MCPcopy Create free account

hub / github.com/Tencent/hpc-ops / functions

Functions10,542 in github.com/Tencent/hpc-ops

⨍ Functions 10,542◇ Types & classes 8,025↳ Endpoints 13
↓ 2 callersMethodRecvFd
src/communicator/channel.cc:103
↓ 2 callersFunctionReserveAddrMapHandleAndSetAccess
src/communicator/multicast_object_manager.cc:41
↓ 2 callersMethodSendFd
src/communicator/channel.cc:67
↓ 2 callersFunction_e2m1_to_bf16_x2
LUT _e2m1_to_bf16_x2: Direct E2M1->BF16 (converts 2 E2M1 to 2 BF16)
3rd/cutlass/include/cutlass/numeric_conversion.h:3741
↓ 2 callersFunction_e2m1_to_bf16_x8
LUT x8 _e2m1_to_bf16_x8: Direct E2M1->BF16 (converts 8 E2M1 to 8 BF16 in one call)
3rd/cutlass/include/cutlass/numeric_conversion.h:3834
↓ 2 callersFunction_is_scalar_zero
(x)
hpc/sampler.py:136
↓ 2 callersFunction_reduce_max_us_list
(local_us)
benchmark/fuse_allreduce_rmsorm/benchmark_fuse_allreduce_rmsnorm.py:73
↓ 2 callersFunctionabs
3rd/cutlass/include/cutlass/quaternion.h:391
↓ 2 callersFunctionabsolute_value
3rd/cutlass/include/cutlass/fast_math.h:1072
↓ 2 callersFunctionadd_byte_offset_
Adds a pointer offset in units of element
3rd/cutlass/include/cutlass/conv/threadblock/conv3d_dgrad_output_gradient_tile_access_iterator_optimized.h:286
↓ 2 callersFunctionadd_byte_offset_
Adds a pointer offset in units of element
3rd/cutlass/include/cutlass/conv/threadblock/conv3d_fprop_activation_tile_access_iterator_optimized.h:281
↓ 2 callersFunctionadd_byte_offset_
Adds a pointer offset in units of element
3rd/cutlass/include/cutlass/conv/threadblock/conv2d_fprop_activation_tile_access_iterator_optimized.h:264
↓ 2 callersFunctionadd_pointer_offset
Adds a pointer offset in units of Element
3rd/cutlass/include/cutlass/transform/threadblock/regular_tile_access_iterator_pitch_linear_direct_conv.h:155
↓ 2 callersFunctionadd_pointer_offset
Adds a pointer offset in units of Element
3rd/cutlass/include/cutlass/transform/threadblock/regular_tile_access_iterator_tensor_op.h:168
↓ 2 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_triangular_matrix.h:723
↓ 2 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_2dthreadtile.h:704
↓ 2 callersMethodadd_pointer_offset
Adds a pointer offset in units of Element
3rd/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_pitch_linear_2dthreadtile.h:243
↓ 2 callersMethodadd_tile_offset
Adds a tile offset
3rd/cutlass/include/cutlass/transform/threadblock/regular_tile_iterator_pitch_linear_2dthreadtile.h:249
↓ 2 callersFunctionadvance
Advances the iterator along the advance dimension
3rd/cutlass/include/cutlass/gemm/warp/mma_tensor_op_tile_access_iterator.h:242
↓ 2 callersFunctionadvance_smem_read_stage
3rd/cutlass/include/cutlass/gemm/threadblock/mma_multistage.h:91
↓ 2 callersFunctionadvance_smem_write_stage
Advance global memory read-iterators and shared memory write-iterators to the stage
3rd/cutlass/include/cutlass/gemm/threadblock/mma_multistage.h:263
↓ 2 callersFunctionadvance_smem_write_stage
3rd/cutlass/include/cutlass/gemm/threadblock/mma_pipelined.h:92
↓ 2 callersFunctionadvance_to_next_work
3rd/cutlass/include/cutlass/gemm/kernel/sm90_tile_scheduler_stream_k.hpp:324
↓ 2 callersFunctionapply_rotary_pos_emb_neox_reference
(x, cos_sin)
tests/test_rope.py:39
↓ 2 callersFunctionarrive
3rd/cutlass/include/cutlass/arch/barrier.h:221
↓ 2 callersFunctionas_position_independent_swizzle_layout
3rd/cutlass/include/cute/pointer_flagged.hpp:91
↓ 2 callersMethodat
Returns a reference to the element at a given Coord
3rd/cutlass/include/cutlass/tensor_ref_planar_complex.h:286
↓ 2 callersMethodbegin_epilogue
Called at the start of the epilogue just before iterating over accumulator slices
3rd/cutlass/include/cutlass/epilogue/threadblock/fusion/visitor_2x.hpp:63
↓ 2 callersFunctionbuild_env
Construct the worker subprocess environment.
