MCPcopy Create free account

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

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

↓ 1 callersFunctionmain
( Q_unpad: T.Tensor(q_shape, dtype), K_unpad: T.Tensor(kv_shape, dtype), V_unpad: T.Te
examples/flash_attention/example_gqa_fwd_varlen.py:27
↓ 1 callersFunctionmain
( Q_unpad: T.Tensor(q_shape, dtype), K_unpad: T.Tensor(k_shape, dtype), V_unpad: T.Ten
examples/flash_attention/example_mha_fwd_varlen.py:38
↓ 1 callersFunctionmain
( BATCH: int = 1, H: int = 32, N_CTX: int = 256, D_HEAD_QK: int = 192, D_HEAD_V: int = 128
examples/flash_attention/example_gqa_bwd_tma_reduce.py:439
↓ 1 callersFunctionmain
( BATCH: int = 1, H: int = 32, N_CTX: int = 256, D_HEAD_QK: int = 192, D_HEAD_V: int = 128
examples/flash_attention/example_gqa_bwd.py:432
↓ 1 callersFunctionmain
( BATCH: int = 8, H: int = 32, N_CTX: int = 1024, D_HEAD: int = 64, causal: bool = False,
examples/flash_attention/example_mha_bwd_bshd.py:298
↓ 1 callersFunctionmain
( BATCH: int = 1, H: int = 32, N_CTX: int = 256, D_HEAD_QK: int = 192, D_HEAD_V: int = 128
examples/flash_attention/example_gqa_bwd_tma_reduce_varlen.py:622
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), K: T.Tensor(kv_shape, dtype), V: T.Tensor(kv_shape, dty
examples/flash_attention/example_mha_fwd_bhsd.py:34
↓ 1 callersFunctionmain
( Q_unpad: T.Tensor(q_shape, dtype), K_unpad: T.Tensor(kv_shape, dtype), V_unpad: T.Te
examples/attention_sink/example_gqa_sink_bwd_varlen.py:61
↓ 1 callersFunctionmain
( Q_unpad: T.Tensor(q_shape, dtype), K_unpad: T.Tensor(kv_shape, dtype), V_unpad: T.Te
examples/attention_sink/example_gqa_sink_fwd_varlen.py:54
↓ 1 callersFunctionmain
(BATCH: int = 1, H: int = 1, N_CTX: int = 512, D_HEAD: int = 128, window_size: Optional[int] = None, dtype: T.
examples/attention_sink/example_mha_sink_bwd_bhsd.py:431
↓ 1 callersFunctionmain
( batch: int = 1, heads: int = 32, seq_q: int = 256, seq_kv: int = 256, dim: int = 128,
examples/attention_sink/benchmark_gqa_sink_fwd.py:128
↓ 1 callersFunctionmain
( batch: int = 1, heads: int = 32, seq_q: int = 256, seq_kv: int = 256, dim: int = 128,
examples/attention_sink/benchmark_mha_sink_fwd.py:125
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), K: T.Tensor(kv_shape, dtype), V: T.Tensor(kv_shape, dty
examples/attention_sink/example_mha_sink_fwd_bhsd.py:53
↓ 1 callersFunctionmain
( BATCH: int = 1, H: int = 8, N_CTX: int = 512, D_HEAD: int = 64, groups: int = 2, win
examples/attention_sink/example_gqa_sink_bwd_bhsd.py:427
↓ 1 callersFunctionmain
()
examples/blockscaled_gemm_sm100/grouped_gemm_mxfp8_blockscaled_1d1d.py:591
↓ 1 callersFunctionmain
()
examples/blockscaled_gemm_sm100/gemm_mxfp8_blockscaled_1d1d.py:644
↓ 1 callersFunctionmain
(*args, **kwargs)
examples/autodd/tilelang_minimized_expected.py:47
↓ 1 callersFunctionmain
()
examples/autodd/tilelang_buggy.py:196
↓ 1 callersFunctionmain
(M=1024, N=1024)
examples/elementwise/example_elementwise_add.py:35
↓ 1 callersFunctionmain
