NG3PY - Refurbished - Dell NVIDIA L4 24GB Passive GPU
Warranty: Lifetime
Condition - Refurbished
Dell NVIDIA L4 24GB Passive GPU (Dell Part Number: NG3PY) is a high-efficiency, low-profile accelerator designed for AI inference, video processing, virtual desktops, and graphics-intensive workloads. Powered by NVIDIA’s Ada Lovelace architecture, it offers significant performance improvements over its predecessor, the T4, while maintaining a compact and energy-efficient design.
Specifications
-
GPU Architecture: NVIDIA Ada Lovelace
-
CUDA Cores: 7,424
-
Tensor Cores: 232 (4th generation)
-
RT Cores: 58 (3rd generation)
-
GPU Memory: 24 GB GDDR6 with ECC
-
Memory Bandwidth: 300 GB/s
-
Interface: PCIe Gen 4.0 x16
-
Form Factor: Low-profile, single-slot (169 mm x 69 mm)
-
Cooling: Passive (requires adequate system airflow)
-
Power Consumption: 72W (no external power connector required)
-
Display Outputs: None
-
Supported APIs: DirectX 12 Ultimate, Shader Model 6.6, OpenGL 4.6, Vulkan 1.3, CUDA 12.0, OpenCL 3.0
Performance Highlights
-
FP32 (Single Precision): 30.3 TFLOPS
-
TF32 Tensor Performance: 60 TFLOPS (up to 120 TFLOPS with sparsity)
-
FP16 Tensor Performance: 121 TFLOPS (up to 242 TFLOPS with sparsity)
-
FP8 Tensor Performance: 242.5 TFLOPS (up to 485 TFLOPS with sparsity)
-
INT8 Performance: 242.5 TOPS (up to 485 TOPS with sparsity)
Compatibility & Deployment
The L4's low-profile, single-slot design and passive cooling make it ideal for dense server environments. It is compatible with systems that have a PCIe Gen 4.0 x16 slot and sufficient airflow. The absence of external power connectors simplifies integration into existing infrastructures.
Condition - Refurbished
Dell NVIDIA L4 24GB Passive GPU (Dell Part Number: NG3PY) is a high-efficiency, low-profile accelerator designed for AI inference, video processing, virtual desktops, and graphics-intensive workloads. Powered by NVIDIA’s Ada Lovelace architecture, it offers significant performance improvements over its predecessor, the T4, while maintaining a compact and energy-efficient design.
Specifications
-
GPU Architecture: NVIDIA Ada Lovelace
-
CUDA Cores: 7,424
-
Tensor Cores: 232 (4th generation)
-
RT Cores: 58 (3rd generation)
-
GPU Memory: 24 GB GDDR6 with ECC
-
Memory Bandwidth: 300 GB/s
-
Interface: PCIe Gen 4.0 x16
-
Form Factor: Low-profile, single-slot (169 mm x 69 mm)
-
Cooling: Passive (requires adequate system airflow)
-
Power Consumption: 72W (no external power connector required)
-
Display Outputs: None
-
Supported APIs: DirectX 12 Ultimate, Shader Model 6.6, OpenGL 4.6, Vulkan 1.3, CUDA 12.0, OpenCL 3.0
Performance Highlights
-
FP32 (Single Precision): 30.3 TFLOPS
-
TF32 Tensor Performance: 60 TFLOPS (up to 120 TFLOPS with sparsity)
-
FP16 Tensor Performance: 121 TFLOPS (up to 242 TFLOPS with sparsity)
-
FP8 Tensor Performance: 242.5 TFLOPS (up to 485 TFLOPS with sparsity)
-
INT8 Performance: 242.5 TOPS (up to 485 TOPS with sparsity)
Compatibility & Deployment
The L4's low-profile, single-slot design and passive cooling make it ideal for dense server environments. It is compatible with systems that have a PCIe Gen 4.0 x16 slot and sufficient airflow. The absence of external power connectors simplifies integration into existing infrastructures.