MCPcopy Create free account

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

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

Functionatomic_addx4
(A: T.Tensor((M, N), T.float32), B: T.Tensor((M, N), T.float32))
testing/python/language/test_tilelang_language_atomic.py:130
Functionatomic_addx4
Perform an atomic addition operation with quad-width operands. Args: dst (Buffer): Destination buffer where the atomic addition will be p
tilelang/language/atomic.py:319
Functionatomic_different_orders
( A: T.Tensor((M, N), dtype), B: T.Tensor((M, N), dtype), C: T.Tensor((M, N), dtype), D: T.Tensor((M,
testing/python/language/test_tilelang_language_atomic.py:95
Functionatomic_load
Load a value from the given buffer using the specified atomic memory ordering. Performs an atomic load from `src` and returns a PrimExpr rep
tilelang/language/atomic.py:357
Functionatomic_load_store
(A: T.Tensor((M, N), dtype), B: T.Tensor((M, N), dtype))
testing/python/language/test_tilelang_language_atomic.py:486
Functionatomic_max
(A: T.Tensor((K, M, N), dtype), B: T.Tensor((M, N), dtype))
testing/python/language/test_tilelang_language_atomic.py:420
Functionatomic_max
Perform an atomic maximum on the value stored at dst with an optional memory-order. Supports scalar/addressed extern atomic max when neither
tilelang/language/atomic.py:21
Functionatomic_min
(A: T.Tensor((K, M, N), dtype), B: T.Tensor((M, N), dtype))
testing/python/language/test_tilelang_language_atomic.py:453
Functionatomic_min
Atomically update the value at dst to the minimum of its current value and value. Supports scalar/addressed extern atomic min when neither a
tilelang/language/atomic.py:103
Functionatomic_store
Perform an atomic store of `src` into `dst` with the given memory ordering. Parameters: dst (Buffer): Destination buffer to store in
tilelang/language/atomic.py:401
Functionatomic_with_memory_order
(A: T.Tensor((K, M, N), dtype), B: T.Tensor((M, N), dtype))
testing/python/language/test_tilelang_language_atomic.py:39
Functionatomic_with_return_prev
(A: T.Tensor((M, N), dtype), B: T.Tensor((M, N), dtype), old_vals: T.Tensor((M, N), dtype))
testing/python/language/test_tilelang_language_atomic.py:162
Methodaug_assign
(self, op: Operator, target: Any, aug_value: Any, name: str | None = None)
tilelang/language/eager/ast.py:211
Methodaug_assign
(self, op, target, aug_value, name: str | None = None)
tilelang/language/eager/builder.py:566
Methodaug_assign_slice
(self, op: Operator, target: Any, sl: slice, aug_value: Any)
tilelang/language/eager/ast.py:214
Methodaug_assign_slice
(self, op, target, sl, aug_value)
tilelang/language/eager/builder.py:612
Functionauto_inline_consumer_chain
( sch: Schedule, block: SBlockRV, )
tilelang/carver/matmul_analysis.py:105
Functionauto_inline_producers
( sch: Schedule, block: SBlockRV, skip_blocks: list[SBlockRV] | None = None, )
tilelang/carver/matmul_analysis.py:61
Functionautotune
Just-In-Time (JIT) compiler decorator for TileLang functions. This decorator can be used without arguments (e.g., `@tilelang.jit`): A
tilelang/autotuner/tuner.py:1328
Methodax
(self)
tilelang/carver/roller/hint.py:41
Methodb_dtype
B operand dtype.
