MCPcopy Create free account

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

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

Methodget_hardware_aware_configs
(self, arch: TileDevice = None, topk: int = 10)
tilelang/carver/template/general_reduce.py:17
Methodget_hardware_aware_configs
Retrieves optimized hardware-aware configurations. Args: arch (TileDevice, optional): The target hardware architecture.
tilelang/carver/template/conv.py:47
Methodget_hardware_aware_configs
Retrieves hardware-aware optimization configurations. Args: arch (TileDevice, optional): The target hardware architectur
tilelang/carver/template/elementwise.py:24
Methodget_host_source
Return a fake host wrapper source.
testing/python/autotune/test_tilelang_autotune_cutedsl_cache.py:39
Methodget_host_source
Returns the source code of the host function.
tilelang/jit/adapter/cython/adapter.py:400
Methodget_host_source
Get the cached host-side source code.
tilelang/jit/adapter/nvrtc/adapter.py:210
Methodget_init_func
(self)
tilelang/jit/adapter/wrapper.py:712
Methodget_input_embeddings
(self)
examples/bitnet-1.58b/modeling_bitnet.py:977
Methodget_input_embeddings
(self)
examples/bitnet-1.58b/modeling_bitnet.py:1142
Methodget_input_embeddings
(self)
examples/bitnet-1.58b/modeling_bitnet.py:1492
Methodget_input_embeddings
(self)
examples/bitnet-1.58b/modeling_bitnet.py:1609
Methodget_input_exprs
(self, output_exprs)
tilelang/carver/roller/shape_inference/common.py:51
Methodget_input_exprs
(self, output_exprs)
tilelang/carver/roller/shape_inference/tir.py:280
Methodget_input_tensors_supply
(with_output: bool)
tilelang/autotuner/tuner.py:724
Methodget_ir
(self)
tilelang/carver/roller/node.py:172
Methodget_ir
(self)
tilelang/carver/roller/node.py:186
Methodget_ir
(self)
tilelang/carver/roller/node.py:519
Methodget_kernel_launch_code
(self, function_name, grid_str, block_str, smem_str, call_args, cluster_dims)
tilelang/jit/adapter/wrapper.py:708
Methodget_kernel_source
Return a fake device source.
testing/python/autotune/test_tilelang_autotune_cutedsl_cache.py:35
Methodget_kernel_source
(self, kernel_only: bool = True)
tilelang/jit/adapter/base.py:102
Methodget_kernel_source
Returns the source code of the compiled kernel.
tilelang/jit/adapter/tvm_ffi.py:325
Methodget_kernel_source
(self, kernel_only: bool = True)
tilelang/jit/adapter/torch/metal.py:56
Methodget_kernel_source
Get the CUDA kernel source code. Returns ------- Optional[str] The kernel source code, or None if not available
tilelang/jit/adapter/nvrtc/adapter.py:197
Functionget_kwarg
(**kwargs)
testing/python/language/test_tilelang_language_eager_ast_signature.py:24
Functionget_ladder_stage3_map
(dtype="float16", index_dtype="int32")
tilelang/carver/matmul_analysis.py:743
Functionget_lane_idx
Return the logical lane index of the calling thread within a warp. Parameters ---------- warp_size : Optional[int, PrimExpr] Logi
tilelang/language/builtin.py:562
Methodget_lib_path
(self)
tilelang/jit/adapter/libgen.py:210
Functionget_max_shared_memory_per_block
(target: Target)
tilelang/carver/analysis.py:193
Functionget_max_threads_per_block
(target: Target)
tilelang/carver/analysis.py:182
Functionget_mma_micro_size
Return the MMA (Tensor Core) micro-tile dimensions for a given data type. This function returns the micro tile sizes (x, y, k) used by MMA/T
tilelang/cuda/intrinsics/layout/utils.py:90
Functionget_multiprocessor_count
Compatible across Triton versions: - 2.0.x - 2.1.0 - 2.2.x and above Supports CUDA and NPU.
