MCPcopy Create free account

hub / github.com/NVIDIA/cutlass / functions

Functions19,553 in github.com/NVIDIA/cutlass

↓ 1,758 callersFunctionprint
examples/77_blackwell_fmha/common/pow_2.hpp:81
↓ 1,617 callersFunctionmake_tensor
include/cute/tensor_impl.hpp:396
↓ 1,287 callersMethodhost_view
Accesses the tensor reference pointing to data
tools/util/include/cutlass/util/host_tensor.h:314
↓ 1,015 callersMethodappend
Inserts the operation. operation_kind -> configuration_name -> []
python/cutlass_library/manifest.py:701
↓ 990 callersMethodsize
examples/77_blackwell_fmha/reference/reference_abs_error.hpp:76
↓ 964 callersMethodk
(self)
python/cutlass_cppgen/shape.py:95
↓ 776 callersMethodat
Accesses an element by coordinate
include/cutlass/matrix.h:148
↓ 747 callersFunctionsize
include/cute/layout.hpp:605
↓ 724 callersFunctionmake_layout
include/cute/layout.hpp:333
↓ 722 callersFunctioncopy
include/cute/algorithm/copy.hpp:171
↓ 716 callersMethodn
(self)
python/cutlass_cppgen/shape.py:91
↓ 696 callersMethodmake_layout
examples/common/gather_tensor.hpp:130
↓ 632 callersMethodget_slice
include/cute/atom/mma_atom.hpp:357
↓ 625 callersMethodm
(self)
python/cutlass_cppgen/shape.py:87
↓ 596 callersFunctionrank
include/cute/layout.hpp:614
↓ 591 callersFunctionmake_smem_ptr
include/cute/pointer.hpp:165
↓ 573 callersFunctionshape
include/cute/int_tuple.hpp:102
↓ 528 callersMethodpartition_S
include/cute/atom/copy_atom.hpp:366
↓ 525 callersMethoddata
include/cute/tensor_impl.hpp:180
↓ 520 callersMethoddata
include/cutlass/gemm/kernel/sm90_tile_scheduler.hpp:64
↓ 517 callersMethodcopy
examples/88_hopper_fmha/kernel/fmha_kernel_bwd_convert.hpp:103
↓ 509 callersMethodcolumn
include/cutlass/matrix.h:2094
↓ 508 callersFunctionmake_layout
(*layouts)
python/pycute/layout.py:116
↓ 500 callersMethodpartition_D
include/cute/atom/copy_atom.hpp:376
↓ 475 callersMethodrow
include/cutlass/matrix.h:229
↓ 441 callersMethodat
Gets the value pointed to
tools/profiler/src/problem_space.cpp:129
↓ 439 callersMethoddata
tools/library/include/cutlass/library/util.h:266
↓ 439 callersFunctionprintf
Print one or more values with optional formatting. This function provides printf-style formatted printing capabilities. It can print values
python/CuTeDSL/cutlass/cute/core.py:2040
↓ 402 callersMethodsync_device
Copies data from host to device
tools/util/include/cutlass/util/host_tensor.h:403
↓ 398 callersMethodto
Convert the ratio to another type. :param dtype: The target type for conversion :type dtype: type :return: The ratio converte
python/CuTeDSL/cutlass/cute/core.py:688
↓ 389 callersMethodadvance
Move to the next pipeline stage.
python/CuTeDSL/cutlass/pipeline/sm90.py:1017
↓ 347 callersMethodconsumer_wait
examples/112_blackwell_ssd/utils/pipeline.h:167
↓ 344 callersMethodconsumer_release
( self, state: PipelineState, *, loc: Optional[ir.Location] = None, ip
python/CuTeDSL/cutlass/pipeline/sm90.py:282
↓ 339 callersFunctionlocal_tile
include/cute/tensor_impl.hpp:1035
↓ 333 callersMethodstride
Returns the stride of the layout
include/cutlass/layout/matrix.h:119
↓ 328 callersMethodbegin
Before entry of the subtile load loop
include/cutlass/epilogue/fusion/sm90_visitor_tma_warpspecialized.hpp:132
↓ 320 callersMethodindex
include/cutlass/pipeline/sm90_pipeline.hpp:188
↓ 320 callersFunctionsynclog_emit_wgmma_reg_smem
include/cutlass/arch/synclog.hpp:898
↓ 320 callersFunctionsynclog_emit_wgmma_smem_smem
include/cutlass/arch/synclog.hpp:916
↓ 311 callersFunctionsize
(layout)
python/pycute/layout.py:125
↓ 306 callersMethodproducer_acquire
examples/112_blackwell_ssd/utils/pipeline.h:136
↓ 297 callersFunctionelect_one_sync
Elect one thread in the warp. The elected thread gets its predicate set to true, all others obtain false.