benchmark/fused_moe/benchmark_fuse_moe.py:157
↓ 2 callersFunctionbuild_runner
(provider, x, w, w_high, w_low, scale, split_flag)
benchmark/route_gemm/benchmark_gemm_bf16xfp32.py:123
↓ 2 callersFunctioncalculate_umma_peer_mask
3rd/cutlass/include/cutlass/pipeline/sm100_pipeline.hpp:95
↓ 2 callersFunctionceil_div_int
src/gemm/sm90/entry.cc:19
↓ 2 callersMethodclear
< Efficiently disables all accesses guarded by mask
3rd/cutlass/include/cutlass/epilogue/threadblock/predicated_tile_iterator_affine.h:227
↓ 2 callersMethodclear
< Efficiently disables all accesses guarded by mask
3rd/cutlass/include/cutlass/epilogue/threadblock/predicated_tile_iterator_direct_conv.h:154
↓ 2 callersMethodclear
< Efficiently disables all accesses guarded by mask
3rd/cutlass/include/cutlass/epilogue/threadblock/predicated_tile_iterator_strided_dgrad.h:155
↓ 2 callersFunctionclear_mask
Clears the predicates
3rd/cutlass/include/cutlass/conv/threadblock/conv2d_fprop_filter_tile_access_iterator_optimized.h:243
↓ 2 callersFunctionclear_mask
Clears the predicates
3rd/cutlass/include/cutlass/conv/threadblock/conv2d_fprop_activation_tile_access_iterator_optimized.h:324
↓ 2 callersMethodclear_mask
Clears the predicate set efficiently
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_triangular_matrix.h:752
↓ 2 callersMethodclear_mask
Clears the predicate set efficiently
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_2dthreadtile.h:733
↓ 2 callersFunctioncluster_arrive
3rd/cutlass/include/cute/arch/cluster_sm90.hpp:57
↓ 2 callersFunctioncoalesce_x
3rd/cutlass/include/cute/layout.hpp:824
↓ 2 callersFunctioncompact_order
3rd/cutlass/include/cute/stride.hpp:410
↓ 2 callersFunctioncomposition_impl
3rd/cutlass/include/cute/layout.hpp:1031
↓ 2 callersFunctioncompute_task_map_len
src/group_gemm/cp_async/entry.cc:35
↓ 2 callersMethodconsumer_release
3rd/cutlass/include/cutlass/pipeline/sm100_pipeline.hpp:1164
↓ 2 callersFunctioncontinue_current_work_for_linear_idx
3rd/cutlass/include/cutlass/gemm/kernel/sm90_tile_scheduler_stream_k.hpp:306
↓ 2 callersFunctionconvert_from_float
3rd/cutlass/include/cutlass/exmy_base.h:1004
↓ 2 callersMethodconvert_to_float
3rd/cutlass/include/cutlass/float8.h:1085
↓ 2 callersFunctioncooperative_gemm_no_predication
3rd/cutlass/include/cute/algorithm/cooperative_gemm.hpp:279
↓ 2 callersFunctioncoprofile
3rd/cutlass/include/cute/layout.hpp:634
↓ 2 callersFunctioncopy_tiles_and_advance
3rd/cutlass/include/cutlass/gemm/threadblock/mma_with_reduction_multistage.h:90
↓ 2 callersFunctioncopy_tiles_and_advance
3rd/cutlass/include/cutlass/gemm/threadblock/mma_planar_complex_multistage.h:94
↓ 2 callersFunctioncopy_tiles_and_advance
3rd/cutlass/include/cutlass/gemm/threadblock/mma_sparse_multistage.h:97
↓ 2 callersFunctioncopy_tiles_and_advance
3rd/cutlass/include/cutlass/gemm/threadblock/mma_blas3_multistage.h:94
↓ 2 callersFunctioncutlassGetStatusString
Convert cutlass status to status strings
3rd/cutlass/include/cutlass/cutlass.h:63
↓ 2 callersMethoddata
3rd/cutlass/include/cutlass/gemm/kernel/sm90_tile_scheduler_stream_k.hpp:81
↓ 2 callersFunctiondepth
3rd/cutlass/include/cute/layout.hpp:623
↓ 2 callersFunctiondevice_to_num
(device)
hpc/allreduce.py:187
↓ 2 callersFunctiondomain_distribute
3rd/cutlass/include/cute/layout.hpp:1449
↓ 2 callersMethodenable_mask
Clears the predicate set efficiently
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_triangular_matrix.h:758
↓ 2 callersMethodenable_mask
Clears the predicate set efficiently
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_2dthreadtile.h:739
↓ 2 callersMethodend_epilogue
Called after all steps have been completed
3rd/cutlass/include/cutlass/epilogue/threadblock/fusion/visitor_2x.hpp:121
↓ 2 callersFunctionepilogue_predication
3rd/cutlass/include/cute/algorithm/cooperative_gemm.hpp:58
↓ 2 callersFunctionexchange_sort
3rd/cutlass/include/cute/int_tuple.hpp:586
↓ 2 callersMethodexpect_transaction
Performs an expected transaction bytes increment without doing an arrive operation
3rd/cutlass/include/cutlass/arch/barrier.h:564
↓ 2 callersFunctionfast_acos
3rd/cutlass/include/cutlass/fast_math.h:822
↓ 2 callersFunctionfast_log
3rd/cutlass/include/cutlass/fast_math.h:903
↓ 2 callersMethodfill
3rd/cutlass/include/cute/container/array.hpp:171
↓ 2 callersFunctionfind_log2
3rd/cutlass/include/cutlass/fast_math.h:244
↓ 2 callersFunctionfloat_exmy_base<T, Derived>
Floating point conversion
3rd/cutlass/include/cutlass/exmy_base.h:1023
↓ 2 callersFunctionfmt_us
(value)
benchmark/attention_decode/bench_attention_decode_fp8.py:438
↓ 2 callersFunctionfpclassify
3rd/cutlass/include/cutlass/half.h:531
↓ 2 callersMethodfront
3rd/cutlass/include/cutlass/array.h:385
↓ 2 callersFunctionfused_sampler
2-kernel fused sampler. Pipeline inside the kernel:: repetition_penalty -> temperature -> [softmax1] -> topk -> [softmax2] -> to
hpc/sampler.py:42
↓ 2 callersFunctiongcd
3rd/cutlass/include/cutlass/fast_math.h:167
↓ 2 callersFunctiongcd_cxx11
3rd/cutlass/include/cutlass/fast_math.h:193
↓ 2 callersFunctiongemm
3rd/cutlass/include/cute/algorithm/cooperative_gemm.hpp:519
↓ 2 callersFunctiongemm_iters
Perform the specified number of threadblock mainloop iterations of matrix multiply-accumulate. Assumes prologue has been initiated.
