MCPcopy Create free account

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

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

Functiongemm_kernel
( A: T.Tensor((M, K), dtype), B: T.Tensor((K, N), dtype), C: T.Tensor((M, N), accum_dt
testing/python/metal/test_metal_gemm_v2.py:17
Functiongemm_kernel
( A: T.Tensor((M, K), dtype), B: T.Tensor((K, N), dtype), C: T.Tensor((M, N), accum_dt
testing/python/metal/test_metal_simdgroup_store.py:18
Functiongemm_lower
( gemm, layout_map, target: Target, thread_bounds: Range, thread_var: tirx.Var, mbar_p
tilelang/tileop/gemm/__init__.py:19
Functiongemm_persistent
( A, B, block_M, block_N, store_block_N, # block_N for C_shared block_K, in_dtype
examples/gemm_sm100/gemm_tcgen5mma_ws_persistent.py:11
Functiongemm_persistent_2cta
( A, B, block_M, block_N, store_block_N, # block_N for C_shared block_K, in_dtype
examples/gemm_sm100/gemm_tcgen5mma_ws_persistent.py:134
Functiongemm_rr
GEMM with both A and B from register/fragment
tilelang/contrib/cutedsl/gemm_v1.py:189
Functiongemm_rs
GEMM with A from register/fragment and B from shared memory
tilelang/contrib/cutedsl/gemm_v1.py:88
Functiongemm_sp
TileLang sparse GEMM operator. This is the default synchronous sparse GEMM interface. On Hopper, if the compiler selects WGMMA SP lowering, T
tilelang/language/experimental/gemm_sp_op.py:119
Functiongemm_sp_fp16
( A_sparse: T.Tensor((M, K // 2), T.float16), E: T.Tensor((M, K // e_factor), e_dtype),
examples/gemm_sp/example_gemm_sp.py:17
Methodgemm_sp_infer_layout
(self, target: Target, thread_bounds: Range)
tilelang/tileop/gemm_sp/__init__.py:47
Methodgemm_sp_lower
(self, target: Target, layout_map: dict, thread_bounds: Range, thread_var: tirx.Var)
tilelang/tileop/gemm_sp/__init__.py:52
Functiongemm_sr
GEMM with A from shared memory and B from register/fragment
tilelang/contrib/cutedsl/gemm_v1.py:148
Functiongemm_ss
GEMM with both A and B from shared memory
tilelang/contrib/cutedsl/gemm_v1.py:28
Functiongemm_swizzled
( A: T.Buffer((M, K), "float16"), B: T.Buffer((K, N), "float16"), C: T.Buffer((M, N),
testing/python/transform/test_tilelang_transform_producer_consumer_ws.py:450
Functiongemv_alloc_reducer
( a, x, block_M=128, block_N=128, num_stages=2, threads=256, dtype: T.dtype = T.float16, accum_dtype: T.dt
examples/gemv/example_gemv.py:230
Functiongen_quant4
(k, n, groupsize=-1)
tilelang/quantize/utils.py:1
Functiongeneral_compress
(lowprecision_weight, source_bits=4, storage_dtype=None)
tilelang/quantize/utils.py:50
Methodgenerate_beam_search
( self, prompts: List[str], beam_width: int, max_tokens: int, )
examples/bitnet-1.58b/vllm_workspace/conftest.py:265
Methodgenerate_beam_search
( self, prompts: List[str], beam_width: int, max_tokens: int, )
examples/bitnet-1.58b/vllm_workspace/conftest.py:521
Methodgenerate_greedy
( self, prompts: List[str], max_tokens: int, images: Optional[List[Image.Image
examples/bitnet-1.58b/vllm_workspace/conftest.py:499
Methodgenerate_greedy_logprobs
( self, prompts: List[str], max_tokens: int, images: Optional[List[Image.Image
examples/bitnet-1.58b/vllm_workspace/conftest.py:285
Methodgenerate_greedy_logprobs
( self, prompts: List[str], max_tokens: int, num_logprobs: int, images
examples/bitnet-1.58b/vllm_workspace/conftest.py:509
Methodgenerate_greedy_logprobs_limit
( self, prompts: List[str], max_tokens: int, num_logprobs: int, images
examples/bitnet-1.58b/vllm_workspace/conftest.py:326
Functiongenerate_random_cu_seqlens
(per_cp_seqlen, cp_size=4, cp_rank=3, kv_stride=1, average_q_len=512)
examples/dsa_hisa/tilelang_utils.py:261
Functionget_abs_err
(y, x)
examples/dsa_hisa/tilelang_utils.py:142
Functionget_ast
(func: Callable)
tilelang/language/eager/utils.py:59
Functionget_autotuned_kernel
( A, B, BLOCK_N=None, reduce_threads=None, )
examples/gemv/example_gemv.py:267
Methodget_avaliable_tensorintrin_shapes
(self)
tilelang/carver/arch/cuda.py:158
Methodget_avaliable_tensorintrin_shapes
(self)
tilelang/carver/arch/arch_base.py:29
Functionget_block
Get the target block from a schedule. Parameters ---------- sch : tirx.Schedule The TIR schedule used to get target block. na
tilelang/carver/common_schedules.py:31
Functionget_block_binding
Returns the block binding for the given dimension.
