MCPcopy Create free account

hub / github.com/NVIDIA/cutlass / functions

Functions19,553 in github.com/NVIDIA/cutlass

⨍ Functions 19,553◇ Types & classes 10,397↳ Endpoints 26
↓ 7 callersMethodinitialize
Initializes GEMM state from arguments.
examples/13_two_tensor_op_fusion/device/b2b_gemm.h:238
↓ 7 callersFunctioninitialize_values
test/unit/conv/device_3x/testbed_conv.hpp:71
↓ 7 callersFunctionis_arg_annotation_constexpr
Check if the argument annotation is a constexpr.
python/CuTeDSL/cutlass/base_dsl/runtime/jit_arg_adapters.py:41
↓ 7 callersMethodis_complex
(self)
python/cutlass_library/gemm_operation.py:129
↓ 7 callersMethodlayout
Returns the layout object
include/cutlass/tensor_ref.h:285
↓ 7 callersMethodlayout_a
Mapping functions (A,B,C -> activation,filter,output)
tools/library/include/cutlass/library/library.h:877
↓ 7 callersMethodlayout_acc_mn
(self, tiled_mma, acc)
examples/python/CuTeDSL/cute/hopper/kernel/attention/fmha.py:1677
↓ 7 callersMethodlayout_b
tools/library/include/cutlass/library/library.h:886
↓ 7 callersFunctionlex_less
include/cute/int_tuple.hpp:809
↓ 7 callersFunctionlog_debug_line
Log one line of debug output
python/cutlass_library/generator.py:55
↓ 7 callersFunctionmake_basis_like
include/cute/numeric/arithmetic_tuple.hpp:319
↓ 7 callersFunctionmake_fragment
( layout_or_shape: Union[Layout, Shape], dtype: Type[Numeric], *, loc: Optional[ir.Location] =
python/CuTeDSL/cutlass/cute/tensor.py:900
↓ 7 callersFunctionmake_tiled_copy_C
include/cute/atom/copy_atom.hpp:440
↓ 7 callersMethodor_
Create a logical OR operation between two values.
python/CuTeDSL/cutlass/base_dsl/tvm_ffi_builder/mlir_builder.py:186
↓ 7 callersMethodout_degree
Get the output degree of node
python/cutlass_cppgen/backend/evt/ir/dag_ir.py:136
↓ 7 callersFunctionpartition_accumulator_shape
Construct A Single Stage's Accumulator Shape
include/cutlass/gemm/collective/sm100_mma_array_warpspecialized.hpp:466
↓ 7 callersFunctionprefetch
examples/41_fused_multi_head_attention/iterators/epilogue_predicated_tile_iterator.h:289
↓ 7 callersMethodpure
include/cutlass/quaternion.h:201
↓ 7 callersFunctionrecast_layout
Recast a layout from one data type to another. :param new_type_bits: The new data type bits :type new_type_bits: int :param old_type
python/CuTeDSL/cutlass/cute/core.py:4157
↓ 7 callersFunctionrepeat_like
Creates an object congruent to target and filled with x. This function recursively creates a nested tuple structure that matches the structure
python/CuTeDSL/cutlass/cute/core.py:3210
↓ 7 callersMethodrun
Runs the kernel currently specified. If it has not already been, the kernel is emitted and compiled. Tensors holding operands and out
python/cutlass_cppgen/op/gemm.py:632
↓ 7 callersMethodrun
Executes one test
test/unit/conv/device/conv2d_testbed.h:218
↓ 7 callersMethodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
examples/41_fused_multi_head_attention/iterators/predicated_tile_iterator_residual_last.h:862
↓ 7 callersMethodset_mask
Sets the predicate mask, overriding value stored in predicate iterator
include/cutlass/transform/threadblock/predicated_tile_iterator.h:1447
↓ 7 callersFunctionshared_load
include/cutlass/arch/memory_sm75.h:231
↓ 7 callersMethodsmem_load_and_partition_delta_d
( self, tiled_s2r_b, local_tidx, smem_delta, smem_tile_coord )
