試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 41 / 65
Which measures are effective for lowering costs when performing large-scale inference processing using GPU instances?
①Provision multiple dedicated GPU instances that run at full capacity continuously.
②Perform all inference on a small number of CPU servers.
③Even for workloads with low inference frequency, assign dedicated high-performance GPUs.
④Adopt a pay-as-you-go model by leveraging spot instances, GPU sharing, and serverless training/serving.
⑤Stop using GPUs and always run the model locally.