引導式測驗 · 中等 等級

晶片佈局規劃和設計測驗中的人工智慧

透過網表圖、順序宏、代理目標、標準單元佈局、可行性和 HPWL 演算法探索學習的晶片佈局。

問題 1 的 8

What physical decision does chip floorplanning or placement make?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. What physical decision does chip floorplanning or placement make?
    • Only the chip’s retail selling price
    • Only its package color
    • Only the company name on the box
    • Where circuit components occupy locations on the chip canvas
  2. How should the cited paper’s report of competitive floorplans produced in hours be interpreted?
    • As a result for its studied setup, netlists and comparisons
    • As proof that every chip can be fully signed off in the same time
    • As a guarantee of global optimality for all circuits
    • As an architecture-independent law about every design tool
  3. AI如何建構佈局問題?
    • 作為一次性猜測
    • 作為順序決策:放置一個區塊,觀察,然後放置下一個區塊
    • 作為烹飪食譜
    • 作為電子表格排序
  4. 圖神經網路學習的「網表」是什麼?
    • 網路密碼列表
    • 漁網設計
    • 組件圖及其連接方式
    • 網際網路路由表
  5. Which fast objective signals are documented by the cited Circuit Training project?
    • Proxy wirelength, congestion and density
    • A proof of all manufacturing and timing requirements
    • 僅產品的市場價格
    • Only a mandatory social-media popularity score
  6. In Google’s cited 2020 method description, what places standard-cell clusters after the RL macro placements?
    • Only an unchanged alphabetical component sort
    • Only a user-interface screenshot
    • A guarantee that no standard cells remain
    • A force-directed procedure
  7. For pins at (0,0),(4,0),(4,3), what is the half-perimeter wirelength estimate in coordinate units?
    • 14
    • 12
    • 7
    • 4
  8. What does the documented can_place_node check evaluate for a proposed grid location?
    • Whether the entire chip is globally optimal
    • Whether the node can be placed there under relevant location constraints
    • Whether every final routed timing path passes signoff
    • Whether the model’s prediction is a factual-confidence probability