MCPcopy Create free account

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

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

↓ 1 callersFunctionmake_store_cluster_3d_multi_tma_kernel
3-D slice copy decomposed into D*M tma_store_cluster calls. Buffer shape is [D, M, N_full]; the copy region is [0:D, 0:M, 0:N_tile]. Because
testing/python/cuda/test_tma_dsmem.py:215
↓ 1 callersFunctionmake_store_cluster_kernel
(N: int)
testing/python/cuda/test_tma_dsmem.py:10
↓ 1 callersFunctionmake_store_cluster_simt_barrier_kernel
2-D slice copy that forces the SIMT fallback even though remote_barrier is set. s_src / s_dst are allocated with inner dimension N_full, but only
testing/python/cuda/test_tma_dsmem.py:73
↓ 1 callersFunctionmake_store_cluster_simt_no_barrier_kernel
No remote_barrier -> SIMT fallback always taken; cluster_sync() orders stores.
testing/python/cuda/test_tma_dsmem.py:38
↓ 1 callersFunctionmake_tcgen05_atom_kernel
(M, N, K, in_dtype, out_dtype, accum_dtype)
testing/python/language/test_tilelang_language_atom_mma.py:245
↓ 1 callersFunctionmake_tensor_at_offset
(ptr: cute.Pointer, offset, shape, div_by=1)
tilelang/contrib/cutedsl/utils.py:103
↓ 1 callersFunctionmake_tensor_from_addr
( addr: PrimExpr, shape: ShapeType, dtype: DType = "float32", strides: tuple[PrimExpr, ...] |
tilelang/language/proxy.py:310
↓ 1 callersFunctionmake_tilelang_binary_kernel
Generate TileLang binary operation kernel (division).
maint/precision/compare_ops.py:222
↓ 1 callersFunctionmake_tilelang_unary_kernel
Generate TileLang unary operation kernel.
maint/precision/compare_ops.py:184
↓ 1 callersMethodmake_tma_atom
( tensor, smem_layout_staged, smem_tile, mcast_dim, )
tilelang/contrib/cutedsl/gemm_v1.py:424
↓ 1 callersFunctionmake_tma_multicast_demo_kernel
(M, N, block_M, block_N, cluster_mask)
testing/python/cuda/test_tma_multicast_demo.py:9
↓ 1 callersFunctionmake_wgmma_atom_kernel
(M, N, K, in_dtype, out_dtype, accum_dtype)
testing/python/language/test_tilelang_language_atom_mma.py:141
↓ 1 callersMethodmanual_assert_close
Validates kernel output against a reference implementation. Args: reference_program: Reference implementation to compare against
tilelang/profiler/__init__.py:164
↓ 1 callersFunctionmanual_check_prog
(C, C_ref)
examples/gemm_fp8/example_tilelang_gemm_amd.py:13
↓ 1 callersFunctionmatch_declare_kernel_cpu
(source: str, annotation: str = "int32_t")
tilelang/jit/adapter/utils.py:87
↓ 1 callersFunctionmatmu_jit_kernel
(M, N, K, block_M, block_N, block_K)
testing/python/issue/test_tilelang_issue_1697.py:7
↓ 1 callersFunctionmatmu_jit_kernel
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_tvm_ffi.py:55
↓ 1 callersFunctionmatmu_jit_kernel
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_gemm_cython.py:55
↓ 1 callersFunctionmatmu_jit_kernel
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_callback.py:125
↓ 1 callersFunctionmatmu_jit_kernel
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_nvrtc.py:55
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation_tcgen05.py:8
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation_sm70.py:8
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation.py:9
↓ 1 callersFunctionmatmul
Create an autotuned matrix multiplication kernel for matrices of shape: - A: (M, K) - B: (N, K) - C: (M, N) Parameters
benchmark/matmul/benchmark_matmul.py:133
↓ 1 callersFunctionmatmul
Create an autotuned matrix multiplication kernel for matrices of shape: - A: (M, K) - B: (N, K) - C: (M, N) Parameters
benchmark/matmul_fp8/benchmark_matmul.py:119
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, in_dtype, out_dtype, accum_dtype,