tilelang/tileop/gemm/gemm_base.py:99
Methodb_dtype
(self)
tilelang/tileop/gemm_sp/gemm_sp_base.py:67
Methodbackward
(ctx, do)
benchmark/blocksparse_attention/benchmark_triton_block_sparse_fmha.py:245
Methodbackward
(ctx, grad)
examples/dsa_sparse_finetune/dsa.py:20
Methodbackward
(ctx, grad_output)
examples/flash_attention/bert_padding.py:20
Methodbackward
(ctx, grad_output)
examples/flash_attention/bert_padding.py:52
Methodbackward
(ctx, grad_output, grad_residual)
examples/flash_attention/bert_padding.py:78
Methodbackward
(ctx, do)
examples/blocksparse_attention/block_sparse_attn_triton.py:245
Methodbackward
(ctx, do)
examples/seer_attention/block_sparse_attn_triton.py:240
Functionballot
Full-warp / full-wavefront ballot. Equivalent to ``ballot_sync(predicate)`` (i.e. with the default full warp mask). Returns: uint64:
tilelang/language/builtin.py:1078
Functionballot_sync
Return a ``uint64`` bitmask of lanes in ``mask`` whose predicate is set. CUDA: ``__ballot_sync(mask, predicate)`` returns ``unsigned int``; codeg
tilelang/language/builtin.py:1061
Functionbar_sync
(barrier_id, number_of_threads)
tilelang/contrib/cutedsl/reduce.py:184
Functionbarrier_arrive
Arrive at a memory barrier. Args: mbarrier: BarrierType The memory barrier to arrive at
tilelang/language/builtin.py:891
Functionbarrier_wait
Wait for a memory barrier to complete. Args: mbarrier: BarrierType The memory barrier to wait on parity: int | Var
tilelang/language/builtin.py:876
Methodbatch_size
(self)
examples/bitnet-1.58b/eval_utils.py:93
Functionbefore
(A_desc: T.handle("uint8x128", "grid_constant"))
testing/python/transform/test_tilelang_transform_fuse_mbarrier_arrive_expect_tx.py:32
Functionbefore
(A: T.Tensor((1024, 32), T.float16), B: T.Tensor((32, 1024), T.float16), C: T.Tensor((1024, 1024), T.float16))
testing/python/transform/test_tilelang_transform_cluster_planning.py:25
Functionbefore
(A: T.Buffer((1,), "float16"))
testing/python/transform/test_tilelang_transform_flatten_buffer.py:20
Functionbefore
( A: T.Tensor((128,), T.float32), B: T.Tensor((128,), T.float32), cond: T.Tensor((1,),
testing/python/transform/test_tilelang_transform_loop_unswitching.py:27
Functionbefore
( A: T.Tensor((16,), T.float32), B: T.Tensor((16,), T.float32), )
testing/python/transform/test_tilelang_transform_lower_tile_op.py:26
Functionbefore
(A: T.Tensor((1024,), T.float32))
testing/python/transform/test_tilelang_transform_legalize_negative_index.py:22
Functionbefore
()
testing/python/transform/test_tilelang_transform_lower_hopper_intrin.py:35
Functionbefore
()
testing/python/transform/test_tilelang_transform_hoist_broadcast_values.py:69
Functionbefore
(A: T.Buffer((4,), "float32"))
testing/python/transform/test_tilelang_transform_if_stmt_binding.py:15
Functionbefore
()
testing/python/transform/test_tilelang_transform_inject_fence_proxy.py:28
Functionbefore
(A: T.Tensor((128, 128), T.float32), B: T.Tensor((128, 128), T.float32))
testing/python/transform/test_tilelang_transform_let_inline.py:16
Functionbefore
( A: T.Tensor((1024, 32), T.float32), B: T.Tensor((32, 1024), T.float32), C: T.Tensor(
testing/python/transform/test_tilelang_transform_pipeline_planning.py:46
Functionbefore
()
testing/python/transform/test_tilelang_transform_inject_tcgen05_fence.py:76
Functionbefore
(b: T.Tensor[(16,), T.float4_e2m1fn])
testing/python/transform/test_tilelang_transform_decouple_type_cast.py:22
Functionbefore
(A: T.Buffer((1,), "float32"))
testing/python/transform/test_tilelang_transform_simplify.py:148
