Function tvm::tirx::gpu_thread_filter_op

Contents

Function tvm::tirx::gpu_thread_filter_op#

Defined in File gpu.h

Function Documentation#

const Op &tvm::tirx::gpu_thread_filter_op()#

Mark an opaque predicate that selects one thread on an active-set axis.

Arguments, in order:

args[0]: var, The CUDA-index-bound thread-axis variable.
args[1]: pred, The runtime predicate used as an If condition. Canonical thread predicates can be used directly without this wrapper.