examples/kda/FLA_KDA/fla_utils.py:126
Methodget_name
(self)
tilelang/autodd.py:108
Methodget_name
(self)
tilelang/autodd.py:130
Methodget_name
(self)
tilelang/autodd.py:145
Methodget_name
(self)
tilelang/autodd.py:171
Methodget_name
(self)
tilelang/autodd.py:189
Methodget_name
(self)
tilelang/autodd.py:349
Functionget_output_blocks
Get the output blocks of a schedule. Parameters ---------- sch : tirx.Schedule The TIR schedule used to get output blocks. bl
tilelang/carver/common_schedules.py:59
Functionget_pass_configs
()
examples/flash_decoding/example_gqa_decode.py:44
Functionget_pass_context
Get the current pass context
tilelang/transform/__init__.py:14
Methodget_peak_tflops
Get the peak TFLOPS for the target device. Args: device: The target device information. Returns:
tilelang/tools/Analyzer.py:171
Methodget_producer_blocks
(self, block: SBlockRV)
tilelang/carver/roller/node.py:80
Functionget_propagate_map
(trans: bool = True, dtype="float16", matrix_name="A", index_dtype="int32")
tilelang/carver/matmul_analysis.py:684
Functionget_ptx_from_source
Compile CUDA C++ source to PTX using NVCC and return as text. Parameters ---------- code : str CUDA C++ kernel source code.
tilelang/contrib/nvcc.py:201
Functionget_rasterization_code
(pannel_width: int = 8)
tilelang/carver/utils.py:14
Functionget_sass_from_source
Compile CUDA C++ source to CUBIN and disassemble to SASS. Uses nvdisasm if available; otherwise falls back to cuobjdump. Parameters
tilelang/contrib/nvcc.py:239
Methodget_source_path
(self)
tilelang/jit/adapter/libgen.py:207
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
benchmark/blocksparse_attention/benchmark_torch_block_sparse_fmha.py:21
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
benchmark/blocksparse_attention/benchmark_triton_block_sparse_fmha.py:26
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
benchmark/blocksparse_attention/benchmark_library_dense_fmha.py:18
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
benchmark/blocksparse_attention/benchmark_tilelang_block_sparse_fmha.py:26
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
examples/blocksparse_attention/example_tilelang_block_sparse_attn.py:21
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
examples/blocksparse_attention/block_sparse_attn_triton.py:25
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
examples/seer_attention/block_sparse_attn_tilelang.py:21
Functionget_sparse_attn_mask_from_threshold
(x, threshold, use_dense_for_last_block=False)
examples/seer_attention/block_sparse_attn_triton.py:26
Functionget_sparse_attn_mask_from_topk
(x, topk, use_dense_for_last_block=False)
benchmark/blocksparse_attention/benchmark_library_dense_fmha.py:6
Methodget_special_tokens_mask
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding special tokens using the
examples/bitnet-1.58b/tokenization_bitnet.py:351
Methodget_stream_type
(self)
tilelang/jit/adapter/wrapper.py:724
Functionget_target_triple
Get target triple according to dumpmachine option of compiler.
tilelang/contrib/cc.py:277
Functionget_tensor
(param: KernelParam)
tilelang/utils/tensor.py:46
Functionget_thread_binding
Returns the thread binding for the given dimension.
tilelang/language/kernel.py:439
Functionget_thread_extent
Returns the thread extent for the given dimension.
tilelang/language/kernel.py:463
Functionget_thread_template_configs
()
examples/gemv/example_gemv.py:256
Methodget_tir
(self, tensor_args, given_tensor_args, kwargs)
tilelang/language/eager/builder.py:1097
Methodget_tma_atom
(tensor, tiler_mk, stages=1)
tilelang/contrib/cutedsl/gemm_v1.py:445
Functionget_tokenizer_pool_config
(tokenizer_group_type)
examples/bitnet-1.58b/vllm_workspace/conftest.py:557
Methodget_tuner_result
Gets the tuning results for this kernel. Returns ------- Dict[str, Any] A dictionary containing:
tilelang/jit/kernel.py:596
Methodget_value
Retrieve the value associated with a variable. Args: var: The variable to look up Returns: The value associa
tilelang/language/frame.py:51
Methodget_vocab
Returns vocab as a dict
examples/bitnet-1.58b/tokenization_bitnet.py:238
Functionget_warp_group_idx
Return the canonical warp group index for the calling thread. Parameters ---------- warp_size : Optional[int, PrimExpr] Logical w
