Glossary3 sections1 min read
What is Ada Lovelace?
Ada Lovelace is the NVIDIA architecture used by the L40S and L4 data center GPUs, RTX Ada workstation cards and the GeForce RTX 40 series. It combines FP8 Tensor Cores with ray-tracing cores.
Why it matters when renting a GPU
Ada GPUs suit inference, image and video generation and 3D rendering, often at a lower hourly price than Hopper.
They use GDDR6 or GDDR6X memory and PCIe, so bandwidth and multi-GPU links are lower than on SXM data center parts.
Worked example
EverXCloud describes 21 Ada Lovelace models. The L40S has 48 GB of GDDR6 at 864 GB/s and 733 dense FP8 TFLOPS.
Frequently asked questions
- Is the L40S good for LLMs?
- For inference of models that fit in 48 GB, especially with FP8 or quantization, yes. Large-scale training favors Hopper or Blackwell.