Searched refs:thread_id_info (Results 1 – 2 of 2) sorted by relevance
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | ir_emitter_unnested.cc | 4003 KernelSupportLibrary* ksl, const ThreadIdInfo& thread_id_info, in EmitTile() argument
4016 mapping_scheme, thread_id_info.thread_id_x, index_ty, &b_); in EmitTile()
4057 ceil_of_ratio(b_.CreateSub(tile_height, thread_id_info.thread_id_y), in EmitTile()
4061 thread_id_info.thread_id_y, b_.CreateMul(y_indvar, num_threads_y)); in EmitTile()
4390 IrEmitterUnnested::ThreadIdInfo thread_id_info = in EmitEpilogueForReduction() local
4395 llvm::Value* x_loc = thread_id_info.thread_id_x; in EmitEpilogueForReduction()
4396 llvm::Value* y_loc = thread_id_info.thread_id_y; in EmitEpilogueForReduction()
4484 b_.CreateUDiv(thread_id_info.thread_id_x, constant(kWarpSize)); in EmitEpilogueForReduction()
4485 ksl.If("intra_warp_reduce_write", is_zero(thread_id_info.lane_id), [&] { in EmitEpilogueForReduction()
4496 {b_.getInt32(0), constant(j), thread_id_info.lane_id})); in EmitEpilogueForReduction()
[all …]
|
| D | ir_emitter_unnested.h | 142 const ThreadIdInfo& thread_id_info, const llvm_ir::IrArray::Index& index,
551 KernelSupportLibrary* ksl, const ThreadIdInfo& thread_id_info,
|