examples/python/CuTeDSL/cute/blackwell/kernel/attention/mamba2_ssd/mamba2_ssd.py:3094
↓ 7 callersMethodsync
include/cutlass/epilogue/collective/sm70_epilogue_vectorized.hpp:296
↓ 7 callersFunctiontapply
test/unit/gemm/device/gemm_testbed_3x_evt.hpp:45
↓ 7 callersMethodtensormaps_perform_update
include/cutlass/gemm/collective/sm90_mma_array_tma_gmma_ss_warpspecialized.hpp:714
↓ 7 callersFunctiontensormaps_replace_global_tensor_properties
include/cutlass/gemm/collective/sm100_mma_array_warpspecialized.hpp:823
↓ 7 callersMethodto
Convert the tensor to a different numeric type. :param dtype: The target numeric type to cast to. :type dtype: Type[Numeric]
python/CuTeDSL/cutlass/cute/tensor.py:2385
↓ 7 callersFunctiontorch_type
(inp)
python/cutlass_cppgen/utils/datatypes.py:208
↓ 7 callersFunctionvalid
Returns whether access is valid or not
include/cutlass/conv/threadblock/predicated_scale_bias_vector_access_iterator.h:274
↓ 7 callersMethodwriteTensor
(self, tensor: torch.Tensor, name: str, stride_names: List[str])
examples/41_fused_multi_head_attention/piped_subprocess.py:77
↓ 6 callersMethodCallbacks
include/cutlass/epilogue/threadblock/fusion/visitor_2x.hpp:179
↓ 6 callersFunctionCreateGemmPlanarComplexOperator
(manifest, layouts, tile_descriptions, data_type, \ alignment_constraints, complex_transforms)
python/cutlass_library/generator.py:358
↓ 6 callersFunctionGemm3x
tools/util/include/cutlass/util/reference/host/gett.hpp:837
↓ 6 callersMethodMatrixCoord
Default ctor
include/cutlass/matrix_coord.h:73
↓ 6 callersFunctionSmId
Computes SM number the thread is running on
include/cutlass/arch/arch.h:58
↓ 6 callersMethodTensor4DCoord
Default ctor
include/cutlass/tensor_coord.h:72
↓ 6 callersMethod__post_init__
(self)
python/CuTeDSL/cutlass/cute/nvgpu/warp/copy.py:36
↓ 6 callersMethod__str__
String representation of the ratio. :return: String in the format "numerator/denominator" :rtype: str
python/CuTeDSL/cutlass/cute/core.py:680
↓ 6 callersMethod_check_can_load_store
(self, vectorized: bool = False)
python/CuTeDSL/cutlass/cute/tensor.py:656
↓ 6 callersMethod_get_pointer
(self, address_as_int, cute_type)
examples/python/CuTeDSL/cute_ext/blackwell/dense_gemm_ptr_array.py:83
↓ 6 callersMethod_insert_cf_symbol_check
Insert a check for range symbol
python/CuTeDSL/cutlass/base_dsl/ast_preprocessor.py:1403
↓ 6 callersMethod_is_scalar_type
Checks if the given type is a scalar numeric type. :param dtype: The type to check. :return: True if the type is a subclass
python/CuTeDSL/cutlass/cute/core.py:5799
↓ 6 callersFunction_loop_and_call_kernel
(iterations: int, workspace_index: int = 0)
python/CuTeDSL/cutlass/cute/testing.py:903
↓ 6 callersMethod_make_trait
( self, *, loc: Optional[ir.Location] = None, ip: Optional[ir.InsertionPoint]
python/CuTeDSL/cutlass/cute/atom.py:56
↓ 6 callersMethod_reset_operations
(self, reset_epilogue: bool = True)
python/cutlass_cppgen/op/gemm.py:284
↓ 6 callersFunction_reverse_tuple
(nested_tuple: tuple)
python/cutlass_cppgen/backend/evt/ir/layout_algorithm.py:106
↓ 6 callersFunction_run_compile
Compile-only helper (iterations=0).
test/examples/CuTeDSL/hopper/test_grouped_gemm.py:84
↓ 6 callersFunction_run_correctness
Correctness helper: runs kernel and validates against torch.einsum reference.
