MCPcopy Create free account

hub / github.com/tile-ai/tilelang / functions

Functions7,254 in github.com/tile-ai/tilelang

Functionis_metal_arch
(arch: TileDevice)
tilelang/carver/arch/metal.py:6
Methodis_non_transform
(self)
tilelang/common/transform_kind.py:11
Methodis_output
(self)
tilelang/carver/roller/node.py:516
Methodis_placeholder
(self)
tilelang/carver/roller/node.py:183
Methodis_rdna_arch
Checks if the current architecture is an RDNA architecture. Returns: bool: True if the architecture is RDNA, False other
tilelang/carver/template/base.py:99
Methodis_scalar
Checks if the parameter represents a scalar value. Returns: bool: True if parameter has no dimensions (empty shape), Fal
tilelang/engine/param.py:70
Methodis_shape_aligned
(shape, factor)
tilelang/carver/roller/policy/default.py:785
Methodis_tcgen05
(self)
tilelang/tileop/gemm/gemm_base.py:164
Methodis_tcgen05
(self)
tilelang/tileop/gemm/__init__.py:114
Methodis_tma_descriptor_arg
(self, arg_name: str)
tilelang/jit/adapter/wrapper.py:274
Methodis_type_allowed
(dtype, vec)
tilelang/carver/roller/policy/default.py:788
Methodis_volta_arch
Checks if the current architecture is a Volta architecture. Returns: bool: True if the architecture is Volta, False othe
tilelang/carver/template/base.py:72
Functionisfinite
Check if input value is finite. Parameters ---------- x : PrimExpr Input argument. span : Optional[Span] The locatio
tilelang/language/tir/op.py:2990
Functionisinf
Check if input value is infinite. Parameters ---------- x : PrimExpr Input argument. span : Optional[Span] The locat
tilelang/language/tir/op.py:3009
Functionisnan
Check if input value is Nan. Parameters ---------- x : PrimExpr Input argument. span : Optional[Span] The location o
tilelang/language/tir/op.py:2952
Functionisnullptr
Check if input value is nullptr. Parameters ---------- x : PrimExpr Input argument. span : Optional[Span] The locati
tilelang/language/tir/op.py:2971
Functioniter_var
The iteration variable. Parameters ---------- var : Union[Var, str] The internal variable that is used for iteration. dom :
tilelang/language/ast/ir.py:1610
Functionjit
(func: Callable[_KP, _T])
tilelang/jit/__init__.py:525
Methodjit_elaborate
(**config_arg)
tilelang/autotuner/tuner.py:1284
Methodk_pack
(self)
tilelang/tileop/gemm/gemm_base.py:156
Methodk_pack
(self)
tilelang/tileop/gemm/__init__.py:106
Methodk_pack
(self)
tilelang/tileop/gemm_sp/__init__.py:43
Methodk_pack
(self)
tilelang/tileop/gemm_sp/gemm_sp_base.py:128
Functionkernel
( block_M=None, block_N=None, block_K=None, num_stages=None, thread_nu
examples/gemm/example_gemm_advanced_autotune.py:172
Functionkernel
(data, kernel, S, D, P, block_M, block_N, block_K, num_stages, threads, dtype=T.float16, accum_dtype=T.float32
examples/analyze/example_conv_analyze.py:20
Functionkernel
()
testing/python/cache/test_tilelang_kernel_cache_atomic_save.py:199
Functionkernel
(x: T.Tensor((N,), T.float32), y: T.Tensor((N,), T.float32))
testing/python/components/test_cuda_restrict_codegen.py:20
Functionkernel
( A: T.Tensor((32,), T.float16), B: T.Tensor((32,), T.float16), C: T.Tensor((32,), T.f
testing/python/components/test_cuda_shared_memory_alias_codegen.py:9
Functionkernel
(a: T.Tensor((length,), dtype), out: T.Tensor((1,), dtype))
testing/python/language/test_tilelang_language_reduce_maxmin_nan.py:22
Functionkernel
()
testing/python/language/test_tilelang_language_sync_threads.py:17
Functionkernel
(inp: T.Tensor((1,), T.int32), out: T.Tensor((1,), T.int32))
testing/python/issue/test_tilelang_issue_2188.py:19
Functionkernel
()
testing/python/issue/test_tilelang_issue_1734.py:10
Functionkernel
(status: T.Tensor((num_tiles,), T.int32), out: T.Tensor((1,), T.int32))
testing/python/issue/test_tilelang_issue_2123.py:13
Functionkernel
Add two vectors using one TileLang kernel.
