed | |
| HDD Storage Capacity | |
| HDD Transfer Rate | |
| Memory Bus Speed | |
| Memory Capacity Per Module | |
| Memory Features | |
| Memory Form Factor | |
| Memory Kit Brand | |
| Memory Kit Breakdown | |
| Memory Kit Speed | |
| Memory Kit Technology | |
| Memory Kit Total Capacity | |
| Memory Kit Voltage | |
| Memory Number of Modules | |
| Memory Number of Pins | |
| Memory Total Capacity | |
| Memory Type | |
| Network Ports | |
| Processor Model | |
| SSD Capacity | |
| SSD Interface | |
| SSD Storage Capacity | |
| SSD Transfer Rate | |
| SSD Type |
NVIDIA L40S 48GB GDDR6 PCIe x16 FHFL Accelerator GPU HPE P68953-B21 P67869-001
The NVIDIA L40S 48GB PCIe Accelerator (HPE Spare Part: P67869-001) is a high-performance data center GPU built on the NVIDIA Ada Lovelace architecture. It is designed for demanding workloads such as generative AI, Large Language Model (LLM) inference, high-performance computing, and advanced graphics rendering in enterprise environments.
Core Technical Specifications
GPU Model: NVIDIA L40S
Architecture: Ada Lovelace
Memory: 48GB GDDR6
Memory Bandwidth: 864 GB/s
Interface: PCIe Gen4 x16 (forward compatible with newer platforms)
Form Factor: Full-Height, Full-Length (FHFL), dual-slot
Cooling: Passive (requires server-grade airflow)
Power Consumption: ~350W
Specialized Compute Features
Tensor Cores (4th Generation):
Optimized for AI training, inference, and deep learning workloads
RT Cores (3rd Generation):
Accelerates real-time ray tracing and advanced rendering
AI & LLM Optimization:
Built for generative AI, large-scale inference, and data analytics
Key Features
High-Performance AI Acceleration:
Ideal for AI model inference, machine learning, and HPC workloads
Enterprise Graphics Capability:
Supports rendering, simulation, and visualization tasks
Scalable Data Center Deployment:
Designed for multi-GPU configurations in enterprise servers
Compatibility
This GPU is validated for high-performance server platforms with adequate power and cooling:
Supported Server Platforms
Gen12 Systems:
DL320, DL340, DL380, DL380a
Gen11 Systems:
DL320, DL380, DL380a, DL385, DL145, ML350
Specialized & HPC Platforms
Edge Servers: e920d Gen10 Plus, e930t Gen11
Scale-Up Systems: Scale-up Server 3200
HPC Platforms: Cray XD295v, ProLiant XL675d Gen10 Plus
Critical Hardware Requirements
Cooling:
Requires high-performance fan configurations; standard cooling is insufficient
Power:
Recommended 800W–1000W overhead per GPU; 2000W+ PSUs for multi-GPU setups
Expansion Requirements:
Requires PCIe x16 riser kits and auxiliary GPU power cables for proper operation
Operating System Support
Windows: Windows Server 2019, 2022, 2025
Linux: RHEL, SUSE, Ubuntu, Oracle Linux
Virtualization: VMware ESXi 7.x / 8.x (GPU pass-through supported)
Related Part Numbers
P68953-B21
P67869-001
S2L70A
S2L70C
The NVIDIA L40S 48GB GPU delivers cutting-edge AI acceleration, exceptional memory capacity, and enterprise-grade reliability, making it an ideal solution for modern data centers, AI workloads, and high-performance computing environments.