tilelang/language/kernel.py:451
Functionget_block_bindings
Returns all three block bindings.
tilelang/language/kernel.py:457
Functionget_block_extent
Returns the block extent for the given dimension.
tilelang/language/kernel.py:475
Methodget_block_size
Determines the optimal block size for a given constraint, based on scoring various factors. Parameters ---------- n
tilelang/carver/roller/policy/default.py:221
Functionget_block_template_configs
()
examples/gemv/example_gemv.py:212
Functionget_bwd_configs
()
examples/amd/example_amd_flash_attn_bwd.py:245
Methodget_cl_shapes
(c_ax_m, c_ax_n, num_nvalid_regions)
tilelang/carver/roller/node.py:424
Functionget_coalesced_veclen
(block_stmt: tirx.SBlock, target_bits: int = 128)
tilelang/carver/analysis.py:287
Methodget_code
(self)
tilelang/carver/roller/rasterization.py:10
Methodget_code
(self)
tilelang/carver/roller/rasterization.py:26
Methodget_code
(self)
tilelang/carver/roller/rasterization.py:46
Methodget_code
(self, panel_width: int = None)
tilelang/carver/roller/rasterization.py:83
Methodget_compile_func
()
tilelang/autotuner/tuner.py:558
Functionget_compiled_object
(source: str | ast.AST, name: str, filename: str = None, globals: dict[str, Any] = None)
tilelang/language/eager/utils.py:72
Functionget_configs
()
benchmark/mamba2/benchmark_mamba_chunk_scan.py:178
Functionget_configs
Generate a list of configuration dictionaries that will be used for tuning. Parameters ---------- with_roller : bool Whether
benchmark/matmul/benchmark_matmul_sp.py:35
Functionget_configs
Generate a list of configuration dictionaries that will be used for tuning. Parameters ---------- with_roller : bool Whether
benchmark/matmul/benchmark_matmul.py:33
Functionget_configs
Generate a list of configuration dictionaries that will be used for tuning. Parameters ---------- with_roller : bool Whether
benchmark/matmul_fp8/benchmark_matmul.py:33
Functionget_configs
()
examples/deepseek_nsa/benchmark/benchmark_nsa_fwd.py:449
Functionget_configs
()
examples/blocksparse_gemm/example_blocksparse_gemm.py:18
Functionget_configs
()
examples/deepseek_v32/fp8_lighting_indexer.py:51
Functionget_configs
()
examples/linear_attention/example_mamba_chunk_state.py:48
Functionget_configs
()
examples/linear_attention/example_mamba_chunk_scan.py:65
Functionget_configs
()
examples/convolution/example_convolution_autotune.py:27
Functionget_configs
()
examples/dequantize_gemm/example_dequant_gemm_bf16_mxfp4_cdna4.py:49
Functionget_configs
Generate a list of hyperparameter configuration dictionaries for tuning. Each configuration is a dict with keys: 'block_M', 'block_N', 'bloc
examples/dequantize_gemm/example_dequant_gemm_bf16_mxfp4_hopper.py:52
Functionget_configs
()
examples/dequantize_gemm/example_dequant_gemm_fp4_hopper.py:95
Functionget_configs
Return a list of tuning configuration dictionaries for the autotuned matmul kernel. Each dictionary is a single combination (Cartesian produ
examples/dequantize_gemm/example_dequant_gemm_bf16_fp4_hopper.py:10
Functionget_configs
()
examples/dequantize_gemm/example_dequant_gemm_w4a8.py:24
Functionget_configs
Generates configurations for the autotuner. For RDNA (gfx11xx/gfx12xx) GPUs using WMMA instructions, block sizes are limited to 32x32 due to
examples/amd/example_amd_flash_attn_fwd.py:55
Functionget_configs
()
examples/flash_decoding/example_gqa_decode_varlen_logits.py:23
Functionget_configs
()
examples/flash_decoding/example_gqa_decode.py:15