testing/python/autotune/test_tilelang_autotune_cutedsl_cache.py:85
Functionkernel
( A: T.Tensor((M, K), T.float16), B: T.Tensor((K, N), T.float16), C: T.Tensor((M, N),
testing/python/transform/test_tilelang_transform_inject_set_max_nreg.py:172
Functionkernel
( A: T.Tensor((M, N), "float16"), B: T.Tensor((M, N), "float16"), )
testing/python/cuda/test_tma_multicast_demo.py:11
Functionkernel_body_annot
(x: T.Tensor((N,), T.float32), y: T.Tensor((N,), T.float32))
testing/python/components/test_cuda_restrict_codegen.py:35
Methodkernel_source
(self)
tilelang/jit/kernel.py:626
Functionkernel_with_scalar_kernel_binding
()
testing/python/issue/test_tilelang_issue_830.py:59
Functionkernel_with_tuple_kernel_binding
()
testing/python/issue/test_tilelang_issue_830.py:53
Functionladder_stage3_permutation_16x16_32x8_32x8_16x16
(kernel_i, kernel_j)
tilelang/carver/matmul_analysis.py:769
Functionladder_stage3_permutation_16x32_32x16_32x16_16x32
(kernel_i, kernel_j)
tilelang/carver/matmul_analysis.py:777
Methodlambda_forward
Args: args: List of input tensors stream: CUDA stream ID, default to -1, will use the current stream if n
tilelang/jit/adapter/cython/adapter.py:350
Functionlaneid_kernel
(A: T.Tensor((num_threads,), T.int32))
testing/python/language/test_tilelang_language_get_warp_info.py:27
Functionlaunch_thread
Launch a thread. Parameters ---------- thread : Union[IterVar, str] The iteration variable. extent : PrimExpr The ex
tilelang/language/ast/ir.py:1204
Methodlauncher
(*args: torch.Tensor)
tilelang/jit/adapter/torch/metal.py:71
Functionlayout_propagate_chain
( sch: Schedule, start_block: SBlockRV, start_buffer: tirx.Buffer, end_block: SBlockRV, in
tilelang/carver/matmul_analysis.py:797
Functionlazy_init
(self, name, mode=ctypes.DEFAULT_MODE, *args, **kwargs)
tilelang/__init__.py:140
Functionldexp
Returns x1 * (2 ** x2). Parameters ---------- x1 : PrimExpr Input argument. x2 : PrimExpr Input argument. Retur
tilelang/language/tir/op.py:2913
Functionldg128
Load 128 bits (16 bytes) from global memory using explicit PTX instructions. Usage: `T.ldg128(x[i])` or `T.ldg128(x[i:i+8])` emits `tl::ldg128(pt
tilelang/language/builtin.py:1557
Functionldg256
Load 256 bits (32 bytes) from global memory using explicit PTX instructions. Usage: `T.ldg256(x[i])` or `T.ldg256(x[i:i+16])` emits `tl::ldg256(p
tilelang/language/builtin.py:1583
Functionldg32
Load 32 bits (4 bytes) from global memory using explicit PTX instructions. Usage: `T.ldg32(x[i])` or `T.ldg32(x[i:i+2])` emits `tl::ldg32(ptr)`.
tilelang/language/builtin.py:1505
Functionldg64
Load 64 bits (8 bytes) from global memory using explicit PTX instructions. Usage: `T.ldg64(x[i])` or `T.ldg64(x[i:i+4])` emits `tl::ldg64(ptr)`.
tilelang/language/builtin.py:1531
Functionldmatrix_32x16_to_shared_16x32_layout_a
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:30
Functionldmatrix_32x16_to_shared_16x32_layout_b
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:36
Functionldmatrix_32x16_to_shared_32x16_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_sp_layout.py:132
Functionldmatrix_32x4_to_shared_16x8_layout_a
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:6
Functionldmatrix_32x4_to_shared_16x8_layout_b
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:12
Functionldmatrix_32x8_to_shared_16x16_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:18
Methodldmatrix_a
(self, A_local_buf, A_buf, ki, rk=0, pid_m=None, pid_n=None)
tilelang/rocm/intrinsics/mfma_macro_generator.py:846
Methodldmatrix_a
Load matrix A fragments from shared memory into per-thread local storage.
tilelang/cuda/intrinsics/macro/mma_sm70_macro_generator.py:190
Methodldmatrix_a
(self, A_local_buf, A_shared_buf, ki, rk=0)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:982
Methodldmatrix_a
(self, A_local_buf: Buffer, A_shared_buf: Buffer | BufferRegion | BufferLoad, ki: PrimExpr, rk: PrimExpr = 0)
tilelang/cuda/intrinsics/macro/mma_sp_macro_generator.py:226
Methodldmatrix_b
(self, B_local_buf, B_buf, ki, rk=0, pid_m=None, pid_n=None)
tilelang/rocm/intrinsics/mfma_macro_generator.py:920
Methodldmatrix_b
Load matrix B fragments from shared memory into per-thread local storage.