examples/dequantize_gemm/example_dequant_gemm_fine_grained.py:17
↓ 1 callersFunctionmatmul
(A, B, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
examples/visual_layout_inference/visual_layout_inference.py:12
↓ 1 callersFunctionmatmul
(A, B, block_M, block_N, block_K, dtype, accum_dtype=T.float32)
examples/gemm_fp8/example_tilelang_gemm_fp8_2xAcc.py:8
↓ 1 callersFunctionmatmul
(A, B, block_M, block_N, block_K, dtype, accum_dtype=T.float32)
examples/gemm_fp8/example_tilelang_gemm_fp8.py:15
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/components/test_tilelang_pass_config_disable_warp_specialized.py:6
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/utils/test_compress_utils.py:12
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/profiler/test_tilelang_profiler.py:7
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/language/test_tilelang_language_pipeline.py:6
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_clear.py:7
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_alias.py:7
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/cpu/test_tilelang_cpu_gemm.py:8
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/amd/test_tilelang_test_amd.py:8
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/issue/test_tilelang_issue_96.py:7
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_callback.py:9
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float32, accum_dtype=T.float32)
testing/python/metal/test_metal_codegen_linux.py:13
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float32, accum_dtype=T.float32)
testing/python/metal/test_metal_codegen.py:9
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/webgpu/test_webgpu_codegen.py:7
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/tilelibrary/test_tilelang_tilelibrary_gemm_sp.py:10
↓ 1 callersFunctionmatmul
(M, N, K, block_M=128, block_N=128, block_K=32, num_stages=0, thread_num=128, enable_rasterization=False)
testing/python/autotune/test_tilelang_autotune_with_inputs.py:41
↓ 1 callersFunctionmatmul
Create an autotuned matrix multiplication kernel for matrices of shape: - A: (M, K) - B: (N, K) - C: (M, N) Parameters
testing/python/autotune/test_tilelang_autotune.py:60
↓ 1 callersFunctionmatmul
(A, B, block_M=128, block_N=128, block_K=32, num_stages=0, thread_num=128, enable_rasterization=False)
testing/python/autotune/test_tilelang_autotune_eager_mode.py:41
↓ 1 callersFunctionmatmul
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/kernel/test_tilelang_kernel_gemm.py:6
↓ 1 callersFunctionmatmul
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/kernel/test_tilelang_kernel_gemm_with_stride.py:7
↓ 1 callersFunctionmatmul_2cta
( M, N, K, block_M, block_N, block_K, in_dtype, out_dtype, accum_dtype,
maint/gemm/correctness_evaluation_tcgen05_2cta.py:9
↓ 1 callersFunctionmatmul_batched
( batch, M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtyp
testing/python/kernel/test_tilelang_kernel_gemm_batched.py:6
↓ 1 callersFunctionmatmul_dynamic
(M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_dtype, accum_dtype, num_stages, threads)
examples/dynamic_shape/example_dynamic.py:54
↓ 1 callersFunctionmatmul_float_variable
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_gemm_cython.py:429
↓ 1 callersFunctionmatmul_gemm_v2
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/metal/test_metal_gemm_v2.py:15
↓ 1 callersFunctionmatmul_gemm_v2
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/metal/test_metal_gemm_v2_linux.py:14
↓ 1 callersFunctionmatmul_int_variable
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_gemm_cython.py:358
↓ 1 callersFunctionmatmul_jit_kernel
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_cutedsl.py:68