test/examples/CuTeDSL/hopper/test_dense_gemm_fp8_2xacc.py:104
↓ 6 callersMethod_thrfrg_A
( self, input: Union[Layout, Tensor], *, loc: Optional[ir.Location] = None,
python/CuTeDSL/cutlass/cute/atom.py:615
↓ 6 callersMethod_to_ir
(self)
python/CuTeDSL/cutlass/cute/nvgpu/cpasync/copy.py:65
↓ 6 callersMethod_verify_scalar
Verifies the following properties: 1) Either ``scalar`` or ``ref_scakar`` must be set (i.e., not ``None``) 2) If ``sc
python/cutlass_cppgen/op/op.py:137
↓ 6 callersFunctiona0
(bv,e)
docs/jquery.js:23
↓ 6 callersMethodaccumulator_type
(self)
python/cutlass_cppgen/backend/gemm_operation.py:1591
↓ 6 callersFunctionadd_pointer_offset
Adds a pointer offset in units of Element
include/cutlass/transform/threadblock/regular_tile_iterator_tensor_op_sm70.h:166
↓ 6 callersFunctionalign_up
examples/41_fused_multi_head_attention/gemm_kernel_utils.h:141
↓ 6 callersFunctionappend
( input: Layout, elem: Layout, up_to_rank: Optional[int] = None, *, loc: Optional[ir.Locat
python/CuTeDSL/cutlass/cute/core.py:3019
↓ 6 callersMethodapply_mask
Apply the appropriate mask to the attention scores. This method modifies the attention scores (acc_qk) based on the mask type
examples/python/CuTeDSL/utils/fmha_helpers.py:902
↓ 6 callersMethodargument_type
Default class for Argument Type
python/cutlass_cppgen/backend/evt/ir/node.py:95
↓ 6 callersFunctionblock_idx
Returns the CTA identifier within a grid.
python/CuTeDSL/cutlass/cute/arch/nvvm_wrappers.py:299
↓ 6 callersMethodcall
include/cute/atom/mma_atom.hpp:92
↓ 6 callersFunctioncan_implement
Determines whether the Implicit GEMM can execute the given problem.
include/cutlass/conv/threadblock/conv3d_fprop_filter_tile_access_iterator_analytic.h:239
↓ 6 callersMethodcapacity
Compute the number of contiguous elements needed to store a tensor with the given size
include/cutlass/layout/tensor_op_multiplicand_sm70.h:188
↓ 6 callersMethodcapacity
Compute the number of contiguous elements needed to store a tensor with the given size
include/cutlass/layout/tensor_op_multiplicand_sm75.h:216
↓ 6 callersFunctioncluster_sync
include/cute/arch/cluster_sm90.hpp:75
↓ 6 callersMethodcompile
Emits and compiles the kernel currently specified. If ``tile_description`` and any of the ``alignment`` parameters are set, the kerne
python/cutlass_cppgen/op/gemm.py:478
↓ 6 callersFunctioncomplement
include/cute/layout.hpp:1178
↓ 6 callersFunctioncompute_lower_corner_whd
include/cutlass/conv/collective/detail.hpp:188
↓ 6 callersFunctioncompute_upper_corner_whd
include/cutlass/conv/collective/detail.hpp:212
↓ 6 callersMethodconsumer_wait
include/cutlass/pipeline/sm100_pipeline.hpp:1172
↓ 6 callersFunctioncopy_host_to_host
tools/util/include/cutlass/util/device_memory.h:149
↓ 6 callersFunctioncopy_tiles_and_advance
include/cutlass/conv/threadblock/implicit_gemm_multistage.h:84
↓ 6 callersFunctioncrd2idx
Convert a multi-dimensional coordinate into a value using the specified layout. This function computes the inner product of the flattened co
python/CuTeDSL/cutlass/cute/core.py:3989
↓ 6 callersFunctioncreate_and_permute_tensor
( l, mode0, mode1, is_mode0_major, dtype, is_dynamic_layout=True )