tilelang/cuda/intrinsics/macro/mma_sm70_macro_generator.py:232
Methodldmatrix_b
(self, B_local_buf, B_shared_buf, ki, rk=0)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:1077
Methodldmatrix_b
(self, B_local_buf: Buffer, B_shared_buf: Buffer | BufferRegion | BufferLoad, ki: PrimExpr, rk: PrimExpr = 0)
tilelang/cuda/intrinsics/macro/mma_sp_macro_generator.py:373
Functionldmatrix_permutation_16x16_32x8_16x16
(kernel_i, kernel_j)
tilelang/carver/matmul_analysis.py:713
Functionldmatrix_permutation_16x32_32x16_32x16
(kernel_i, kernel_j)
tilelang/carver/matmul_analysis.py:723
Functionldmatrix_trans_32x16_to_shared_16x32_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_sp_layout.py:138
Functionldmatrix_trans_32x32_to_shared_shared_16x64_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_sp_layout.py:144
Functionldmatrix_trans_32x8_to_shared_16x16_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_sp_layout.py:126
Functionldmatrix_trans_32x8_to_shared_16x16_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:24
Functionldmatrix_trans_permutation_16x16_32x8_16x16
(kernel_i, kernel_j)
tilelang/carver/matmul_analysis.py:718
Functionlet
Create a new let binding. Parameters ---------- v : Var The variable to bind. value : PrimExpr The value to be bound
tilelang/language/ast/ir.py:914
Methodlib_code
Returns the code of the compiled library.
tilelang/jit/adapter/cython/adapter.py:378
Methodlibpath
Returns the path to the compiled library.
tilelang/jit/adapter/cython/adapter.py:373
Functionlikely
Mark condition as likely. Parameters ---------- cond : PrimExpr Input argument. span : Optional[Span] The location
tilelang/language/tir/op.py:2932
Functionlinear_index
Compute a flat (linear) index from multi-dimensional coordinates and strides. The function accepts a sequence of PrimExpr arguments where th
tilelang/language/utils.py:51
Methodload_lib
Pretend to load a cached Python module from disk.
testing/python/jit/test_tilelang_jit_cutedsl_host_codegen.py:231
Methodload_lib
(self, lib_path: str | None = None)
tilelang/jit/adapter/libgen.py:53
Functionlog
(x: Union[TensorSSA, Numeric], fastmath: bool = False, **kwargs)
tilelang/contrib/cutedsl/math.py:90
Functionlog
Take log of input x. Parameters ---------- x : PrimExpr Input argument. Returns ------- y : PrimExpr The res
tilelang/language/tir/op.py:2321
Functionlog10
(x: Union[TensorSSA, Numeric], fastmath: bool = False, **kwargs)
tilelang/contrib/cutedsl/math.py:98
Functionlog10
Take log10 of input x. Parameters ---------- x : PrimExpr Input argument. Returns ------- y : PrimExpr The r
tilelang/language/tir/op.py:2353
Functionlog1p
Take log(x + 1) with respect to input x. Parameters ---------- x : PrimExpr Input argument. Returns ------- y : Prim
tilelang/language/tir/op.py:2369
Functionlog2
(x: Union[TensorSSA, Numeric], fastmath: bool = False, **kwargs)
tilelang/contrib/cutedsl/math.py:94
Functionlog2
Take log2 of input x. Parameters ---------- x : PrimExpr Input argument. Returns ------- y : PrimExpr The re
tilelang/language/tir/op.py:2337
Methodloglikelihood
(self, requests)
examples/bitnet-1.58b/eval_utils.py:106
Functionlookup_param
Returns the param by name Parameters ---------- param_name : str The name of param. span : Optional[Span] The locati
tilelang/language/tir/op.py:500
Functionloop_break
Break out of the innermost loop.
tilelang/language/builtin.py:1253
Functionloop_break
Break out of the current loop. Returns: tir.Call: A call to the `tl.loop_break` intrinsic.
tilelang/language/customize.py:76
Functionloop_layout_fn
(i, j)
testing/python/layout/test_tilelang_annotate_loop_layout.py:19
Functionloop_layout_kernel
(A, B, loop_layout)
testing/python/layout/test_tilelang_annotate_loop_layout.py:7
Methodlower
( self, layout_map: dict, target: Target, thread_bounds: Range, thread_var: tirx.Var, mbar_phase_expr:
tilelang/rocm/op/gemm/gemm_wmma.py:73
Methodlower
( self, layout_map: dict, target: Target, thread_bounds: Range, thread
tilelang/rocm/op/gemm/gemm_mfma.py:67
Methodlower
( self, layout_map: dict, target: Target, thread_bounds: Range, thread
tilelang/cpu/op/gemm/gemm_scalar.py:19
Methodlower
( self, layout_map: dict, target: Target, thread_bounds: Range, thread
tilelang/tileop/gemm/gemm_base.py:43
Methodlower
( self, layout_map: dict, target: Target, thread_bounds: Range, thread
tilelang/tileop/gemm_sp/gemm_sp_wgmma.py:73
Methodlower
(self, layout_map: dict, target: Target, thread_bounds: Range, thread_var: tirx.Var)
tilelang/tileop/gemm_sp/gemm_sp_base.py:18
← previousnext →4,901–5,000 of 7,254, ranked by callers