Description
PNY NVIDIA Tesla L40S 48GB GDDR6 Graphics Accelerator – PG133G
NVIDIA L40S 48GB AI & HPC Professional GPU – Ada Lovelace Architecture
The NVIDIA L40S is a high-performance professional graphics accelerator built on the advanced Ada Lovelace architecture. Designed for AI training, high-performance computing (HPC), advanced visualization, and data center workloads, the L40S delivers exceptional compute power, ray tracing performance, and large-memory capacity for enterprise environments.
With 48GB of high-speed GDDR6 memory and boost clocks up to 2520 MHz, the L40S is engineered for demanding AI, simulation, and rendering applications.
Built for AI & High-Performance Computing
Optimized for:
AI model training and inference
Scientific computing and simulation
Advanced 3D visualization
Digital twin and real-time rendering
Data center and enterprise GPU workloads
Machine learning and deep learning applications
Key Features
NVIDIA L40S GPU (Ada Lovelace Architecture)
48GB GDDR6 Memory
Up to 2520 MHz Boost Clock
PCI Express 4.0 x16 Interface
384-bit Memory Interface
18176 CUDA Cores
568 Tensor Cores
142 RT Cores
300W Maximum Power Draw
1× 16-Pin Power Connector
Dual-Slot Cooling Design
1× HDMI 2.1
3× DisplayPort 1.4a
Ada Lovelace Architecture Performance
Built on a 5nm process, the L40S leverages NVIDIA’s Ada Lovelace architecture to deliver exceptional efficiency and performance improvements over previous generations.
With thousands of CUDA cores, dedicated Tensor cores for AI acceleration, and RT cores for real-time ray tracing, the L40S enables:
Faster AI model training
High-speed inference processing
Advanced photorealistic rendering
Large-scale engineering simulations
Massive 48GB Memory Capacity
The 48GB GDDR6 memory with a 384-bit interface provides the bandwidth and capacity required for:
Large AI datasets
Complex simulation models
High-resolution rendering workloads
Data-intensive enterprise applications
Enterprise-Ready Design
The L40S is designed for workstation and data center deployment, offering:
PCIe 4.0 x16 connectivity
300W maximum power rating
Professional dual-slot cooling solution
Full-height, dual-slot form factor
Ideal Applications
AI & deep learning development
High-performance computing (HPC)
Scientific research
Engineering simulation
Enterprise visualization
Cloud GPU infrastructure
Technical Specifications
Brand: PNY / NVIDIA
Model: PG133G
Graphics Coprocessor: NVIDIA L40S
Architecture: NVIDIA Ada Lovelace
CUDA Cores: 18176
Tensor Cores: 568
RT Cores: 142
Memory: 48GB GDDR6
Memory Interface: 384-bit
Boost Clock: Up to 2520 MHz
Interface: PCI Express 4.0 x16
Power Consumption: 300W
Power Connector: 1× 16-Pin
Video Outputs: 1× HDMI 2.1, 3× DisplayPort 1.4a
Form Factor: Dual-Slot
Card Length: 267 mm
Date First Available: October 25, 2023
The NVIDIA L40S 48GB is a powerful AI and HPC accelerator delivering next-generation Ada architecture performance, massive memory capacity, and enterprise-class reliability for professional and data center environments.
USD 