examples/python/CuTeDSL/cute/hopper/kernel/dense_gemm/dense_gemm.py:1472
↓ 6 callersFunctioncreate_and_permute_tensor
(mode0, mode1, is_mode0_major, dtype)
examples/python/CuTeDSL/cute/ampere/kernel/dense_gemm/sgemm.py:682
↓ 6 callersFunctioncreate_and_permute_tensor
(l, mode0, mode1, is_mode0_major, dtype)
examples/python/CuTeDSL/cute/ampere/kernel/dense_gemm/tensorop_gemm.py:878
↓ 6 callersFunctioncreate_tensor
(shape, dtype)
examples/python/CuTeDSL/cute/blackwell/kernel/attention/mixed_input_fmha/mixed_input_fmha_prefill_d512.py:1811
↓ 6 callersFunctioncreate_tensor
(shape, dtype)
examples/python/CuTeDSL/cute/blackwell/kernel/attention/mixed_input_fmha/mixed_input_fmha_prefill_d256.py:1671
↓ 6 callersFunctiondrain_cp_asyncs
examples/41_fused_multi_head_attention/gemm/mma_from_smem.h:535
↓ 6 callersMethodepilogue_tma_store
Epilogue implementation for TMA store version. :param epi_tidx: Thread index :type epi_tidx: cutlass.Int32 :param wa
examples/python/CuTeDSL/cute/blackwell/kernel/dense_gemm/dense_gemm.py:940
↓ 6 callersMethodexec_mma
( self, tiled_mma, tCtAcc, tCrA, tCrB, acc_producer_state,
examples/python/CuTeDSL/cute/blackwell/kernel/attention/mamba2_ssd/mamba2_ssd.py:2892
↓ 6 callersMethodexponent_biased
Returns the biased exponent
include/cutlass/tfloat32.h:173
↓ 6 callersMethodexponent_biased
Returns the biased exponent
include/cutlass/half.h:475
↓ 6 callersMethodexponent_biased
Returns the biased exponent
include/cutlass/bfloat16.h:211
↓ 6 callersFunctionfast_tanh
include/cutlass/fast_math.h:921
↓ 6 callersMethodfilter
Filter the archs by the given criterion.
python/CuTeDSL/cutlass/base_dsl/arch.py:150
↓ 6 callersFunctionfind_if
( t: XTuple, pred_fn: Callable[[XTuple, int], bool], hierarchical: bool = True, )
python/CuTeDSL/cutlass/cute/tuple.py:226
↓ 6 callersFunctionflatten
(a: Layout)
python/CuTeDSL/cutlass/cute/core.py:3241
↓ 6 callersFunctionfull_like
Return a full TensorSSA with the same shape and type as a given array. :param a: The shape and data-type of `a` define these same attributes
python/CuTeDSL/cutlass/cute/tensor.py:2654
↓ 6 callersMethodget
( self, field: Any, *, loc: Optional[ir.Location] = None, ip: Optional
python/CuTeDSL/cutlass/cute/nvgpu/warp/mma.py:372
↓ 6 callersMethodget_block_coord
examples/88_hopper_fmha/kernel/fmha_tile_scheduler.hpp:72
↓ 6 callersMethodget_block_coord
examples/111_hopper_ssd/kernel/sm90_ssd_tile_scheduler.hpp:100
↓ 6 callersFunctionget_cudnn_layout
Maps a CUTLASS tensor layout to a cuDNN cudnnTensorFormat_t
tools/profiler/src/cudnn_helpers.cpp:95
↓ 6 callersMethodget_current_work
include/cutlass/gemm/kernel/static_tile_scheduler.hpp:163
↓ 6 callersMethodget_edge_weight
Get the edge weight of edge src->dst
python/cutlass_cppgen/backend/evt/ir/dag_ir.py:154
↓ 6 callersMethodget_grid_shape
Computes the grid shape based on the maximum active clusters allowed. :param max_active_clusters: The maximum number of active clust
python/CuTeDSL/cutlass/utils/static_persistent_tile_scheduler.py:291
↓ 6 callersMethodget_grid_shape
Computes the grid shape
examples/111_hopper_ssd/device/ssd.hpp:101
← previousnext →1,101–1,200 of 19,553, ranked by callers