MCPcopy Create free account

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

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

↓ 1 callersFunctionrun_kernel_only
()
examples/deepseek_mhc/example_mhc_post.py:113
↓ 1 callersFunctionrun_matmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_clear.py:42
↓ 1 callersFunctionrun_matmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_alias.py:46
↓ 1 callersFunctionrun_matmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_ptr.py:112
↓ 1 callersFunctionrun_matmul_float_variable
(M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, num_stages, threads)
testing/python/jit/test_tilelang_jit_gemm_cython.py:478
↓ 1 callersFunctionrun_matmul_int_variable
(M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, num_stages, threads)
testing/python/jit/test_tilelang_jit_gemm_cython.py:407
↓ 1 callersFunctionrun_nvrtc_im2col_tma_desc
Test im2col TMA descriptor functionality in NVRTC backend.
testing/python/jit/test_tilelang_jit_nvrtc.py:353
↓ 1 callersFunctionrun_nvrtc_kernel_do_bench
( M, N, K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, block_M, block_N, block_K, num_stages=3, num
testing/python/jit/test_tilelang_jit_nvrtc.py:176
↓ 1 callersFunctionrun_nvrtc_kernel_multi_stream
( M, N, K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, block_M, block_N, block_K, num_stages=3, num
testing/python/jit/test_tilelang_jit_nvrtc.py:216
↓ 1 callersFunctionrun_pointer_table_copy
(N, dtype=T.float16)
testing/python/language/test_tilelang_language_ptr.py:131
↓ 1 callersFunctionrun_pointer_table_grouped_matmul
(batch_sizes_list, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_ptr.py:166
↓ 1 callersFunctionrun_pointer_table_multi_copy
(G, N, dtype=T.float16)
testing/python/language/test_tilelang_language_ptr.py:148
↓ 1 callersFunctionrun_single_bitwise_reduce
( name, func, a, clear=True, )
testing/python/math/test_math_bitwise_reduce.py:44
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, chunk_size, input_dtype, output_dtype, gate_dtype,
examples/gdn/example_wy_fast.py:134
↓ 1 callersFunctionrun_test
( B, S, H, chunk_size, reverse, head_first, input_dtype, output_dtype, thr
examples/gdn/example_cumsum.py:94
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/gdn/example_wy_fast_bwd_split.py:413
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/gdn/example_chunk_delta_h.py:228
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/gdn/example_chunk_delta_bwd.py:405
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, chunk_size, input_dtype, output_dtype, accum_dtype,
examples/gdn/example_chunk_o.py:153
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/gdn/example_chunk_o_bwd.py:362
↓ 1 callersFunctionrun_test
( B, S, H, DK, chunk_size, input_dtype, output_dtype, accum_dtype, use_g,
examples/gdn/example_chunk_scaled_dot_kkt.py:130
↓ 1 callersFunctionrun_test
( B, S, H, DK, scale, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/kda/chunk_inter_solve_fused.py:481
↓ 1 callersFunctionrun_test
( B, S, H, DK, scale, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/kda/chunk_intra_token_parallel.py:176
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/kda/wy_fast_bwd.py:241
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/kda/chunk_delta_bwd.py:203
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, input_dtype, output_dtype, accum_dtype, gate_dtype,
examples/kda/chunk_delta_h_fwd.py:185
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, scale, input_dtype, gate_dtype, qk_dtype, chunk_siz
examples/kda/chunk_bwd_dqkwg.py:186
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, chunk_size, input_dtype, output_dtype, gate_dtype,
examples/kda/wy_fast.py:145
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, chunk_size, input_dtype, output_dtype, accum_dtype,
examples/kda/chunk_o.py:169
↓ 1 callersFunctionrun_test
( B, S, H, DK, input_dtype, output_dtype, accum_dtype, gate_dtype, state_d
