MCPcopy Create free account

hub / github.com/bytedance/ByteTransformer / functions

Functions317 in github.com/bytedance/ByteTransformer

Methodinfer
bytetransformer/include/bert_transformer.h:215
Methodinitial_offset
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/output_tile_thread_map_for_attr_mask.h:222
Methodinitialize
bytetransformer/include/bert_transformer.h:188
Methodinitialize
Initializes GEMM state from arguments.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:149
Methodinitialize
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:728
Functioninput_compress_layernorm_kernel_launcher
bytetransformer/include/layernorm.h:393
Functioninput_compress_layernorm_v2
bytetransformer/include/layernorm.h:198
Functioninput_layernorm_kernel_launcher
bytetransformer/include/layernorm.h:376
Functioninput_layernorm_v2
bytetransformer/include/layernorm.h:162
Functionis_source_needed
bytetransformer/include/cutlass_attention_operators.h:218
Methodis_source_needed
Returns true if source is needed
bytetransformer/include/cutlass_attention_operators.h:621
Functionlayernorm
bytetransformer/include/layernorm.h:25
Functionlayernorm_v2
bytetransformer/include/layernorm.h:87
Functionload
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:307
Methodload
bytetransformer/include/cutlass_attention_operators.h:838
Methodload
bytetransformer/include/cutlass_attention_operators.h:990
Methodload
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:105
Methodload
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:261
Methodload
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:479
Methodload
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:677
Methodload
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:867
Methodload
Loads a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_volta_tensor_op.h:123
Methodload
Loads a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_tensor_op.h:133
Methodload_with_byte_offset
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:233
Methodload_with_byte_offset
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:473
Methodload_with_pointer_offset
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:228
Methodload_with_pointer_offset
Loads a fragment from memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:467
Functionmain
unit_test/bert_transformer_test.cc:33
Methodmake_align
bytetransformer/include/cutlass_attention.h:120
Methodmaximum_active_blocks
Computes the maximum number of active blocks per multiprocessor
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:90
Methodoperand_B_ref
Returns a TensorRef to the B operand
cutlass_contrib/include/cutlass/contrib/gemm/threadblock/mma_base.h:202
Functionoperator()
bytetransformer/include/cutlass_attention_operators.h:289
Functionoperator()
Perform a threadblock-scoped matrix multiply-accumulate
cutlass_contrib/include/cutlass/contrib/gemm/threadblock/mma_multistage.h:291
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:83
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:165
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:201
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:636
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:863
Methodoperator()
bytetransformer/include/cutlass_attention_operators.h:1010
Methodoperator()
bytetransformer/include/gemm_bias_act.h:81
Methodoperator()
bytetransformer/include/gemm_bias_act.h:91
Methodoperator()
bytetransformer/include/gemm_bias_act.h:100
Methodoperator()
bytetransformer/include/gemm_bias_act.h:187
Methodoperator()
cutlass_contrib/include/cutlass/contrib/functional.h:36
Methodoperator()
cutlass_contrib/include/cutlass/contrib/functional.h:46
Methodoperator()
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:74
Methodoperator()
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:109
Methodoperator()
Runs the kernel using initialized state.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:228
Methodoperator()
Executes one GEMM
cutlass_contrib/include/cutlass/contrib/gemm/kernel/gemm_grouped.h:431
Methodoperator()
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/softmax_epilogue.h:107
Functionoperator++
bytetransformer/include/cutlass_attention_operators.h:507
Methodoperator++
bytetransformer/include/cutlass_attention_operators.h:812
Methodoperator++
bytetransformer/include/cutlass_attention_operators.h:978
Methodoperator++
cutlass_contrib/include/cutlass/contrib/args_pack_def.h:97
Methodoperator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:747
Methodoperator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:198
Methodoperator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:436
Methodoperator++
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:634
Functionparallel_prefix
bytetransformer/include/remove_padding.h:37
Functionprint_vec
unit_test/helper.h:88
Methodpush
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/epilogue.h:241
Methodpush
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/epilogue.h:356
Methodrun
Runs the kernel using initialized state.
cutlass_contrib/include/cutlass/contrib/gemm/device/gemm_grouped.h:193
Methodset_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:704
Methodset_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:893
Methodset_iteration_index
Overrides the internal iteration index
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:909
Methodset_k_partition
Functionally required for serial reduction in the epilogue
bytetransformer/include/cutlass_attention_operators.h:627
Methodset_k_partition
Functionally required for serial reduction in the epilogue
bytetransformer/include/gemm_bias_act.h:178
Functionset_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:512
Methodset_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:819
Methodset_kgroup_index
bytetransformer/include/cutlass_attention_operators.h:984
Functionset_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:552
Functionset_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:434
Methodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:766
Methodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_access_iterator.h:955
Methodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:217
Methodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:455
Methodset_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:673
Methodset_mask
< Sets the mask
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:949
Functionsoftmax_et_kernelLauncher
bytetransformer/include/softmax.h:398
Functionsoftmax_kernelLauncher
bytetransformer/include/softmax.h:319
Functionsoftmax_kernel_warp
bytetransformer/include/softmax.h:26
Functionsoftmax_kernel_warp_et
bytetransformer/include/softmax.h:108
Functionsoftmax_kernel_warp_half2
bytetransformer/include/softmax.h:65
Functionsoftmax_kernel_warp_half2_et
bytetransformer/include/softmax.h:148
Functionsoftmax_kernel_warp_half2_register
bytetransformer/include/softmax.h:192
Functionsoftmax_kernel_warp_half2_register_et
bytetransformer/include/softmax.h:245
Functionsoftmax_reduction_kernel
bytetransformer/include/cutlass_attention_operators.h:378
Functionsoftmax_reduction_kernel_launcher
bytetransformer/include/cutlass_attention_operators.h:412
Functionstore
Stores a fragment to memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:368
Methodstore
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:299
Methodstore
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:497
Methodstore
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:695
Methodstore
Stores a fragment to memory
cutlass_contrib/include/cutlass/contrib/epilogue/threadblock/predicated_tile_iterator.h:888
Methodstore
Stores a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_volta_tensor_op.h:161
Methodstore
Stores a fragment from the referenced part of the accumulator tile
cutlass_contrib/include/cutlass/contrib/epilogue/warp/softmax_fragment_iterator_tensor_op.h:167
Methodstore_with_byte_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:273
Methodstore_with_byte_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:491
Methodstore_with_pointer_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:267
Methodstore_with_pointer_offset
Store a fragment to memory
cutlass_contrib/include/cutlass/contrib/transform/threadblock/predicated_tile_iterator.h:485
← previousnext →201–300 of 317, ranked by callers