Functions — github.com/SHI-Labs/NATTEN

CodeHub MCPcopy Create free account

hub / github.com/SHI-Labs/NATTEN / functions

Functions2,059 in github.com/SHI-Labs/NATTEN

⨍ Functions 2,059◇ Types & classes 573

Methodget_workspace_size

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_kernel_bwd_sum_OdO.hpp:81

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/utils/generic_cutlass_device.hpp:97

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fna_hopper/device/fna_bwd_sm90.hpp:281

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fna_hopper/device/fna_sm90.hpp:96

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fna_blackwell/device/fna_sm100.hpp:96

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fna_blackwell/device/fna_bwd_sm100.hpp:286

Methodget_workspace_size

csrc/include/natten/cuda/fna_blackwell/kernel/sm100_fna_fwd_kernel_tma_warpspecialized.hpp:198

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fmha_hopper/device/fmha_sm90.hpp:96

Methodget_workspace_size

Gets the workspace size

csrc/include/natten/cuda/fmha_hopper/device/fmha_bwd_sm90.hpp:255

Methodget_workspace_size

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_tma_warpspecialized.hpp:147

Methodget_workspace_size

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_bwd_convert.hpp:76

Methodget_workspace_size

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_tma.hpp:113

Methodget_workspace_size

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_bwd_sum_OdO.hpp:83

Methodget_workspace_size

csrc/include/natten/cuda/reduction/fmha_kernel_bwd_sum_OdO.hpp:67

scripts/autogen_fna.py:233

scripts/autogen_fmha.py:206

csrc/include/natten/cuda/fmha/epilogue/epilogue_pipelined.h:277

csrc/include/natten/cuda/fmha/epilogue/epilogue_pipelined.h:421

csrc/include/natten/cuda/fna/epilogue/epilogue_pipelined.h:277

csrc/include/natten/cuda/fna/epilogue/epilogue_pipelined.h:421

Functionhopper_fmha_backward_torch_fake_op

( query: Tensor, key: Tensor, value: Tensor, output: Tensor, d_output: Tensor, logsume

src/natten/_libnatten/torch_wrappers.py:425

Functionhopper_fmha_backward_torch_op

( query: Tensor, key: Tensor, value: Tensor, output: Tensor, d_output: Tensor, logsume

src/natten/_libnatten/torch_wrappers.py:368

Functionhopper_fmha_forward_torch_fake_op

( query: Tensor, key: Tensor, value: Tensor, is_causal: bool, scale: float, q_tile_siz

src/natten/_libnatten/torch_wrappers.py:338

Functionhopper_fmha_forward_torch_op

( query: Tensor, key: Tensor, value: Tensor, is_causal: bool, scale: float, q_tile_siz

src/natten/_libnatten/torch_wrappers.py:284

Functionhopper_fna_backward_torch_fake_op

( query: Tensor, key: Tensor, value: Tensor, output: Tensor, d_output:

src/natten/_libnatten/torch_wrappers.py:984

Functionhopper_fna_backward_torch_op

( query: Tensor, key: Tensor, value: Tensor, output: Tensor, d_output:

src/natten/_libnatten/torch_wrappers.py:932

Functionhopper_fna_forward_torch_fake_op

( query: Tensor, key: Tensor, value: Tensor, kernel_size: list[int], s

src/natten/_libnatten/torch_wrappers.py:899

Functionhopper_fna_forward_torch_op

( query: Tensor, key: Tensor, value: Tensor, kernel_size: list[int], s

src/natten/_libnatten/torch_wrappers.py:852

csrc/include/natten/cuda/fmha/iterators/warp_iterator_from_smem.h:177

csrc/include/natten/cuda/fna/iterators/warp_iterator_from_smem.h:177

MethodincrIteration

Returns the next block to process

csrc/include/natten/cuda/fmha/kernel_backward.h:2024

csrc/include/natten/cuda/fna_hopper/collective/fna_collective_load.hpp:66

csrc/include/natten/cuda/fmha_hopper/collective/fmha_collective_load.hpp:68

Methodinitialize

Initializes GEMM state from arguments.

csrc/include/natten/cuda/fmha_blackwell/device/fmha_sm100.hpp:150

Methodinitialize

Initializes GEMM state from arguments.

csrc/include/natten/cuda/utils/generic_cutlass_device.hpp:151

Methodinitialize

Initializes GEMM state from arguments.

csrc/include/natten/cuda/fna_hopper/device/fna_sm90.hpp:150

Methodinitialize

Initializes GEMM state from arguments.

csrc/include/natten/cuda/fna_blackwell/device/fna_sm100.hpp:150

Methodinitialize

csrc/include/natten/cuda/fna/iterators/predicated_tile_access_iterator.h:99

Methodinitialize

csrc/include/natten/cuda/fna/epilogue/predicated_tile_iterator_params.h:87

Methodinitialize

Initializes GEMM state from arguments.

csrc/include/natten/cuda/fmha_hopper/device/fmha_sm90.hpp:150

Methodinitialize_split

Initializes state from arguments.

csrc/include/natten/cuda/fmha_blackwell/device/fmha_bwd_sm100.hpp:317

Methodinitialize_split

Initializes state from arguments.

csrc/include/natten/cuda/fna_hopper/device/fna_bwd_sm90.hpp:299

Methodinitialize_split

Initializes state from arguments.

