가이드 연계 퀴즈 · 중간 레벨

AI 퀴즈를 위한 GPU 대 TPU

하드웨어를 선택하기 전에 팀이 필요로 하는 GPU 및 TPU 아키텍처, 소프트웨어 지원, 시스템 제약 조건, 워크로드별 증거를 비교하세요.

관련 가이드 경로AI를 위한 GPU 대 Tpu
질문 1 의 8

Which frameworks does the guide list as supported for Cloud TPU workloads?

모든 질문 미리보기8 질문

퀴즈를 풀면 정답과 설명이 나타납니다.

  1. Which frameworks does the guide list as supported for Cloud TPU workloads?
    • JAX, TensorFlow, and PyTorch/XLA
    • CUDA C++ only
    • PyTorch CPU and ONNX Runtime only
    • TensorFlow Lite only
  2. TPU는 누가 디자인하나요?
    • NVIDIA
    • Google
    • 인텔
    • Apple
  3. 행렬 곱셈에서 TPU를 효율적으로 만드는 핵심 하드웨어 구조는 무엇입니까?
    • 회전하는 자기 디스크
    • 대형 기계식 릴레이 뱅크
    • 곱셈-누산 단위로 구성된 수축기 배열
    • 단일 초고속 코어
  4. Which system factors can affect a GPU-versus-TPU performance comparison?
    • Only the accelerator’s peak operation count
    • Only the number of parameters in the model
    • Only the cloud region name
    • Model support, software, interconnects, memory, utilization, and deployment cost
  5. What is a practical strength of GPUs relative to TPUs for many teams?
    • A broad ecosystem of CUDA libraries, tools, and existing GPU-supported workloads
    • A guarantee that every model is faster on a GPU
    • A requirement that all training use TensorFlow
    • A built-in advantage in every cost comparison
  6. Why does the guide recommend benchmarking the intended workload on the specific platform?
    • Peak operation counts always predict end-to-end speed exactly
    • Measured results depend on workload shape, kernels, input pipeline, memory, and system scaling
    • All workloads use the same hardware path and precision
    • A benchmark removes the need to check software support
  7. What should a team compare before choosing between a GPU and a TPU for training?
    • Peak FLOPS alone for an unrelated precision
    • The chip’s name and product launch date
    • End-to-end time and cost for its model, framework, input pipeline, and target scale
    • A benchmark from a different model and software stack
  8. How does Google describe a TPU pod?
    • A single GPU with no network
    • A CPU-only cluster
    • A model format
    • A group of TPU chips connected through a high-speed interconnect