Specifications
Memory Size :
varies by model (e.g., 8 GB)
Maximum Resolution :
7680x4320 (8K)
Supply Ability :
200pcs
Cuda Cores :
varies by model (e.g., 3584 for RTX 3070)
Power Consumption :
varies by model (e.g., 220W)
Price :
$7999-8299 pcs Negotiable
Display Outputs :
HDMI 2.1, DisplayPort 1.4a
Boost Clock Speed :
varies by model (e.g., 1725 MHz)
Ray Tracing Cores :
Yes
MOQ :
1pc
Base Clock Speed :
varies by model (e.g., 1500 MHz)
Memory Interface Width :
varies by model (e.g., 256-bit)
Brand Name :
NVIDIA
Payment Terms :
D/A,D/P
Model Number :
L40 48G
Tensor Cores :
Yes
Directx Support :
DirectX 12 Ultimate
Memory Type :
GDDR6 or GDDR6X
Gpu Architecture :
Ampere
Place of Origin :
USA
Packaging Details :
Carton Packaging
Recommended Psu :
varies by model (e.g., 650W)
Description

NVIDIA L40 48G GPU – 48GB High-Speed GDDR6, Ada Lovelace, for Data Center & Professional Visualization

Redefine Data Center Visual Computing & AI Acceleration
The NVIDIA L40 48G, built on the revolutionary Ada Lovelace architecture, is a flagship data center GPU engineered for professional visualization, 3D rendering, AI workloads, and virtualization. It delivers unprecedented performance and versatility to tackle the most demanding enterprise workflows.

Core Advantages

  • 48GB Ultra-Fast GDDR6 ECC Memory
    Equipped with 48GB high-capacity GDDR6 memory (ECC-enabled) and a 384-bit bus, achieving 864GB/s bandwidth. It effortlessly handles memory-intensive tasks like 3D modeling, simulation, and large-scale AI datasets, supporting multi-user vGPU allocation for team collaboration.
  • Next-Gen Core Architecture
    • 18,176 CUDA Cores: 90.5 TFLOPS FP32 performance for robust parallel computing.
    • 142 3rd-Gen RT Cores: 209 TFLOPS ray tracing performance, doubling real-time ray-tracing speed vs. prior gen, ideal for high-fidelity rendering and virtual production.
    • 568 4th-Gen Tensor Cores: Delivering up to 362 TFLOPS BF16/FP16 and 724 TOPS INT8, accelerating AI training, inference, and generative AI tasks.
  • Versatile Data Center Design
    • PCIe 4.0 x16 Interface: High-speed data transmission for seamless server integration.
    • Dual-Slot Passive Cooling: 300W TDP, optimized for 24/7 data center operation with reliable thermal efficiency.
    • 4×DP 1.4a Ports: Support multi-display setups for professional workstations.
    • vGPU & AV1 Support: Enable virtual workstations (vWS) and 3×NVENC/NVDEC with AV1 encode/decode for media workflowsNVIDIA.

Key Applications

  • Professional Visualization: 3D design, CAD, real-time rendering, and NVIDIA Omniverse™ Enterprise.
  • AI & Data Science: Training/inference for generative AI, computer vision, and data analytics.
  • Virtualization & Cloud: High-performance virtual workstations (vGPU) and cloud gaming.
  • Media & Entertainment: Video streaming, batch rendering, and virtual production.

Summary

The NVIDIA L40 48G is the universal data center accelerator that unifies graphics, compute, and AI capabilities. It empowers enterprises to consolidate workloads, simplify infrastructure, and unlock new levels of productivity across creative, scientific, and virtualization workflows.
Send your message to this supplier
Send Now

GDDR6 L40 NVIDIA GeForce Graphics Card 48G GPU Ada Lovelace For Data Center High Speed

Ask Latest Price
Memory Size :
varies by model (e.g., 8 GB)
Maximum Resolution :
7680x4320 (8K)
Supply Ability :
200pcs
Cuda Cores :
varies by model (e.g., 3584 for RTX 3070)
Power Consumption :
varies by model (e.g., 220W)
Price :
$7999-8299 pcs Negotiable
Contact Supplier
GDDR6 L40 NVIDIA GeForce Graphics Card 48G GPU Ada Lovelace For Data Center High Speed

Beijing Hongkun Jiasheng Technology Co., Ltd.

Verified Supplier
1 Years
beijing, beijing
Since 2018
Business Type :
Distributor/Wholesaler, Exporter, Trading Company, Seller
Certification Level :
Verified Supplier
Contact Supplier
Submit Requirement