include/cute/arch/cluster_sm90.hpp:180
↓ 292 callersMethodwith
include/cute/atom/mma_atom.hpp:76
↓ 282 callersFunctionenum_auto
()
python/cutlass_library/library.py:49
↓ 276 callersMethoddevice_data
Gets pointer to device data
tools/util/include/cutlass/util/host_tensor.h:286
↓ 269 callersMethoddevice_ref
Accesses the tensor reference pointing to data
tools/util/include/cutlass/util/host_tensor.h:304
↓ 269 callersMethodresize
Changes the size of the logical tensor. Only allocates memory if new capacity exceeds reserved capacity. To force allocation, call reset().
tools/util/include/cutlass/util/host_tensor.h:228
↓ 264 callersFunctionmin
include/cute/int_tuple.hpp:154
↓ 263 callersMethodmake_pipeline_state
include/cutlass/pipeline/sm90_pipeline.hpp:247
↓ 263 callersFunctionmax
include/cute/int_tuple.hpp:134
↓ 248 callersMethodimag
Set the imaginary part of the complex number
include/cutlass/complex.h:315
↓ 247 callersFunctionto_underlying_arguments
include/cutlass/gemm/kernel/sm100_tile_scheduler_stream_k.hpp:114
↓ 245 callersMethoddata_ptr
Get the data pointer from the tensor value. Extracts the data pointer (first field) from the LLVM struct value. :param loc: Optional
examples/python/CuTeDSL/dsl_tutorials/ffi/jit_argument.py:77
↓ 243 callersMethodsize_in_bytes
Returns the size of the union in bytes. :return: The size of the union.
python/CuTeDSL/cutlass/cute/core.py:6134
↓ 241 callersFunctionat
()
docs/jquery.js:23
↓ 239 callersFunctionprint
include/cute/layout.hpp:1918
↓ 237 callersFunctionElement
Unpacks an element from memory
include/cutlass/subbyte_reference.h:187
↓ 235 callersFunctionfrom_dlpack
Convert from tensor object supporting __dlpack__() to a CuTe Tensor. :param tensor_dlpack: Tensor object that supports the DLPack protocol :t
python/CuTeDSL/cutlass/cute/runtime.py:749
↓ 234 callersMethodreal
Set the real part of the complex number
include/cutlass/complex.h:311
↓ 230 callersMethodindex
(self)
python/CuTeDSL/cutlass/pipeline/helpers.py:697
↓ 230 callersMethodproducer_commit
NOP for TMA based mainloop
examples/112_blackwell_ssd/utils/pipeline.h:142
↓ 223 callersMethodget
Returns a pointer to the managed object
tools/util/include/cutlass/util/device_memory.h:251
↓ 220 callersMethodat
Returns a reference to the element at the logical Coord in host memory
tools/util/include/cutlass/util/host_tensor.h:375
↓ 216 callersFunctionmake_gmem_ptr
include/cute/pointer.hpp:102
↓ 215 callersFunctionappend
include/cute/layout_composed.hpp:308
↓ 214 callersFunctionprint
include/cute/atom/mma_atom.hpp:656
↓ 211 callersMethodget_tma_tensor
include/cute/atom/copy_traits_sm90_tma.hpp:150
↓ 209 callersMethodcreate
Create a MoE persistent tile scheduler. :param params: Scheduler parameters (from host) :param offs: Cumsum tensor o
examples/python/CuTeDSL/cute/blackwell/kernel/moe/moe_persistent_scheduler.py:417
↓ 208 callersMethodmn
Obtains a Coord<2> from GemmCoord
include/cutlass/gemm_coord.h:162
↓ 206 callersMethodn