examples/kda/chunk_bwd_intra.py:380
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, scale, input_dtype, do_dtype, da_dtype, chunk_size,
examples/kda/chunk_bwd_gla_dA.py:95
↓ 1 callersFunctionrun_test
( B, S, H, DK, DV, scale, input_dtype, do_dtype, output_dtype, chunk_s
examples/kda/chunk_bwd_dv.py:99
↓ 1 callersFunctionrun_test
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/jit/test_tilelang_jit_nullptr.py:40
↓ 1 callersFunctionrun_tile_atomic_add
(K, M, N, block_M, block_N, dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:334
↓ 1 callersFunctionrun_tile_atomic_add_expr
(M, N, block_M, block_N, dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:355
↓ 1 callersFunctionrun_tile_atomic_add_scalar
(dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:387
↓ 1 callersFunctionrun_tile_atomic_max
(K, M, N, block_M, block_N, dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:538
↓ 1 callersFunctionrun_tile_atomic_max_expr
(M, N, block_M, block_N, dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:596
↓ 1 callersFunctionrun_tile_atomic_min
(K, M, N, block_M, block_N, dtype=T.float32)
testing/python/language/test_tilelang_language_atomic.py:569
↓ 1 callersFunctionrun_tilelang_copy
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16, pad_value=0)
testing/python/language/test_tilelang_language_annotate_safe_value.py:29
↓ 1 callersFunctionrun_tilelang_copy_buffer_load_with_parallel
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_copy.py:124
↓ 1 callersFunctionrun_tilelang_copy_bufferload
(num_tokens=128, dtype=T.float16)
testing/python/language/test_tilelang_language_copy.py:96
↓ 1 callersFunctionrun_tilelang_copy_fp8_e8m0
(M=1024, N=1024, block_M=128, block_N=128, src_dtype=T.float8_e8m0fnu, dst_dtype=T.float8_e8m0fnu)
testing/python/language/test_tilelang_language_copy.py:169
↓ 1 callersFunctionrun_tilelang_copy_mask_copy
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_mask_op.py:63
↓ 1 callersFunctionrun_tilelang_copy_mask_copy_range
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_mask_op.py:132
↓ 1 callersFunctionrun_tilelang_copy_mask_parallel
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_mask_op.py:29
↓ 1 callersFunctionrun_tilelang_copy_mask_parallel_range
(M=1024, N=1024, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_mask_op.py:98
↓ 1 callersFunctionrun_tilelang_copy_shape_mismatched
(M=1024, N=1024, dtype=T.float16)
testing/python/language/test_tilelang_language_copy.py:153
↓ 1 callersFunctionrun_tilelang_grouped_gemm
( batch_sizes_list, K, M, block_M, block_N, block_K, trans_b=False, num_stages
examples/grouped_gemm/example_grouped_gemm_bwd.py:185
↓ 1 callersFunctionrun_tilelang_if_range
(M=128, N=128, block_M=32, block_N=32, dtype=T.float16)
testing/python/language/test_tilelang_language_if_range.py:30
↓ 1 callersFunctionrun_tilelang_mla
(q, block_table, blocked_k, max_seqlen_pad, block_size, b, s_q, cache_seqlens, h_q, h_kv, d, dv, causal, dtype
examples/deepseek_mla/example_mla_decode_paged.py:258
↓ 1 callersFunctionrun_tilelang_ternary
(M=128, N=128, block_M=32, block_N=32, dtype=T.float16)
testing/python/language/test_tilelang_language_ternary.py:24
↓ 1 callersFunctionrun_tilelang_transpose
(M=128, N=128, block_M=128, block_N=128, dtype=T.float16)
testing/python/language/test_tilelang_language_transpose.py:45
↓ 1 callersFunctionrun_tma_store_wait_read_option_codegen
()
testing/python/language/test_tilelang_language_tma_store.py:174
↓ 1 callersFunctionrun_tmp_var_test
(N=1024, block_N=128)
testing/python/issue/test_tilelang_issue_814.py:24
↓ 1 callersFunctionrun_torch_mla
(q, block_table, blocked_k, max_seqlen_pad, block_size, b, s_q, cache_seqlens, h_q, h_kv, d, dv, causal, dtype
examples/deepseek_mla/example_mla_decode_paged.py:229
↓ 1 callersFunctionrun_tvm_ffi_cluster_launch
()
testing/python/language/test_tilelang_language_cluster.py:80
↓ 1 callersFunctionrun_tvm_ffi_im2col_tma_desc
Test im2col TMA descriptor functionality in tvm_ffi backend.
