NVIDIA L40S Unparalleled AI and graphics performance for the data center

NVIDIA L40S is a powerful graphics processor based on the Ada Lovelace architecture, which is designed to solve the tasks of artificial intelligence (AI), graphic visualization, media content processing and high-performance computing (HPC) based on DELL PowerEdge R760 and DELL PowerEdge R760xa servers . It is the successor to the NVIDIA L40 and has improved performance, especially aimed at heavy workloads in data centers and cloud environments.

Key features of NVIDIA L40S:

1. Architecture :

   • Ada Lovelace is the latest architecture from NVIDIA, which provides a significant increase in performance compared to previous generations (Ampere). It is optimized for AI, graphics and computing tasks.

2. CUDA kernel :

   • 18,176 CUDA cores are responsible for performing parallel calculations, making this GPU highly efficient for AI inference, graphics rendering, and computational tasks.

3. Tensor kernels :

   • 568 4th generation Tensor cores – These cores accelerate neural networks and machine learning tasks by supporting new FP8 number formats, enabling high performance in low precision tasks.

4. RT-kernel (Ray Tracing Cores) :

   • 142 RT-cores of the 3rd generation - designed to process graphics with real-time ray tracing, which allows for high-quality rendering in applications that require photorealistic graphics.

5. Memory :

   • 48 GB GDDR6 ECC is enough memory for large machine learning models, intensive graphics tasks and calculations.
   • Memory bandwidth: 1.7 TB/s , which provides efficient processing of large volumes of data and models.

6. Energy consumption (TDP) :

   • 350W - The L40S is optimized for high performance with reasonable power consumption, making it suitable for data centers.

7. Interface :

   • PCIe Gen4 is a modern interface for data transfer between the GPU and other system components, which provides fast data exchange for intensive tasks.

8. NVLink :

   • NVLink support for combining multiple GPUs, which allows you to increase the bandwidth of data exchange between several GPUs in a single system.

9. Multi-instance GPU (MIG) :

   • MIG (Multi-Instance GPU) — the ability to distribute one GPU to several logical instances for simultaneous execution of several tasks, which allows better use of resources in cloud environments or data centers.

Purpose of NVIDIA L40S:

1. Artificial intelligence (AI) and machine learning :

   • The NVIDIA L40S is ideal for machine learning and large AI model inference tasks such as language models, computer vision, and generative AI. With the help of Tensor kernels, the GPU accelerates the processes of training and inference of neural networks, especially tasks that require calculations with low accuracy (FP8, FP16).

2. Video and media content processing :

   • The L40S supports hardware acceleration for video encoding and decoding, making it ideal for real-time video processing and streaming. Support for video formats such as H.264 and H.265 (HEVC) allows you to work with large media files and provides high encoding/decoding efficiency.

3. Visualization and rendering :

   • Thanks to the powerful RT cores, the NVIDIA L40S is capable of rendering photorealistic graphics using ray tracing. This makes it an ideal solution for architectural applications, animations, virtual reality (VR) creation and 3D design.

4. Virtualized workstations (VDI) :

   • NVIDIA vGPU support allows you to use the L40S to create virtual workstations that can provide multiple users with simultaneous access to powerful graphics resources for complex graphics and computing tasks.

5. High performance computing (HPC) :

   • The L40S is also used in high-performance computing such as modeling, scientific research and data analysis. Thanks to its high computing power and support for large amounts of memory, this GPU can be used to solve complex scientific problems.

6. Cloud computing and data centers :

   • With low power consumption and the ability to work in distributed environments with NVLink and MIG, the L40S is an excellent choice for use in cloud services and data centers, where it can serve many users or applications simultaneously.

Key benefits of NVIDIA L40S:

• Versatility : Suitable for a wide range of tasks — from artificial intelligence to processing graphics and media content.
• High performance in AI and HPC : With a large number of CUDA and Tensor cores, the L40S is capable of processing large AI models and performing complex computational tasks.
• Low power consumption : Despite its high performance, the L40S is relatively power efficient, making it suitable for large-scale deployment.
• Virtualization capability : MIG technology allows GPU resources to be shared between multiple users or applications for increased efficiency in virtualized environments.

Fields of application of NVIDIA L40S:

• Training and inference of large AI models .
• Media processing and video analytics .
• Photorealistic rendering and visualization (architecture, design, animation).
• Virtualized workstations for graphic and computing tasks .
• Computational problems in the field of HPC and scientific research .

NVIDIA L40S is a powerful tool for enterprises, data centers and cloud platforms, which provides high performance for tasks of artificial intelligence, media processing and visualization.

How can we help?

For more detailed information about the DELL PowerEdge R760 server with DDR5 4800 or the DELL PowerEdge R750 server with DDR4 3200, you can find it on our website SERVER SOLUTIONS , to find out the cost of the server, go to the DELL Server Configurator link .