CODEWITHSUNDEEP
parallel-processingcudagpu
Huan.Chen
Huan.Chen

Reputation: 1

Whether block scheduling is done by GPU?

Can we assign a block to a specific SM programmatically? Can we get runtime information(number of blocks or warps,a block's or warp's execution time etc.) of SM?

Upvotes: 0

Views: 67

Answers (1)

talonmies
talonmies

Reputation: 72349

  1. No.
  2. No.

Nothing of this sort is exposed by the standard CUDA runtime and driver APIs.

Upvotes: 3

Related Questions