()
examples/hadamard_transform/example_hadamard.py:131
↓ 1 callersFunctionmain
()
examples/gemm_sm100/gemm_tcgen5mma_ws.py:138
↓ 1 callersFunctionmain
()
examples/gemm_sm100/gemm_tcgen5mma_ws_persistent.py:269
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/benchmark_inference_latency.py:37
↓ 1 callersFunctionmain
(args)
examples/bitnet-1.58b/eval_ppl.py:31
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/load_from_quantized.py:50
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/benchmark_generate.py:77
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/eval_gpu_memory.py:37
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/benchmark_model_10k_loops.py:44
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/eval_correctness.py:75
↓ 1 callersFunctionmain
()
examples/bitnet-1.58b/maint/create_bitblas_ckpt.py:64
↓ 1 callersFunctionmain
( Q: T.Tensor(shape_q, dtype), K: T.Tensor(shape_k, dtype), V: T.Tensor(shape_v, dtype
examples/blocksparse_attention/example_tilelang_sparse_gqa_decode_varlen_indice.py:38
↓ 1 callersFunctionmain
(batch=64, heads=32, heads_kv=8, max_cache_seqlen=8192, dim=128, dim_v=128, sparse_ratio=0.8, block_size=32)
examples/blocksparse_attention/example_triton_sparse_gqa_decode_varlen_mask.py:317
↓ 1 callersFunctionmain
()
examples/blocksparse_attention/example_tilelang_block_sparse_attn.py:175
↓ 1 callersFunctionmain
( Q: T.Tensor(shape_q, dtype), K: T.Tensor(shape_k, dtype), V: T.Tensor(shape_v, dtype
examples/blocksparse_attention/example_tilelang_sparse_gqa_decode_paged.py:43
↓ 1 callersFunctionmain
( Q: T.Tensor(shape_q, dtype), K: T.Tensor(shape_k, dtype), V: T.Tensor(shape_v, dtype
examples/blocksparse_attention/example_tilelang_sparse_gqa_decode_varlen_mask.py:38
↓ 1 callersFunctionmain
(batch=64, heads=32, heads_kv=8, max_cache_seqlen=8192, dim=128, dim_v=128, sparse_ratio=0.8, block_size=32)
examples/blocksparse_attention/example_triton_sparse_gqa_decode_varlen_indice.py:321
↓ 1 callersFunctionmain
()
examples/blocksparse_attention/block_sparse_attn_triton.py:354
↓ 1 callersFunctionmain
()
examples/analyze/example_gemm_analyze.py:51
↓ 1 callersFunctionmain
()
examples/analyze/example_conv_analyze.py:53
↓ 1 callersFunctionmain
()
examples/gdn/example_wy_fast.py:195
↓ 1 callersFunctionmain
()
examples/gdn/example_cumsum.py:147
↓ 1 callersFunctionmain
()
examples/gdn/example_wy_fast_bwd_split.py:511
↓ 1 callersFunctionmain
()
examples/gdn/example_chunk_delta_h.py:353
↓ 1 callersFunctionmain
()
examples/gdn/example_chunk_delta_bwd.py:548
↓ 1 callersFunctionmain
()
examples/gdn/example_chunk_o.py:213
↓ 1 callersFunctionmain
()
examples/gdn/example_chunk_o_bwd.py:466
↓ 1 callersFunctionmain
()
examples/gdn/example_chunk_scaled_dot_kkt.py:171
↓ 1 callersFunctionmain
()
examples/visual_layout_inference/visual_layout_inference.py:35
↓ 1 callersFunctionmain
(M=8192, N=8192, blk_m=8)
examples/cast/example_per_token_cast_to_fp8.py:71
↓ 1 callersFunctionmain
(M=8192, N=8192, BG=2, blk_m=8, batch_sizes=None)