↓ 1 callersFunctionmatmul_jit_test
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/cpu/test_tilelang_cpu_gemm.py:64
↓ 1 callersFunctionmatmul_kernel_jit
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/jit/test_tilelang_jit_gemm.py:10
↓ 1 callersFunctionmatmul_nested_papipa
( M, N, K, block_M, block_N, block_K, in_dtype, out_dtype, accum_dtype,
testing/python/analysis/test_tilelang_nested_loop_checker.py:407
↓ 1 callersFunctionmatmul_nested_pipelines
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_dtype, accum_dtype, threads, extra_p
testing/python/analysis/test_tilelang_nested_loop_checker.py:102
↓ 1 callersFunctionmatmul_nt
(M, N, K, bM, bN, bK, in_dtype, out_dtype, accum_dtype)
testing/python/kernel/test_tilelang_kernel_int8_gemm_tcgen5.py:13
↓ 1 callersFunctionmatmul_nt_int4
(M, N, K, block_M, block_N, block_K, threads=128)
examples/gemm_int4/example_tilelang_gemm_int4.py:17
↓ 1 callersFunctionmatmul_relu_kernel
( A: T.Tensor((M, K), T.float16), B: T.Tensor((K, N), T.float16), C: T.Tensor((M, N),
testing/python/language/test_tilelang_language_async_copy_gemm_sm80.py:26
↓ 1 callersFunctionmatmul_rr
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation.py:297
↓ 1 callersFunctionmatmul_rr
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/tilelibrary/test_tilelang_tilelibrary_gemm_sp.py:526
↓ 1 callersFunctionmatmul_rs
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation_sm70.py:127
↓ 1 callersFunctionmatmul_rs
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation.py:128
↓ 1 callersFunctionmatmul_rs
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/amd/test_tilelang_test_amd.py:147
↓ 1 callersFunctionmatmul_rs
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/tilelibrary/test_tilelang_tilelibrary_gemm_sp.py:189
↓ 1 callersFunctionmatmul_rs
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/kernel/test_tilelang_kernel_gemm.py:431
↓ 1 callersFunctionmatmul_sp
Create an autotuned matrix multiplication kernel for matrices of shape: - A: (M, K) - B: (K, N) - C: (M, N) Parameters
benchmark/matmul/benchmark_matmul_sp.py:84
↓ 1 callersFunctionmatmul_sp_fp16
(M, N, K, accum_dtype, e_dtype, block_M, block_N, block_K, num_stages, thread_num, policy, enable_rasterizatio
examples/gemm_sp/example_gemm_sp.py:13
↓ 1 callersFunctionmatmul_sr
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
maint/gemm/correctness_evaluation.py:212
↓ 1 callersFunctionmatmul_sr
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/tilelibrary/test_tilelang_tilelibrary_gemm_sp.py:357
↓ 1 callersFunctionmatmul_sr
( M, N, K, block_M, block_N, block_K, trans_A, trans_B, in_dtype, out_
testing/python/kernel/test_tilelang_kernel_gemm.py:311
↓ 1 callersFunctionmatmul_ss
SS GEMM baseline: verifies tcgen05.ld + cast pipeline works.
testing/python/kernel/test_tilelang_kernel_bf16_gemm_tcgen5_ts.py:17
↓ 1 callersFunctionmatmul_test
(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32)
testing/python/language/test_tilelang_language_ptr.py:9
↓ 1 callersFunctionmatmul_tma_copy
( M, N, K, block_M, block_N, block_K, in_dtype, out_dtype, accum_dtype,
testing/python/language/test_tilelang_language_tma_copy.py:23
↓ 1 callersFunctionmatmul_tma_copy_store
GEMM using T.tma_copy for both load (global->shared) and store (shared->global).
testing/python/language/test_tilelang_language_tma_copy.py:180
↓ 1 callersFunctionmatmul_tma_store
GEMM with T.copy for loads and T.tma_copy for the final store (shared -> global).
testing/python/language/test_tilelang_language_tma_store.py:18
↓ 1 callersFunctionmatmul_windowed_pipelined
A pipelined GEMM whose K-loop has a dynamic lower bound.
