NVIDIAAI GPU · PCIe
L20
NVIDIA L20 is a Ada Lovelace data-center accelerator with 48 GB GDDR6 and 0.864 TB/s of memory bandwidth.
Specifications
Dense throughput, no sparsity| Architecture | Ada Lovelace |
|---|---|
| Form factor | PCIe |
| Process | TSMC 4N |
| FP16 / BF16 tensor (dense) | 119.5 TFLOPS |
| FP8 tensor (dense) | — |
| FP4 tensor (dense) | — |
| Memory | 48 GB GDDR6 |
| Memory bandwidth | 0.864 TB/s |
| Power | 275 W |
| Launch | Sep 2023 |
AI index
Tensor × bandwidth100 × √((119.5 ÷ 989.4) × (0.864 ÷ 3.35)) = 17.6
The AI index is the geometric mean of dense 16-bit tensor throughput and memory bandwidth, each relative to the H100 SXM. It is a spec-sheet comparison, not a measured benchmark.