examples/cast/example_group_per_split_token_cast_to_fp8.py:156
↓ 1 callersFunctionmain
( A: T.Tensor((M, K), T.int4), B: T.Tensor((N, K), T.int4), C: T.Tensor((M, N), T.int3
examples/gemm_int4/example_tilelang_gemm_int4.py:19
↓ 1 callersFunctionmain
()
examples/deepseek_mhc/example_mhc_pre.py:482
↓ 1 callersFunctionmain
()
examples/deepseek_mhc/example_mhc_bwd.py:184
↓ 1 callersFunctionmain
(M=16384, N=16384, K=16384)
examples/warp_specialize/example_warp_specialize_gemm_barrierpipe_stage2.py:53
↓ 1 callersFunctionmain
(M=16384, N=16384, K=16384)
examples/warp_specialize/example_warp_specialize_gemm_copy_1_gemm_0.py:42
↓ 1 callersFunctionmain
(batch=1, heads=64, kv_heads=1, kv_ctx=1024, dim=512, pe_dim=64)
examples/warp_specialize/example_warp_specialize_flashmla.py:352
↓ 1 callersFunctionmain
(M=1024, N=1024, K=1024)
examples/warp_specialize/example_warp_specialize_gemm_copy_0_gemm_1.py:41
↓ 1 callersFunctionmain
(M=16384, N=16384, K=16384)
examples/warp_specialize/example_warp_specialize_gemm_softpipe_stage2.py:43
↓ 1 callersFunctionmain
()
examples/warp_specialize/example_warp_specialize_gemm_copy_gemm_0_1.py:47
↓ 1 callersFunctionmain
( M=1024, N=1024, K=1024, accum_dtype=T.float, e_dtype=T.int16, block_M=128, block
examples/gemm_sp/example_gemm_sp.py:45
↓ 1 callersFunctionmain
(d_hidden=7168, d_expert=2048, n_routed_experts=8, n_shared_experts=1, n_experts_per_token=4, batch_size=1, se
examples/fusedmoe/example_fusedmoe_tilelang.py:476
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), K: T.Tensor(kv_shape, dtype), V: T.Tensor(k
examples/seer_attention/block_sparse_attn_tilelang.py:51
↓ 1 callersFunctionmain
(argv=None)
examples/topk/example_topk.py:59
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), K: T.Tensor(kv_shape, dtype), V: T.Tensor(kv_shape, dty
examples/flash_attention_sm100/gqa_bwd_bshd.py:31
↓ 1 callersFunctionmain
( Q: T.Tensor(shape, dtype), K: T.Tensor(shape, dtype), V: T.Tensor(shape, dtype),
examples/flash_attention_sm100/mha_bwd_bshd.py:26
↓ 1 callersFunctionmain
( Q: T.Tensor(shape, dtype), K: T.Tensor(shape, dtype), V: T.Tensor(shape, dtype),
examples/flash_attention_sm100/mha_fwd_bshd.py:42
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), K: T.Tensor(kv_shape, dtype), V: T.Tensor(kv_shape, dty
examples/flash_attention_sm100/gqa_fwd_bshd.py:45
↓ 1 callersFunctionmain
()
examples/kda/chunk_inter_solve_fused.py:549
↓ 1 callersFunctionmain
()
examples/kda/chunk_intra_token_parallel.py:256
↓ 1 callersFunctionmain
()
examples/kda/wy_fast_bwd.py:327
↓ 1 callersFunctionmain
()
examples/kda/chunk_delta_bwd.py:283
↓ 1 callersFunctionmain
()
examples/kda/chunk_delta_h_fwd.py:281
↓ 1 callersFunctionmain
()
examples/kda/chunk_bwd_dqkwg.py:250
↓ 1 callersFunctionmain
()
examples/kda/wy_fast.py:211
↓ 1 callersFunctionmain
()
examples/kda/chunk_o.py:222
↓ 1 callersFunctionmain
()
examples/kda/chunk_bwd_intra.py:473
↓ 1 callersFunctionmain
()
examples/kda/chunk_bwd_gla_dA.py:131