testing/python/jit/test_tilelang_jit_tvm_ffi.py:351
↓ 1 callersFunctionrun_tvm_ffi_kernel_do_bench
( M, N, K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, block_M, block_N, block_K, num_stages=3, num
testing/python/jit/test_tilelang_jit_tvm_ffi.py:175
↓ 1 callersFunctionrun_tvm_ffi_kernel_multi_stream
( M, N, K, trans_A, trans_B, in_dtype, out_dtype, dtypeAccum, block_M, block_N, block_K, num_stages=3, num
testing/python/jit/test_tilelang_jit_tvm_ffi.py:214
↓ 1 callersFunctionrun_two_arg_mathop_test
Test two-argument mathops to ensure they generate non-fastmath CUDA code.
testing/python/fastmath/test_mathops_fastmath.py:98
↓ 1 callersFunctionrun_two_arg_mathop_test
Test two-argument mathops to ensure they generate non-fastmath CUDA code.
testing/python/math/test_math_fast_math.py:98
↓ 1 callersFunctionsample
Samples a token from the logits using temperature scaling. Args: logits (torch.Tensor): The logits tensor for token predictions.
examples/deepseek_v32/inference/generate.py:14
↓ 1 callersFunctionscaled_dot_product_attention
(query, key, value, h_q, h_kv, is_causal=False)
examples/deepseek_mla/example_mla_decode_paged.py:208
↓ 1 callersFunctionscaled_dot_product_attention
(query, key, value, h_q, h_kv, is_causal=False)
examples/deepseek_mla/benchmark_mla.py:15
↓ 1 callersFunctionscaled_dot_product_attention
(query, key, value, h_q, h_kv, is_causal=False)
examples/deepseek_mla/amd/benchmark_mla_decode_amd_aiter.py:20
↓ 1 callersFunctionscaled_dot_product_attention
(query, key, value, h_q, h_kv, is_causal=False)
examples/deepseek_mla/amd/benchmark_mla_decode_amd_triton.py:11
↓ 1 callersFunctionscatter_kern
(src, dst, perm)
testing/python/language/test_tilelang_language_subtype.py:375
↓ 1 callersFunctionserial_count_kernel
()
testing/python/amd/test_tilelang_hip_codegen.py:461
↓ 1 callersFunctionset_log_level
Set the logging level for the module's logger. Args: level (str or int): Can be the string name of the level (e.g., 'INFO') or the actual
tilelang/__init__.py:54
↓ 1 callersMethodset_output_nodes
Sets the output nodes for this template and returns itself. Args: output_nodes (List[OutputNode]): The output nodes to a
tilelang/carver/template/base.py:140
↓ 1 callersMethodset_value
Record a variable binding that is not represented by a scoped frame.
tilelang/language/frame.py:62
↓ 1 callersFunctionshared_16x16_to_local_32x16_layout_A_colmajor_gfx11
Forward: A_T[i=K, j=M] -> (thread=M, local=K%16). ATTN: Here we only reflect (i, j) to the lower-half-lane of threads in a warp.
tilelang/rocm/intrinsics/wmma_layout.py:190
↓ 1 callersFunctionshared_16x16_to_local_32x16_layout_A_gfx11
Forward: A[i=M, j=K] -> (thread=i, local=j%16). ATTN: Here we only reflect (i, j) to the lower-half-lane of threads in a warp.
tilelang/rocm/intrinsics/wmma_layout.py:168
↓ 1 callersFunctionshared_16x16_to_local_32x16_layout_B_colmajor_gfx11
Forward: B_T[i=N, j=K] -> (thread=i, local=j%16). ATTN: Here we only reflect (i, j) to the lower-half-lane of threads in a warp.
tilelang/rocm/intrinsics/wmma_layout.py:234
↓ 1 callersFunctionshared_16x16_to_local_32x16_layout_B_gfx11
Forward: B[i=K, j=N] -> (thread=N, local=K%16). ATTN: Here we only reflect (i, j) to the lower-half-lane of threads in a warp.
