One group wants a packaged, portable microservice for a standard model; another already serves several custom models and wants a dedicated inference server. Which NVIDIA pair matches those jobs?
Select an answer to reveal the explanation.
Short Explanation
One group wants a packaged portable microservice. The other already serves several custom models and wants a dedicated inference server. That pair is NIM, then Triton. Not cuDF/cuML, not Megatron/XGBoost, and not NCCL for both.
Full Explanation
NIM simplifies deploying models as microservices. Triton is the named inference server on the cert-page skill list. RAPIDS, Megatron, XGBoost, and NCCL are the wrong pairing.