Functionget_configs
()
examples/flash_attention/example_mha_fwd_bshd.py:11
Functionget_configs
(user_config=None)
examples/flash_attention/example_gqa_fwd_bshd.py:31
Functionget_configs
()
examples/flash_attention/example_mha_fwd_varlen.py:15
Functionget_configs
()
examples/flash_attention/example_mha_fwd_bhsd.py:11
Functionget_configs
()
examples/attention_sink/example_mha_sink_fwd_bhsd.py:13
Functionget_configs
()
examples/gdn/example_chunk_delta_h.py:87
Functionget_configs
()
examples/topk/example_topk.py:8
Functionget_configs
()
examples/kda/chunk_intra_token_parallel.py:46
Functionget_configs
()
examples/kda/wy_fast_bwd.py:63
Functionget_configs
()
examples/kda/chunk_delta_bwd.py:64
Functionget_configs
()
examples/kda/chunk_delta_h_fwd.py:65
Functionget_configs
()
examples/kda/chunk_bwd_dqkwg.py:53
Functionget_configs
()
examples/kda/wy_fast.py:41
Functionget_configs
()
examples/kda/chunk_o.py:47
Functionget_configs
()
examples/kda/chunk_bwd_intra.py:66
Functionget_configs
()
examples/kda/chunk_bwd_gla_dA.py:39
Functionget_configs
()
examples/kda/chunk_bwd_dv.py:43
Functionget_configs
()
examples/deepseek_mla/amd/benchmark_mla_decode_amd_tilelang.py:9
Functionget_configs
()
examples/gemm_fp8/example_tilelang_gemm_amd_fp8_preshuffle.py:15
Functionget_configs
()
examples/gemm_fp8/example_tilelang_gemm_amd.py:27
Functionget_configs
()
testing/python/autotune/test_tilelang_autotune_with_inputs.py:32
Functionget_configs
()
testing/python/autotune/test_tilelang_autotune_do_not_specialize.py:16
Functionget_configs
()
testing/python/autotune/test_tilelang_autotune_eager_mode.py:32
Methodget_current_stream_functor
Return a callable that reads Torch's current CUDA stream pointer. The returned lambda yields the raw CUDA stream handle of the current
tilelang/jit/adapter/base.py:61
Methodget_declaration
(self, declare_kernel_code: str)
tilelang/jit/adapter/wrapper.py:703
Methodget_decoder
(self)
examples/bitnet-1.58b/modeling_bitnet.py:1157
Functionget_dequantize_block
(sch, blocks)
tilelang/carver/matmul_analysis.py:419
Functionget_dll_directories
()
tilelang/libinfo.py:8
Methodget_elaborate_func
()
tilelang/autotuner/tuner.py:565
Functionget_err_ratio
(y, x)
examples/dsa_hisa/tilelang_utils.py:148
Methodget_forward_index
(self)
tilelang/layout/layout.py:90
Functionget_func_nonlocals
A modified version of `inspect.getclosurevars`
tilelang/language/eager/utils.py:35
Functionget_fused_mapping_kernel
(topk_idx: T.Tensor[(1,), T.int32])
testing/python/transform/test_tilelang_transform_loop_unswitching.py:399
Functionget_fwd_configs
()
examples/amd/example_amd_flash_attn_bwd.py:42
Functionget_global_symbol_section_map
Get global symbols from a library via nm -g Parameters ---------- path : str The library path nm: str The path to nm
tilelang/contrib/cc.py:218
Methodget_hardware_aware_configs
Retrieves optimized hardware-aware configurations. Args: arch (TileDevice, optional): The target hardware architecture.
tilelang/carver/template/gemv.py:27
Methodget_hardware_aware_configs
Retrieves optimized hardware-aware configurations. Args: arch (TileDevice, optional): The target hardware architecture.
tilelang/carver/template/flashattention.py:27
Methodget_hardware_aware_configs
Retrieves optimized hardware-aware configurations. Args: arch (TileDevice, optional): The target hardware architecture.
tilelang/carver/template/matmul.py:40
← previousnext →4,601–4,700 of 7,254, ranked by callers