Functionbefore
()
testing/python/transform/test_tilelang_transform_make_packed_api.py:59
Functionbefore
( A: T.Tensor((16,), T.float32), B: T.Tensor((16,), T.float32), )
testing/python/transform/test_tilelang_transform_lower_ptx_async_copy.py:38
Functionbefore
( A: T.Tensor((16, 16), T.float32), B: T.Tensor((16, 16), T.float32), C: T.Tensor((16,
testing/python/transform/test_tilelang_transform_Inject_software_pipeline.py:158
Functionbefore
(A: T.Tensor((512, 512), T.float16), B: T.Tensor((512, 512), T.float16))
testing/python/transform/test_tilelang_transform_inject_set_max_nreg.py:42
Functionbefore_no_set_max_nreg
(A: T.Tensor((512, 512), T.float16))
testing/python/transform/test_tilelang_transform_inject_set_max_nreg.py:133
Functionbenchmark_fn
()
benchmark/blocksparse_attention/benchmark_torch_block_sparse_fmha.py:50
Functionbenchmark_fn
()
benchmark/blocksparse_attention/benchmark_triton_block_sparse_fmha.py:275
Functionbenchmark_fn
()
benchmark/blocksparse_attention/benchmark_library_dense_fmha.py:40
Functionbenchmark_fn
()
benchmark/blocksparse_attention/benchmark_tilelang_block_sparse_fmha.py:191
Functionbenchmark_pytorch
PyTorch benchmark - not used
examples/autodd/tilelang_buggy.py:94
Functionbind_assign_value
Value binding methods when parsing assign statement. e.g. binding vi, vj, vk with T.axis.remap("SSR", [i, j, k]), when parsing vi, vj, vk
tilelang/language/parser/parser.py:114
Functionbind_for_value
Value binding methods when parsing for statement. e.g. binding i, j, k with T.grid(128, 128, 128), when parsing for i, j, k in T.grid(128,
tilelang/language/parser/parser.py:78
Functionbind_with_value
Value binding methods when parsing with statement. e.g. binding i, j, k with T.grid(128, 128, 128), when parsing with T.grid(128, 128, 18)
tilelang/language/parser/parser.py:42
Functionbitcast
Reinterpret the bits of a value as a different type. Equivalent to C's (*(target_type *)(&value)). Args: value: Source value (Nu
tilelang/contrib/cutedsl/utils.py:65
Methodbits
(self)
tilelang/language/dtypes.py:16
Functionbitwise_and
Take bitwise and of two values Parameters ---------- x : PrimExpr Left operand y : PrimExpr Right operand span
tilelang/language/tir/op.py:2708
Functionbitwise_not
Take bitwise not of input value Parameters ---------- x : PrimExpr Input operand span : Optional[Span] The location
tilelang/language/tir/op.py:2730
Functionbitwise_or
Take bitwise or of two values Parameters ---------- x : PrimExpr Left operand y : PrimExpr Right operand span :
tilelang/language/tir/op.py:2749
Functionbitwise_reduce_input
()
testing/python/math/test_math_bitwise_reduce.py:96
Functionbitwise_xor
Take bitwise xor of two values Parameters ---------- x : PrimExpr Left operand y : PrimExpr Right operand span
tilelang/language/tir/op.py:2771
Functionblock
The block declaration statement. Parameters ---------- name : str The name of the block. no_realize : bool The flag
tilelang/language/ast/ir.py:344
Functionblock_attr
The block annotation statement. Parameters ---------- attrs : Dict[str, Any] The annotation of the block.
tilelang/language/ast/ir.py:432
Functionblock_rank_in_cluster
Return the 1-D rank of the calling CTA within its cluster (%%cluster_ctarank).
tilelang/language/cluster.py:47
Functionblock_sparse_matmul
( A: T.Tensor((M, K), dtype), B: T.Tensor((K, N), dtype), BlockMask: T.Tensor(block_ma
testing/python/language/test_tilelang_language_pipeline.py:129
Methodblocks
Returns the block indices from the topmost frame.