testing/python/transform/test_tilelang_transform_producer_consumer_ws.py:40
↓ 1 callersFunctionmatmul_with_parallel
( M, N, K, block_M, block_N, block_K, in_dtype, out_dtype, accum_dtype,
testing/python/analysis/test_tilelang_nested_loop_checker.py:526
↓ 1 callersFunctionmay_create_cyclic_reference
()
testing/python/language/test_tilelang_memory_leak.py:60
↓ 1 callersFunctionmaybe_desc_name
Check if a parameter name corresponds to a TMA descriptor. Args: name: The parameter name to check. matches: List of all mat
tilelang/jit/adapter/utils.py:288
↓ 1 callersFunctionmbarrier_arrive
Arrive at memory barrier. Args: mbarrier: BarrierType The memory barrier to arrive at cta_id: int | Var | None
tilelang/language/builtin.py:489
↓ 1 callersFunctionmbarrier_wait_parity
Wait for memory barrier parity condition. Args: mbarrier: BarrierType The memory barrier to wait on parity: int |
tilelang/language/builtin.py:448
↓ 1 callersFunctionmerge_if_test
()
testing/python/issue/test_tilelang_issue_merge_if.py:8
↓ 1 callersFunctionmetadata_16bit_load_32x2_to_shared_16x2_layout_32bit
(thread_id: int, local_id: int)
tilelang/cuda/intrinsics/layout/mma_sp_layout.py:82
↓ 1 callersFunctionmhc_post
( x: torch.Tensor, residual: torch.Tensor, post_layer_mix: torch.Tensor, comb_res_mix: torch.T
examples/deepseek_mhc/example_mhc_post.py:52
↓ 1 callersFunctionmhc_post_ref
( x: torch.Tensor, residual: torch.Tensor, post_layer_mix: torch.Tensor, comb_res_mix: torch.T
examples/deepseek_mhc/example_mhc_post.py:68
↓ 1 callersFunctionmhc_pre
Forward pass for mHC pre block. Args: residual: shape (..., hc_mult, hidden_size), dtype torch.bfloat16 fn: shape (hc_mult3,
examples/deepseek_mhc/example_mhc_pre.py:196
↓ 1 callersFunctionmhc_pre_ref
( residual: torch.Tensor, fn: torch.Tensor, hc_scale: torch.Tensor, hc_base: torch.Tensor,
examples/deepseek_mhc/example_mhc_pre.py:303
↓ 1 callersFunctionmla_aiter
()
examples/deepseek_mla/amd/benchmark_mla_decode_amd_aiter.py:97
↓ 1 callersFunctionmla_decode_triton
( q_nope, q_pe, kv_c_cache, k_pe_cache, o, req_to_tokens, b_seq_len, attn_logi
examples/deepseek_mla/benchmark_mla.py:349
↓ 1 callersFunctionmla_decode_triton
( q_nope, q_pe, kv_c_cache, k_pe_cache, o, req_to_tokens, b_seq_len, attn_logi
examples/deepseek_mla/amd/benchmark_mla_decode_amd_triton.py:270
↓ 1 callersFunctionmma_load_a_32x16_to_shared_16x32_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:148
↓ 1 callersFunctionmma_load_a_32x4_to_shared_16x8_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:136
↓ 1 callersFunctionmma_load_a_32x8_to_shared_16x16_layout
groupID = %laneid >> 2 threadID_in_group = %laneid % 4 row = groupID for ai where 0 <= i < 2 || 4 <= i < 6
tilelang/cuda/intrinsics/layout/mma_layout.py:154
↓ 1 callersFunctionmma_store_32x2_to_shared_8x8_layout_fp64
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:48
↓ 1 callersFunctionmma_store_32x8_to_shared_16x16_layout
(thread_id, local_id)
tilelang/cuda/intrinsics/layout/mma_layout.py:42
← previousnext →2,601–2,700 of 7,254, ranked by callers