tools/profiler/include/cutlass/profiler/grouped_gemm_operation_profiler.h:111
↓ 204 callersMethodget_tma_descriptor
Return TmaDescriptor/TensorMap
include/cute/atom/copy_traits_sm90_tma.hpp:198
↓ 200 callersFunctionprint
include/cutlass/conv/convnd_problem_shape.hpp:579
↓ 200 callersMethodset
Sets runtime fields of the Atom. Some Atoms have runtime state, for example a tcgen05 MMA Atom .. code-block:: python
python/CuTeDSL/cutlass/cute/atom.py:200
↓ 198 callersFunctioncan_implement
include/cutlass/gemm/kernel/sm100_tile_scheduler_stream_k.hpp:220
↓ 195 callersMethodir_value
( self, *, loc: Optional[ir.Location] = None, ip: Optional[ir.InsertionPoint]
python/CuTeDSL/cutlass/cute/experimental/core.py:24
↓ 194 callersFunctionfilter_zeros
include/cute/layout.hpp:915
↓ 193 callersMethodnumel
Total number of elements (product of all shape dimensions).
python/CuTeDSL/cutlass/base_dsl/_mlir_helpers/arith.py:1184
↓ 188 callersFunctiongemm
include/cute/algorithm/gemm.hpp:68
↓ 186 callersMethodstride
Returns the layout object's stride vector
tools/util/include/cutlass/util/host_tensor.h:350
↓ 179 callersMethodreset
Clears the HostTensor allocation to size/capacity = 0
tools/util/include/cutlass/util/host_tensor.h:168
↓ 177 callersMethodm
tools/profiler/include/cutlass/profiler/grouped_gemm_operation_profiler.h:110
↓ 172 callersMethodfrom_tensor
(tensor: cute.Tensor)
python/CuTeDSL/cutlass/utils/layout.py:51
↓ 171 callersFunctiontransform
include/cute/algorithm/tuple_algorithms.hpp:205
↓ 170 callersMethodadd_tile_offset
include/cutlass/epilogue/threadblock/shared_load_iterator_mixed.h:185
↓ 168 callersFunctionTensorNorm
tools/util/include/cutlass/util/reference/host/tensor_reduce.h:154
↓ 168 callersMethodhost_ref
Accesses the tensor reference pointing to data
tools/util/include/cutlass/util/host_tensor.h:298
↓ 168 callersFunctionround_nearest
include/cutlass/fast_math.h:151
↓ 165 callersMethodshape
include/cute/tensor_impl.hpp:198
↓ 164 callersFunctionget_workspace_size
include/cutlass/gemm/kernel/sm90_tile_scheduler.hpp:139
↓ 159 callersMethoddata
include/cutlass/epilogue/collective/detail.hpp:171
↓ 158 callersFunctionconvert
(src: Tensor, dst: Tensor)
python/CuTeDSL/cutlass/cute/testing.py:617
↓ 157 callersMethodpartition_A
include/cute/atom/mma_atom.hpp:476
↓ 154 callersMethodk
tools/profiler/include/cutlass/profiler/grouped_gemm_operation_profiler.h:112
↓ 153 callersMethodempty
include/cute/container/array.hpp:153
↓ 153 callersMethodpartition_B
include/cute/atom/mma_atom.hpp:487
↓ 150 callersMethodw
Accesses the real part of the quaternion
include/cutlass/quaternion.h:188
↓ 149 callersFunctioncast_smem_ptr_to_uint
CUTE helper to cast SMEM pointer to unsigned
include/cute/arch/util.hpp:91
↓ 149 callersMethodget
examples/77_blackwell_fmha/reference/reference_abs_error.hpp:61
↓ 148 callersFunctionadd_tile_offset
include/cutlass/epilogue/threadblock/shared_load_iterator.h:155
↓ 148 callersFunctionproduct
(a)
python/pycute/int_tuple.py:65
next →1–100 of 19,553, ranked by callers