GPUDatacenterReleased —

NVIDIA L2

24 GB GDDR6 ECC

A single-slot, low-profile Ada Lovelace data center GPU that NVIDIA offered to the Chinese market from early 2024 as an export-compliant counterpart to the L4. It carries 24 GB of ECC-protected GDDR6 for video processing and mainstream AI inference.

General

Manufacturer: NVIDIA
Model: L2
Segment: Datacenter
GPU chip: —
Architecture: Ada Lovelace
Release date: —
Price: Search on Amazon

Memory

Memory size: 24 GB
Memory type: GDDR6 ECC
Bus width: —
Bandwidth: 300 GB/s
Memory clock: —

Clocks & silicon

GPU clock: —
Boost clock: —
Shader units: —
Tensor / AI cores: —
RT cores: —
Transistors: —
Technology (nm): 4 nm
TSMC 4N
TDP / TBP: —

Interface & APIs

Bus interface: —
SLI / multi-GPU: —
DirectX: 12 Ultimate (12_2)
OpenGL: 4.6
Vulkan: 1.3
Other APIs: CUDA 8.9, OpenCL 3.0

AI support

Precisions: FP8FP16BF16INT8INT4TF32FP64
AI accelerator: —
Peak AI TOPS: 193 TOPS
Notes: Fourth-generation Tensor Cores. Peak dense throughput: INT8 and FP8 193 TOPS, FP16 and BFLOAT16 96.5 TFLOPS, TF32 48.3 TFLOPS. FP32 is 24.1 TFLOPS, and FP64 is not supported.