tilelang/language/builtin.py:641
Functionget_warp_idx
Return the canonical warp index without synchronizing the warp. Parameters ---------- warp_size : Optional[int, PrimExpr] Logical
tilelang/language/builtin.py:615
Functionget_warp_idx_sync
Return the canonical warp index, assuming the warp's threads are converged. Parameters ---------- warp_size : Optional[int, PrimExpr]
tilelang/language/builtin.py:589
Functiongrid
(meta)
examples/kda/FLA_KDA/fla_chunk_delta.py:499
Functiongrid
(meta)
examples/kda/FLA_KDA/fla_chunk_inter.py:168
Functiongrid
(meta)
examples/kda/FLA_KDA/cumsum.py:305
Functiongrid
(meta)
examples/kda/FLA_KDA/fla_chunk_o.py:320
Functiongrid
(meta)
examples/kda/FLA_KDA/fla_chunk_intra_token_parallel.py:149
Functiongrid
The grid For statement. Parameters ---------- extents : PrimExpr The extents of the iteration. Returns ------- res :
tilelang/language/ast/ir.py:836
Functiongrid
The grid For statement. Parameters ---------- extents : PrimExpr The extents of the iteration. Returns ------- res :
tilelang/language/tir/ir.py:136
Functiongrouped_mxfp8_blockscaled_gemm_2cta
Grouped 2CTA MXFP8 blockscaled GEMM. Logical scale shape follows tilelang_gemm.py: SFA [M_total, sf_k_packed], SFB [E, N, sf_k_packed]
examples/blockscaled_gemm_sm100/grouped_gemm_mxfp8_blockscaled_1d1d.py:14
Functiongrouped_mxfp8_blockscaled_gemm_2cta_persistent
Persistent grouped 2CTA MXFP8 blockscaled GEMM with one accumulator TMEM.
examples/blockscaled_gemm_sm100/grouped_gemm_mxfp8_blockscaled_1d1d.py:196
Functionguarded_copy_loop
( x: T.Tensor((M, K), T.float16), y: T.Tensor((M, K), T.float16), valid_m: T.int32,
testing/python/issue/test_tilelang_issue_tma_no_ws.py:124
Functionhandle_add_byte_offset
(handle, byte_offset)
tilelang/contrib/cutedsl/utils.py:116
Methodhas_arch
Checks whether the architecture is set. Returns: bool: True if the architecture is set, False otherwise.
tilelang/carver/template/base.py:63
Functionhas_mma_support
(arch: TileDevice)
tilelang/carver/arch/cuda.py:46
Methodhas_value
Check if a variable has an associated value. Args: var: The variable to check Returns: bool: True if the var
tilelang/language/frame.py:70
Functionhave_bf16
Either bf16 support is provided in the compute capability or not Parameters ---------- compute_version : str compute capability o
tilelang/contrib/nvcc.py:534
Functionhave_cudagraph
Either CUDA Graph support is provided
tilelang/contrib/nvcc.py:524
Functionhave_fp16
Either fp16 support is provided in the compute capability or not Parameters ---------- compute_version: str compute capability of
tilelang/contrib/nvcc.py:464
Functionhave_fp8
Whether fp8 support is provided in the specified compute capability or not Parameters ---------- compute_version : str GPU capabi
tilelang/contrib/nvcc.py:547
Functionhave_int8
Either int8 support is provided in the compute capability or not Parameters ---------- compute_version : str compute capability o
tilelang/contrib/nvcc.py:481
Functionhave_matrixcore
Either MatrixCore support is provided in the compute capability or not Parameters ---------- compute_version : str, optional comp
tilelang/contrib/rocm.py:204
Functionhave_tensorcore
Either TensorCore support is provided in the compute capability or not Parameters ---------- compute_version : str, optional comp
tilelang/contrib/nvcc.py:493
Functionhelion_mamba2_chunk_scan_kernel
Argument: cb: (batch, nchunks, ngroups, chunk_size, chunk_size) x: (batch, seqlen, nheads, headdim) dt: (
benchmark/mamba2/benchmark_mamba_chunk_scan.py:79
Functionhf_runner
()
examples/bitnet-1.58b/vllm_workspace/conftest.py:407
Methodhost_func
(self)
tilelang/jit/adapter/wrapper.py:651
Methodhost_func
Override parent's host_func to return generated Python code.
tilelang/jit/adapter/nvrtc/wrapper.py:273
Methodhost_func
Override parent's host_func to return generated Python code.
tilelang/jit/adapter/cutedsl/wrapper.py:804
Methodhost_source
(self)
tilelang/jit/kernel.py:635
Functionhypot
Equivalent to sqrt(x1**2 + x2**2), element-wise. Parameters ---------- x1 : PrimExpr Input argument. x2 : PrimExpr I
tilelang/language/tir/op.py:2875
← previousnext →4,701–4,800 of 7,254, ranked by callers