↓ 1 callersFunctionmain
()
examples/kda/chunk_bwd_dv.py:134
↓ 1 callersFunctionmain
( A: T.Tensor((M, K), T.float4_e2m1fn), B: T.Tensor((N, K), T.float8_e4m3fn), SFA: T.T
examples/deepseek_v4/fp8_fp4_gemm_1d1d_sm100.py:40
↓ 1 callersFunctionmain
( Q: T.Tensor(q_shape, dtype), KV: T.Tensor(kv_shape, dtype), TopkIndices: T.Tensor(id
examples/deepseek_v4/sparse_attn_fwd_sm90.py:64
↓ 1 callersFunctionmain
( batch=1, heads=128, kv_heads=1, kv_ctx=8192, dim=512, pe_dim=64, )
examples/deepseek_mla/example_mla_decode.py:218
↓ 1 callersFunctionmain
( batch=1, heads=128, kv_heads=1, kv_ctx=8192, dim=512, pe_dim=64, )
examples/deepseek_mla/example_mla_decode_ws.py:548
↓ 1 callersFunctionmain
()
examples/deepseek_mla/example_mla_decode_persistent.py:175
↓ 1 callersFunctionmain
()
examples/gemm_fp8/example_tilelang_gemm_fp8_2xAcc.py:74
↓ 1 callersFunctionmain
()
examples/gemm_fp8/example_tilelang_gemm_fp8.py:56
↓ 1 callersFunctionmain
()
testing/python/language/test_tilelang_language_tma_1d.py:49
↓ 1 callersFunctionmain
()
cmake/generate_windows_import_lib.py:21
↓ 1 callersFunctionmain
()
cmake/ccache_strip_pep517.py:55
↓ 1 callersFunctionmain
()
cmake/find_pip_cuda.py:188
↓ 1 callersFunctionmain
(args: Args)
tilelang/autodd.py:1257
↓ 1 callersMethodmain
(A: T.Buffer(1, "float32"))
testing/python/transform/test_tilelang_transform_make_packed_api.py:86
↓ 1 callersFunctionmake_cutlass_metadata_layout_sm8x
Make a layout of metadata that is compatible with cutlass sm8x compression kernel. Note that layout atom is the same for smem and gmem. ref: h
tilelang/layout/gemm_sp.py:121
↓ 1 callersFunctionmake_cutlass_metadata_layout_sm90
Make a layout of metadata that is compatible with cutlass sm90 compression kernel. Note that layout atom is the same for smem and gmem. Args:
tilelang/layout/gemm_sp.py:21
↓ 1 callersFunctionmake_kernel
()
testing/python/language/test_tilelang_language_reduce.py:287
↓ 1 callersFunctionmake_matmul_prim
(M, N, K, block_M=128, block_N=128, block_K=32, dtype=T.float16, accum_dtype=T.float32)
maint/host_checks/common.py:6
↓ 1 callersFunctionmake_mfma_load_base_layout
Create a layout function for storing MFMA results into a fragment buffer. This layout is used in conjunction with `inverse_mfma_store_layout`
examples/plot_layout/fragment_mfma_load_a.py:14
↓ 1 callersFunctionmake_mma_atom_kernel
(M, N, K, in_dtype, out_dtype, accum_dtype)
testing/python/language/test_tilelang_language_atom_mma.py:47
↓ 1 callersFunctionmake_mma_load_base_layout
Create a layout function for storing MMA results into a fragment buffer. This layout is used in conjunction with `inverse_mma_store_layout` t
examples/plot_layout/fragment_mma_load_a.py:8
↓ 1 callersFunctionmake_offsets
(batch_sizes, device)
examples/blockscaled_gemm_sm100/grouped_gemm_mxfp8_blockscaled_1d1d.py:530
↓ 1 callersMethodmake_reverse
(self, input_name: str, input_iter: list[tirx.PrimExpr])
tilelang/carver/roller/shape_inference/tir.py:17
← previousnext →2,501–2,600 of 7,254, ranked by callers