Searched refs:mapping_scheme (Results 1 – 3 of 3) sorted by relevance
/external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
D | kernel_mapping_scheme.h | 171 explicit ReductionCodegenInfo(KernelMappingScheme mapping_scheme, in ReductionCodegenInfo() argument 173 : mapping_scheme_(mapping_scheme), in ReductionCodegenInfo() 177 CHECK_EQ(num_partial_results, (mapping_scheme.GetTileSizeX() / in ReductionCodegenInfo() 178 mapping_scheme.GetNumThreadsX())); in ReductionCodegenInfo()
|
D | ir_emitter_unnested.cc | 3941 static llvm::Value* GetStartOffsetX(const KernelMappingScheme& mapping_scheme, in GetStartOffsetX() argument 3948 if (mapping_scheme.GetIndexingOrder() == kStridedIndexingX) { in GetStartOffsetX() 3950 } else if (mapping_scheme.GetIndexingOrder() == kStridedLinearIndexingX) { in GetStartOffsetX() 3951 return b->CreateMul(thread_id_x, constant(mapping_scheme.GetVectorSize())); in GetStartOffsetX() 3953 CHECK_EQ(mapping_scheme.GetIndexingOrder(), kLinearIndexingX); in GetStartOffsetX() 3955 mapping_scheme.GetTileSizeX() / mapping_scheme.GetNumThreadsX(); in GetStartOffsetX() 4001 const KernelMappingScheme& mapping_scheme, in EmitTile() argument 4010 int64 num_threads_x = mapping_scheme.GetNumThreadsX(); in EmitTile() 4011 llvm::Value* num_threads_y = constant(mapping_scheme.GetNumThreadsY()); in EmitTile() 4012 int64 tile_size_x = mapping_scheme.GetTileSizeX(); in EmitTile() [all …]
|
D | ir_emitter_unnested.h | 520 const KernelMappingScheme& mapping_scheme, llvm::Type* index_ty, 549 const KernelMappingScheme& mapping_scheme, 562 const KernelMappingScheme& mapping_scheme, llvm::Value* y_loc, 574 const KernelMappingScheme& mapping_scheme, llvm::Value* y_loc,
|