tilelang/rocm/intrinsics/wmma_layout.py:212
↓ 1 callersFunctionshared_16x16_to_local_64x4_layout_C
(i, j)
tilelang/rocm/intrinsics/mfma_layout.py:29
↓ 1 callersFunctionshared_16x16_to_mma_a_32x8_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:82
↓ 1 callersFunctionshared_16x16_to_mma_b_32x8_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:91
↓ 1 callersFunctionshared_16x32_to_mma_a_32x16_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:106
↓ 1 callersFunctionshared_16x32_to_mma_b_32x16_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:115
↓ 1 callersFunctionshared_16x8_to_mma_a_32x4_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:57
↓ 1 callersFunctionshared_16x8_to_mma_b_32x4_layout
(i, j)
tilelang/cuda/intrinsics/layout/mma_layout.py:67
↓ 1 callersFunctionshared_symbolic_kernel
Kernel with shared symbolic variable across multiple buffers.
testing/python/language/test_tilelang_language_subtype.py:132
↓ 1 callersFunctionshould_enable_ast_print
(pass_ctx: PassContext | None = None)
tilelang/engine/semantic_check.py:15
↓ 1 callersFunctionshould_enable_layout_visual
(pass_ctx: PassContext | None = None)
tilelang/backend/pass_pipeline/pipeline_utils.py:37
↓ 1 callersFunctionshould_enable_prelower_semantic_check
(pass_ctx: PassContext | None = None)
tilelang/engine/semantic_check.py:9
↓ 1 callersMethodshow_source
Print generated source code to stdout. Parameters ---------- which : Literal["kernel", "host", "both"], optional
tilelang/jit/kernel.py:472
↓ 1 callersFunctionshuffle_weight
( x: torch.Tensor, layout=(16, 32), k_pack=1, is_transpose=False, )
examples/gemm_fp8/example_tilelang_gemm_amd_fp8_preshuffle.py:150
↓ 1 callersMethodsimd_load
Load shared/global memory into simdgroup local buffer.
tilelang/metal/intrinsics/metal_macro_generator.py:226
↓ 1 callersMethodsimd_store
Store simdgroup local buffer to shared/global memory.
tilelang/metal/intrinsics/metal_macro_generator.py:222
↓ 1 callersFunctionsimple_invalid_loop
(dtype: T.dtype = T.bfloat16, accum_dtype: T.dtype = T.float32, num_threads: int = 128)
testing/python/analysis/test_tilelang_fragment_loop_checker.py:8
↓ 1 callersFunctionsinkhorn_forward
(M, iters=20)
examples/deepseek_mhc/example_mhc_bwd.py:21
↓ 1 callersFunctionsinkhorn_normalize_ref
(x: torch.Tensor, repeat: int, eps: float)
examples/deepseek_mhc/example_mhc_pre.py:294
↓ 1 callersFunctionsparse_mla_fwd
( Q, KV, Indices, q_start_index_s, heads, dim, tail_dim, topk, kv_stride,
examples/deepseek_v32/sparse_mla_fwd_seesaw.py:22
↓ 1 callersFunctionsparse_mla_fwd
( Q, KV, Indices, Offsets, TokenIndices, heads, dim, tail_dim, topk, k
examples/dsa_sparse_finetune/sparse_mla_fwd.py:16
↓ 1 callersFunctionsparse_mla_fwd_interface
( q, kv, indices, q_start_index_s, kv_stride, sm_scale=None, is_casual=True, return_kernel=False, print_ke
examples/deepseek_v32/sparse_mla_fwd_pipelined.py:306
↓ 1 callersFunctionsparse_mla_fwd_interface
( q, kv, indices, q_start_index_s, kv_stride, sm_scale=None, is_casual=True, return_kernel=False, print_ke
examples/deepseek_v32/sparse_mla_fwd_seesaw.py:469
↓ 1 callersMethodspatial
The spatial block axis defining function. Parameters ---------- dom : Union[Range, List[PrimExpr], Tuple[PrimExpr]]
tilelang/language/ast/ir.py:536
↓ 1 callersFunctionspeed_benchmark_decode_comparison
Speed benchmark for decode kernel
examples/flash_decoding/example_gqa_decode_varlen_logits.py:226
← previousnext →3,001–3,100 of 7,254, ranked by callers