Guide-yakabatana mibvunzo · Zvakaoma Nhanho

Kukurukurirana pamwe chete uye NCCL Quiz

Edzai zvese-kudzikisa, zvese-kuunganidza, kutepfenyura uye kuderedza-kuparadzira mabasa, NCCL identity, network offloads, uye diki sum-kuderedza muenzaniso.

Nzira dzekutungamira dzinoenderanaCollective Kukurukurirana Uye Nccl
Mubvunzo 1 ye 8

Chii chinoitwa nekuita kwekudzikisa-sese mukudzidziswa kwakagoverwa?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinoitwa nekuita kwekudzikisa-sese mukudzidziswa kwakagoverwa?
    • Iyo inosanganisa kukosha kune ese maGPU uye inopa yega GPU mhedzisiro yakasanganiswa
    • Iyo inobvisa yakawandisa data kubva kune imwe GPU
    • Iyo inoderedza iyo GPU wachi kumhanya kuchengetedza simba
    • Inotumira data kubva kumaGPU ese kune imwechete yekuchengetedza dhisiki
  2. Chii chinonzi acronym NCCL?
    • Neural Compute Cluster Layer
    • Network Cache Control Logic
    • NVIDIA Collective Communications Library
    • Numeric Convolution Compute Library
  3. Why can chunked ring communication be useful for a large tensor?
    • It guarantees constant latency for any number of devices
    • Chunks can be streamed across participating links instead of waiting for one monolithic transfer
    • It makes topology irrelevant
    • It prevents all network contention
  4. Ndeipi kushanda pamwe chete kunounganidza chidimbu cheGPU chega chega mukopi yakazara inobatwa nemaGPU ese?
    • Broadcast
    • Deredza-kuparadzira
    • Chipingamupinyi
    • Zvose-unganidza
  5. What does a supported network reduction offload such as SHARP do?
    • Guarantee exactly half the traffic for every collective
    • Remove the need for participating ranks
    • Perform some reduction work in compatible network hardware
    • Turn all-gather into a local memory copy
  6. Ndeipi yekubatanidza iyo NCCL inowanzo shandisa kutaurirana PAKATI pemaseva (node)?
    • InfiniBand kana RoCE
    • A USB tambo
    • Iyo PCIe slot chete
    • Wi-Fi
  7. Three ranks contribute scalar values 2, 3 and 5 to a sum all-reduce. What value does each receive?
    • Only its own original value
    • 3 for the middle rank and nothing elsewhere
    • The average 3.33 unless requested separately
    • 10
  8. Which collective combines contributions but distributes reduced pieces instead of the entire reduced buffer to every rank?
    • Zvose-unganidza
    • Deredza-kuparadzira
    • Broadcast
    • A standalone model checkpoint