3rd/cutlass/include/cutlass/gemm/threadblock/mma_pipelined.h:283
↓ 2 callersFunctiongenerate_kv_block_indices
(kcache, req_length)
tests/test_rope.py:21
↓ 2 callersMethodgetParams
3rd/cutlass/include/cutlass/conv/threadblock/conv2d_tile_iterator.h:102
↓ 2 callersMethodgetStagePtr
src/allreduce/fuse_allreduce_rmsnorm_low_latency.h:181
↓ 2 callersFunctiongetVersionBuild
3rd/cutlass/include/cutlass/version.h:64
↓ 2 callersMethodget_1d_coord
3rd/cutlass/include/cute/layout.hpp:272
↓ 2 callersMethodget_abs_max
3rd/cutlass/include/cutlass/epilogue/threadblock/epilogue_with_absmax.h:189
↓ 2 callersMethodget_callbacks
3rd/cutlass/include/cutlass/epilogue/threadblock/fusion/visitor_2x.hpp:134
↓ 2 callersFunctionget_current_work_cta_m_n_in_cluster
Given raster order and current work tile linear index, reset cta m and n index in the cluster.
3rd/cutlass/include/cutlass/gemm/kernel/sm90_tile_scheduler_stream_k.hpp:735
↓ 2 callersMethodget_grid_dims
Returns the grid extents in thread blocks to launch
3rd/cutlass/include/cutlass/gemm/kernel/params_universal_base.h:216
↓ 2 callersMethodget_hier_coord
3rd/cutlass/include/cute/layout.hpp:250
↓ 2 callersMethodget_mask
Gets the mask
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_triangular_matrix.h:770
↓ 2 callersMethodget_mask
Gets the mask
3rd/cutlass/include/cutlass/transform/threadblock/predicated_tile_iterator_2dthreadtile.h:751
↓ 2 callersMethodget_ptr_aux_output_abs_max
3rd/cutlass/include/cutlass/epilogue/thread/linear_combination_generic_with_scaling.h:305
↓ 2 callersMethodget_ptr_output_abs_max
3rd/cutlass/include/cutlass/epilogue/thread/linear_combination_generic_with_scaling.h:300
↓ 2 callersMethodget_shape_C
Get C extents. fprop: C extents array contains [N,Z,P,Q,K]. Turn that into ((Q,P,Z,N), (K)) dgrad: C extents array contains [N,D,H,W,C]. Turn that int
3rd/cutlass/include/cutlass/conv/convnd_problem_shape.hpp:457
↓ 2 callersFunctionget_slice
3rd/cutlass/include/cute/atom/copy_atom.hpp:340
↓ 2 callersFunctionget_stride
3rd/cutlass/include/cutlass/transform/threadblock/ell_predicated_tile_access_iterator.h:402
↓ 2 callersFunctionget_strided_dgrad_tile_m
////////////////////////////////////////////////////////////////////////////////////////////
3rd/cutlass/include/cutlass/conv/threadblock/threadblock_swizzle.h:53
↓ 2 callersFunctionget_tma_swizzle_base
3rd/cutlass/include/cute/atom/copy_traits_sm90_tma_swizzle.hpp:108
↓ 2 callersFunctionget_tma_swizzle_bits
3rd/cutlass/include/cute/atom/copy_traits_sm90_tma_swizzle.hpp:70
↓ 2 callersMethodget_work_k_tile_count
3rd/cutlass/include/cutlass/gemm/kernel/sm100_tile_scheduler.hpp:499
↓ 2 callersFunctiongmem_wait
Wait until we have at least one completed global fetch stage
3rd/cutlass/include/cutlass/gemm/threadblock/mma_multistage.h:485
← previousnext →801–900 of 10,542, ranked by callers