tilelang/language/kernel.py:246
Methodbody_rs
GEMM body_rs for SM90/Hopper: A from register, B from shared memory. Based on cute::tl_wgmma::GemmTensorOp::body_rs from gemm_sm90.h
tilelang/contrib/cutedsl/gemm_v1.py:521
Functionboolean
Construct a new tirx.Var with type boolean or cast expression to type boolean. Parameters ---------- expr: PrimExpr The expressio
tilelang/language/ast/ir.py:1445
Methodboolop
(self, op: BoolOp, left: Any, right: Callable[[], Any] | None = None)
tilelang/language/eager/ast.py:217
Methodboolop
(self, op, left, right=None)
tilelang/language/eager/builder.py:619
Functionbuffer_decl
(*args, **kwargs)
tilelang/language/ast/ir.py:167
Functionbuffer_store
Buffer store node. Parameters ---------- buffer : Buffer The buffer. value : PrimExpr The value to be stored. i
tilelang/language/ast/ir.py:1259
Functionbuffer_var
The pointer declaration function. Parameters ---------- dtype : str The data type of the pointer. storage_scope : str
tilelang/language/ast/ir.py:1553
Functionbuggy_kernel
(x: T.Tensor[(num_tokens,), T.float])
testing/python/components/test_storage_rewrite_detect_inplace.py:11
Functionbuggy_kernel
(x: T.Tensor[(num_tokens,), T.float32])
testing/python/issue/test_tilelang_issue_830.py:37
Functionbuggy_kernel
(x: T.Tensor[(num_tokens, hidden), T.float])
testing/python/issue/test_tilelang_issue_1001.py:14
Functionbuggy_kernel
(A, dtype=T.float32)
testing/python/issue/test_tilelang_issue_1719.py:83
Functionbuggy_kernel
(x: T.Tensor[(num_tokens,), "int64"])
testing/python/issue/test_tilelang_issue_1008.py:14
Functionbuild_date
Extract build date (YYYYMMDD) from version string. Args: version_str: Version string like "0.1.7.post3+cuda.d20260127.gita17230e4".
tilelang/utils/version.py:8
Methodbuild_inputs_with_special_tokens
(self, token_ids_0, token_ids_1=None)
examples/bitnet-1.58b/tokenization_bitnet.py:340
Methodbytes
(self)
tilelang/language/dtypes.py:18
Functionc2d_im2col
Deprecated alias for :func:`im2col`. Deprecated: Use :func:`im2col` instead. This alias is scheduled for removal in TileLang 0.14
tilelang/language/copy_op.py:586
Methodcache_dir
(self)
tilelang/autotuner/tuner.py:259
Functioncache_dirs
(tmp_path, monkeypatch)
testing/python/cache/test_tilelang_kernel_cache_atomic_save.py:31
Functioncache_dirs
(tmp_path, monkeypatch)
testing/python/autotune/test_tilelang_autotune_atomic_save.py:45
Functioncal_ks_ke_from_cu_seqlen_qk
seq_len: seq len per cp rank balanced cp slice assignment: 0 1 2 3 3 2 1 0
examples/deepseek_v32/utils.py:130
Functioncall_cpacked
Build expression by call an external packed function. Same as call_packed, except that the first argument is the function name (as in call_ex
tilelang/language/tir/op.py:55
Functioncall_cpacked_lowered
Lowered version of call c-packed. Same as call_packed, except that the first argument is the function name (as in call_extern), and the last a
tilelang/language/tir/op.py:109
Functioncall_extern
Build expression by calling a extern function. Parameters ---------- dtype : str The data type of the result. func_name: str
tilelang/language/tir/op.py:189
Functioncall_llvm_intrin
Build expression by calling a llvm intrinsic function Parameters ---------- dtype : str The data type of the result. name : s
tilelang/language/tir/op.py:215
Functioncall_llvm_pure_intrin
Build expression by calling a pure llvm intrinsic function Parameters ---------- dtype : str The data type of the result. nam
tilelang/language/tir/op.py:240
← previousnext →4,101–4,200 of 7,254, ranked by callers