NVIDIA L4 900-2G193-0000-001 - Video Card, 24GB, GDDR6, 192-bit, PCIe 4.0 x16
- — Garantie: ans de garantie
- — Retour et échange jusqu'à 30 jours
- — Meilleur prix
- Processeur graphiqueL4 Tensor Core
- Mémoire carte graphique distincte24 Go
- Mémoire Bus192 bit
- Type d'interfacePCI Express 4.0 x16
- Toutes les caractéristiques
The NVIDIA L4 GPU is a data-center-grade graphics card tailored for AI inference, video processing, and virtualized graphics workloads, featuring the following specifications:
- Architecture: Built on NVIDIA’s Ada Lovelace architecture, which enhances efficiency and performance, the L4 includes third-generation RT (Ray Tracing) cores and fourth-generation Tensor Cores. The architecture supports advanced AI tasks, such as natural language processing (NLP), generative AI, and high-performance graphics processing.
- Memory: Equipped with 24 GB of GDDR6 memory with optional ECC (Error Correction Code) to maintain data integrity. The 192-bit memory bus enables a peak bandwidth of 300 GB/s, allowing high-speed data transfers essential for real-time AI inference and video processing tasks.
- Tensor and RT Cores: The fourth-generation Tensor Cores deliver up to 485 TFLOPS for FP8 operations and 485 TOPS for INT8, making them efficient for deep learning inference and video processing. The third-generation RT cores support real-time ray tracing with Shader Execution Reordering (SER), which optimizes rendering efficiency, enhancing frame generation for high-quality graphics.
- Video Capabilities: This GPU supports AV1 encoding and decoding, offering double NVENC encode engines and four NVDEC decode engines, which provide a 40% improvement in efficiency over H.264 at FHD resolution. Additionally, it supports 4x JPEG decoders, optimizing static image processing and making it suitable for applications in video analytics, virtual conferencing, and broadcasting.
- Virtualization Support: With its 24 GB memory and support for NVIDIA vGPU software, the L4 can manage virtualized GPU workloads. This increases performance for AI, professional visualization, and remote desktop applications, boosting user experience and reducing IT overhead.
- Energy Efficiency: The L4 is designed to enhance energy efficiency, offering up to 99% improved power efficiency compared to traditional CPU infrastructures. It provides a reduced total cost of ownership (TCO), with benefits in cooling, power consumption, and space utilization, helping data centers minimize their carbon footprint.
- Enterprise Features: The L4 is optimized for continuous operation in enterprise data centers. It integrates secure boot and root-of-trust technology, essential for secure data center workflows. Additionally, the L4 GPU integrates with NVIDIA’s AI Enterprise software, a suite of tools and libraries for accelerated AI and data analytics, providing scalability across cloud, edge, and on-premises data centers.
The NVIDIA L4 GPU serves as a universal accelerator, making it suitable for a range of applications in AI, graphics, and video processing for modern data center demands.