csrc/include/natten/cuda/fna_blackwell/device/fna_bwd_sm100.hpp:304

Methodinitialize_split

Initializes state from arguments.

csrc/include/natten/cuda/fmha_hopper/device/fmha_bwd_sm90.hpp:273

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_blackwell/kernel/sm100_fmha_fwd_kernel_tma_warpspecialized.hpp:244

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_kernel_bwd_convert.hpp:83

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_blackwell/kernel/sm100_fmha_bwd_kernel_tma_warpspecialized.hpp:480

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_kernel_bwd_sum_OdO.hpp:84

Methodinitialize_workspace

csrc/include/natten/cuda/fna_blackwell/kernel/sm100_fna_bwd_kernel_tma_warpspecialized.hpp:483

Methodinitialize_workspace

csrc/include/natten/cuda/fna_blackwell/kernel/sm100_fna_fwd_kernel_tma_warpspecialized.hpp:201

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_tma_warpspecialized.hpp:150

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_bwd_convert.hpp:79

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_tma.hpp:116

Methodinitialize_workspace

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_kernel_bwd_sum_OdO.hpp:86

Methodinitialize_workspace

csrc/include/natten/cuda/reduction/fmha_kernel_bwd_sum_OdO.hpp:70

csrc/include/natten/cuda/fmha_blackwell/collective/fmha_fusion.hpp:301

csrc/include/natten/cuda/fmha_hopper/collective/fmha_varlen.hpp:47

Methodis_contributing

csrc/include/natten/cuda/fmha_hopper/collective/fmha_fusion.hpp:271

Functionis_full

(dtype: torch.dtype)

src/natten/utils/dtype.py:27

Functionis_half

(dtype: torch.dtype)

src/natten/utils/dtype.py:31

Methodis_initialized

csrc/include/natten/cuda/fmha_blackwell/device/fmha_sm100.hpp:73

Methodis_initialized

csrc/include/natten/cuda/utils/generic_cutlass_device.hpp:74

Methodis_initialized

csrc/include/natten/cuda/fna_hopper/device/fna_sm90.hpp:73

Methodis_initialized

csrc/include/natten/cuda/fna_blackwell/device/fna_sm100.hpp:73

Methodis_initialized

csrc/include/natten/cuda/fmha_hopper/device/fmha_sm90.hpp:73

Functionis_memory_usage_default

Returns whether memory usage preference for KV parallelism in `"cutlass-fna"` and `"cutlass-fmha"` backends is the default setting.

src/natten/context.py:78

Methodis_self_attn

src/natten/profiling_utils/problem.py:78

Methodis_source_needed

csrc/include/natten/cuda/fmha/epilogue/epilogue_rescale_output.h:146

Methodis_source_needed

csrc/include/natten/cuda/fna/epilogue/epilogue_rescale_output.h:146

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_causal_tile_scheduler.hpp:88

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_causal_tile_scheduler.hpp:194

csrc/include/natten/cuda/fmha_blackwell/kernel/fmha_tile_scheduler.hpp:151

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_tile_scheduler.hpp:155

csrc/include/natten/cuda/fmha_hopper/kernel/fmha_tile_scheduler.hpp:207

Functioniterable_to_static_cute_tuple

scripts/autogen_reference_fna.py:184

MethoditerateRows

csrc/include/natten/cuda/fmha/gemm/mma_accum_lambda_iterator.h:50

MethoditerateRows

csrc/include/natten/cuda/fmha/gemm/mma_accum_lambda_iterator.h:158

MethoditerateRows

csrc/include/natten/cuda/fmha/gemm/mma_accum_lambda_iterator.h:231

MethoditerateRows

csrc/include/natten/cuda/fna/gemm/mma_accum_lambda_iterator.h:84

MethoditerateRows

csrc/include/natten/cuda/fna/gemm/mma_accum_lambda_iterator.h:195

MethoditerateRows

csrc/include/natten/cuda/fna/gemm/mma_accum_lambda_iterator.h:270

Methoditerative_softmax

csrc/include/natten/cuda/fmha/kernel_forward.h:923

Methoditerative_softmax

csrc/include/natten/cuda/fna/kernel_forward.h:980

Methoditerator_V

csrc/include/natten/cuda/fmha/kernel_forward.h:602

Functionkernel_type_int_to_enum_type

csrc/include/natten/cuda/hopper_fmha_fna.h:54

scripts/autogen_fna.py:222

scripts/autogen_fmha.py:196

csrc/include/natten/cuda/fmha/kernel_forward.h:1062

csrc/include/natten/cuda/fna/kernel_forward.h:1125

Functionlibnatten_import_error

(*args, **kwargs)

src/natten/_libnatten/stubs.py:25

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/epilogue_predicated_tile_iterator.h:414

csrc/include/natten/cuda/fmha_blackwell/collective/sm100_fmha_fwd_mainloop_tma_warpspecialized.hpp:278

csrc/include/natten/cuda/fmha_blackwell/collective/sm100_fmha_load_tma_warpspecialized.hpp:147

load a tile from global memory into shared memory

csrc/include/natten/cuda/fmha/transform/tile_smem_loader.h:55

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:409

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:664

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:894

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1161

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1411

Loads a fragment from memory

csrc/include/natten/cuda/fmha/iterators/predicated_tile_iterator_residual_last.h:1638

← previous next →1,401–1,500 of 2,059, ranked by callers