Skip to main content

NVIDIA L4 (Ada Lovelace)

Overview

The NVIDIA L4 is a low-power GPU designed for cloud AI inference, released in 2023 and based on the Ada Lovelace architecture. With just 72W TDP, it supports passive cooling, delivers 485 TFLOPS FP8 compute, and offers 3× the inference performance of the T4. The L4 supports AV1 hardware encode/decode, making it suitable for cloud gaming, real-time video AI, and generative AI inference.

Core Specifications

ItemSpec
ArchitectureAda Lovelace AD104
ProcessTSMC 4N
Memory24 GB GDDR6
Memory Bandwidth300 GB/s
CUDA Cores7,680
Tensor Cores240 (4th Gen)
FP3230.3 TFLOPS
FP8 Tensor Core485 TFLOPS
INT8 Tensor Core485 TOPS
TDP72 W (passive cooling capable)
PCIeGen 4 x16
Video Codec2× NVENC (AV1) + 4× NVDEC (AV1)

T4 vs L4 Performance Comparison

MetricT4 (Turing)L4 (Ada)Improvement
FP16 Compute65 TFLOPS121 TFLOPS1.9×
INT8 Compute130 TOPS242 TOPS1.9×
FP8 ComputeN/A485 TFLOPSNew
Memory16 GB24 GB+50%
Bandwidth320 GB/s300 GB/sSlightly lower
TDP70 W72 WSame
Video Codec1× NVENC2× NVENC (AV1)+100%

Vendor Information

ItemDetail
ManufacturerNVIDIA Corporation
Websitehttps://www.nvidia.com
Product Pagehttps://www.nvidia.com/en-us/data-center/l4/
ReleaseMarch 2023 GTC

Use Cases

  • Cloud AI inference (generative AI, LLM)
  • Cloud gaming (high-density deployment)
  • Video AI (transcoding, content moderation, AV1 streaming)
  • Edge AI inference