Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/SHI-Labs/NATTEN
/ functions
Functions
2,059 in github.com/SHI-Labs/NATTEN
⨍
Functions
2,059
◇
Types & classes
573
Method
IndividualTileScheduler
csrc/include/natten/cuda/fmha_hopper/kernel/fmha_tile_scheduler.hpp:50
Method
Layout
csrc/include/natten/cuda/fmha/gemm/custom_mma_base.h:115
Method
Layout
csrc/include/natten/cuda/fna/gemm/custom_mma_base.h:115
Method
LayoutAccum
Returns a layout object for the Accum matrix
csrc/include/natten/cuda/fmha/gemm/mma_from_smem.h:118
Method
LayoutAccum
Returns a layout object for the Accum matrix
csrc/include/natten/cuda/fna/gemm/mma_from_smem.h:119
Method
LayoutB
csrc/include/natten/cuda/fmha/gemm/mma_from_smem.h:228
Method
LayoutB
csrc/include/natten/cuda/fna/gemm/mma_from_smem.h:229
Method
Mask
Mask
csrc/include/natten/cuda/fmha/iterators/epilogue_predicated_tile_iterator.h:156
Method
Mask
Mask
csrc/include/natten/cuda/fna/iterators/epilogue_predicated_tile_iterator.h:165
Method
Mask
Mask
csrc/include/natten/cuda/fna/epilogue/predicated_tile_iterator.h:162
Function
MatrixCoord thread_start
csrc/include/natten/cuda/fmha/iterators/epilogue_predicated_tile_iterator.h:659
Method
MmaBaseFromSharedMemory
Construct from tensor references
csrc/include/natten/cuda/fmha/gemm/mma_from_smem.h:252
Method
MmaBaseFromSharedMemory
Construct from tensor references
csrc/include/natten/cuda/fna/gemm/mma_from_smem.h:253
Method
NoOpWarpIteratorScale
csrc/include/natten/cuda/fmha/gemm/mma_from_smem.h:283
Method
NoOpWarpIteratorScale
csrc/include/natten/cuda/fna/gemm/mma_from_smem.h:284
Function
PYBIND11_MODULE
csrc/natten.cpp:39
Method
Params
Default ctor
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:153
Method
Params
Default ctor
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:514
Method
Params
Default ctor
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:739
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:990
Method
Params
Default ctor
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1295
Method
Params
Default ctor
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1514
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1736
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1963
Method
Params
csrc/include/natten/cuda/fmha/iterators/epilogue_predicated_tile_iterator.h:133
Method
Params
Construct the Params object given a pitch-linear tensor's layout
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:245
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:530
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:762
Method
Params
Construct the Params object given a pitch-linear tensor's layout
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:999
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1282
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1509
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1740
Method
Params
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1969
Method
Params
Default ctor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator_residual_last.h:159
Method
Params
Default ctor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator_residual_last.h:506
Method
Params
Default ctor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator_residual_last.h:820
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator.h:578
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator.h:953
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator.h:1286
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator.h:1509
Method
Params
Construct the Params object given a pitch-linear tensor's layout
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator.h:153
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator.h:441
Method
Params
Default constructor
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator.h:672
Method
Params
csrc/include/natten/cuda/fna/iterators/epilogue_predicated_tile_iterator.h:140
Method
Params
Construct the Params object given a pitch-linear tensor's layout
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator_residual_last.h:166
Method
Params
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator_residual_last.h:456
Method
Params
csrc/include/natten/cuda/fna/epilogue/predicated_tile_iterator.h:137
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:539
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:764
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1315
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1534
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1760
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:1987
Method
PredicatedTileAccessIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator_residual_last.h:844
Function
PredicatedTileAccessIteratorResidualLast operator++
Increment and return an instance to self.
csrc/include/natten/cuda/fmha/iterators/predicated_tile_access_iterator_residual_last.h:404
Function
PredicatedTileAccessIteratorResidualLast operator++
Increment and return an instance to self.
csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator_residual_last.h:402
Method
PredicatedTileIteratorPrefetch
Constructor
csrc/include/natten/cuda/fna/iterators/epilogue_predicated_tile_iterator.h:236
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:270
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:554
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:786
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1021
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1302
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1529
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1764
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1993
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator_residual_last.h:191
Method
PredicatedTileIteratorResidualLast
Constructs a TileIterator from its precomputed state, threadblock offset, and thread ID
csrc/include/natten/cuda/fna/iterators/predicated_tile_iterator_residual_last.h:479
Method
Sm100FmhaFwdEpilogueTmaWarpspecialized
csrc/include/natten/cuda/fmha_blackwell/collective/sm100_fmha_fwd_epilogue_tma_warpspecialized.hpp:148
Method
Sm100FnaFwdEpilogueTmaWarpspecialized
csrc/include/natten/cuda/fna_blackwell/collective/sm100_fna_fwd_epilogue_tma_warpspecialized.hpp:135
Method
TileSchedulerBwdAdapter
csrc/include/natten/cuda/fmha_hopper/kernel/fmha_tile_scheduler.hpp:186
Method
WarpIteratorFromSmem
Constructor from TensorRef
csrc/include/natten/cuda/fmha/iterators/warp_iterator_from_smem.h:166
Method
WarpIteratorFromSmem
Constructor from TensorRef
csrc/include/natten/cuda/fna/iterators/warp_iterator_from_smem.h:166
Method
__add__
(self, other)
src/natten/profiling_utils/formatting.py:190
Method
__eq__
(self, other)
src/natten/profiling_utils/formatting.py:181
Method
__ge__
(self, other)
src/natten/profiling_utils/formatting.py:172
Function
__global__ fna_bwd_reference_dK_kernel
csrc/include/natten/cuda/reference/fna_reference_backward.hpp:177
Function
__global__ fna_bwd_reference_dQ_kernel
csrc/include/natten/cuda/reference/fna_reference_backward.hpp:62
Function
__global__ fna_bwd_reference_dV_kernel
csrc/include/natten/cuda/reference/fna_reference_backward.hpp:294
Function
__global__ fna_reference_kernel
csrc/include/natten/cuda/reference/fna_reference_forward.hpp:63
Method
__gt__
(self, other)
src/natten/profiling_utils/formatting.py:178
Method
__init__
( self, na_dim: int, embed_dim: int, num_heads: int, kernel_size: Dime
src/natten/modules.py:43
Method
__init__
( self, embed_dim: int, num_heads: int, kernel_size: Dimension2DTypeOrDed,
src/natten/modules.py:317
Method
__init__
( self, embed_dim: int, num_heads: int, kernel_size: Dimension3DTypeOrDed,
src/natten/modules.py:426
Method
__init__
( self, na_dim: int, batch_size: int, heads: int, dim: int, in
src/natten/profiling_utils/problem.py:35
Method
__init__
( self, op: Optional[LibNattenOp], op_str: str, time: float, index: in
src/natten/profiling_utils/formatting.py:102
Method
__init__
(self, name: str)
src/natten/utils/log.py:100
Method
__init__
(self, name, short_name, torch_name, bits)
scripts/autogen_hopper_fna_bwd.py:133
Method
__init__
( self, na_dim: int, dtype: DataType, gemm_shape: Tuple[int, int, int],
scripts/autogen_hopper_fna_bwd.py:155
Method
__init__
( self, )
scripts/autogen_hopper_fna_bwd.py:269
Method
__init__
(self, na_dim: int)
scripts/autogen_hopper_fna_bwd.py:304
Method
__init__
(self, na_dim: int, dtype: DataType)
scripts/autogen_hopper_fna_bwd.py:339
Method
__init__
( self, na_dim: int, dtype: DataType, head_dim: int, )
scripts/autogen_hopper_fna_bwd.py:379
Method
__init__
( self, na_dim: int, dtype: DataType, head_dim: int, causal_mask: tupl
scripts/autogen_hopper_fna_bwd.py:439
Method
__init__
(self, na_dim: int, sm: int, gemm_shape: Tuple[int, int, int])
scripts/autogen_fna.py:34
Method
__init__
(self, na_dim: int, sm: int, gemm_shapes: List[Tuple[int, int, int]])
scripts/autogen_fna.py:48
Method
__init__
(self, name, torch_name, short_name, bits, min_sm)
scripts/autogen_fna.py:68
Method
__init__
( self, is_backward: bool, dtype: DataType, config: KernelConfig, caus
scripts/autogen_fna.py:118
Method
__init__
( self, is_backward: bool, na_dim: int, sm: int, dtype: DataType,
scripts/autogen_fna.py:195
Method
__init__
(self, is_backward: bool)
scripts/autogen_fna.py:299
Method
__init__
(self, is_backward: bool, na_dim: int)
scripts/autogen_fna.py:354
← previous
next →
801–900 of 2,059, ranked by callers