MCPcopy Create free account

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

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

Function_fp4_eager_kernel
(A)
testing/python/issue/test_tilelang_issue_2292.py:12
Function_fp4_eager_strided_kernel
(A)
testing/python/issue/test_tilelang_issue_2292.py:24
Method_ftransform
(f, mod, ctx)
tilelang/tools/Analyzer.py:115
Function_fwd_kernel
( Q, K, V, sm_scale, block_mask_ptr, Out, stride_qz, stride_qh, stride_qm,
benchmark/blocksparse_attention/benchmark_triton_block_sparse_fmha.py:93
Function_fwd_kernel
( Q, K, V, sm_scale, block_mask_ptr, Out, stride_qz, stride_qh, stride_qm,
examples/blocksparse_attention/block_sparse_attn_triton.py:93
Function_fwd_kernel
( Q, K, V, sm_scale, block_mask_ptr, Out, stride_qz, stride_qh, stride_qm,
examples/seer_attention/block_sparse_attn_triton.py:90
Function_ge
(a, b)
tilelang/language/parser/operation.py:110
Method_gemm_blockscaled
()
tilelang/cuda/op/gemm/gemm_tcgen05.py:293
Method_gemm_blockscaled_cond
()
tilelang/cuda/op/gemm/gemm_tcgen05.py:277
Method_gemm_rrr
()
tilelang/rocm/op/gemm/gemm_wmma.py:149
Method_gemm_rrr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma.py:204
Method_gemm_rrr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm_sp/gemm_sp_mma.py:224
Method_gemm_rsr
()
tilelang/rocm/op/gemm/gemm_wmma.py:134
Method_gemm_rsr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Matrix Co
tilelang/rocm/op/gemm/gemm_mfma.py:193
Method_gemm_rsr
()
tilelang/tileop/gemm_sp/gemm_sp_wgmma.py:145
Method_gemm_rsr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma_sm70.py:145
Method_gemm_rsr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_wgmma.py:155
Method_gemm_rsr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma.py:176
Method_gemm_rsr
()
tilelang/cuda/op/gemm_sp/gemm_sp_wgmma.py:145
Method_gemm_rsr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm_sp/gemm_sp_mma.py:186
Method_gemm_scalar
()
tilelang/cpu/op/gemm/gemm_scalar.py:45
Method_gemm_srr
()
tilelang/rocm/op/gemm/gemm_wmma.py:120
Method_gemm_srr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Matrix Co
tilelang/rocm/op/gemm/gemm_mfma.py:162
Method_gemm_srr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma.py:146
Method_gemm_srr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm_sp/gemm_sp_mma.py:147
Method_gemm_ss
()
tilelang/cuda/op/gemm/gemm_tcgen05.py:217
Method_gemm_ss_cond
()
tilelang/cuda/op/gemm/gemm_tcgen05.py:210
Method_gemm_ss_shared
()
tilelang/metal/op/gemm/gemm_metal.py:90
Method_gemm_ss_simdgroup
()
tilelang/metal/op/gemm/gemm_metal.py:75
Method_gemm_ssr
()
tilelang/rocm/op/gemm/gemm_wmma.py:104
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Matrix Co
tilelang/rocm/op/gemm/gemm_mfma.py:127
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/tileop/gemm_sp/gemm_sp_wgmma.py:129
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma_sm70.py:108
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_wgmma.py:140
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm/gemm_mma.py:111
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm_sp/gemm_sp_wgmma.py:129
Method_gemm_ssr
The inner macro that loads data from shared buffers A_shared and B_shared into local fragments, then issues Tensor Co
tilelang/cuda/op/gemm_sp/gemm_sp_mma.py:102
Function_get_element_size
Get element size in bits from BufferLikeType. Args: buffer_or_load_or_region: BufferLikeType Returns: int: Element size
tilelang/layout/swizzle.py:50
Method_get_required_files
(self, cache_path: str)
tilelang/jit/adapter/nvrtc/kernel_cache.py:11
Method_get_required_files
Return the CuTeDSL cache files required for a runnable reload.
tilelang/jit/adapter/cutedsl/kernel_cache.py:68
Function_get_stride_continuous
Get stride (product of all dims except the last) and continuous (last dimension) from BufferLikeType. Args: buffer_or_load_or_re
tilelang/layout/swizzle.py:31
Function_gt
(a, b)
tilelang/language/parser/operation.py:107
Function_hip_target
()
testing/python/amd/test_tilelang_mxfp4_gfx950.py:26
Method_init_a
(desc_a, A_ptr)
tilelang/cuda/intrinsics/macro/tcgen05_macro_generator.py:810
Method_init_a_desc
(desc_a, A_ptr)
tilelang/cuda/intrinsics/macro/wgmma_macro_generator.py:454
Method_init_b
(desc_b, B_ptr)
tilelang/cuda/intrinsics/macro/tcgen05_macro_generator.py:786
Method_init_b_desc
(desc_b, B_ptr)
tilelang/cuda/intrinsics/macro/wgmma_macro_generator.py:426
Method_init_weights
(self, module)
examples/bitnet-1.58b/modeling_bitnet.py:847
Method_init_with_prim_func
(self, func: tvm.tirx.PrimFunc, name: str | None = None)
tilelang/carver/roller/policy/tensorcore.py:25
Method_initialize_abbrev
(self, a_dtype, b_dtype, accum_dtype)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:946
Method_initialize_k_dim
(self, a_dtype=T.float16)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:938
Method_initialize_k_dim
(self, a_dtype=T.float16)
tilelang/cuda/intrinsics/macro/tcgen05_macro_generator.py:172
Method_initialize_k_dim
(self, a_dtype=T.float16)
tilelang/cuda/intrinsics/macro/mma_sm75_macro_generator.py:12
Method_initialize_local_size
(self, m_dim=16, n_dim=16, k_dim=16, warp_size=32)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:941
Method_initialize_m_dim
(self, a_dtype=T.float16)
tilelang/cuda/intrinsics/macro/mma_sm75_macro_generator.py:24
Method_initialize_micro_size
(self, m_dim=16, k_dim=16)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:959
Method_initialize_micro_size
(self, m_dim: int = 16, k_dim: int = 16)
tilelang/cuda/intrinsics/macro/tcgen05_macro_generator.py:167
Method_initialize_micro_size
(self, m_dim: int = 16, k_dim: int = 16)
tilelang/cuda/intrinsics/macro/mma_sm75_macro_generator.py:41
Method_initialize_micro_size
(self, m_dim: int = 16, k_dim: int = 16)
tilelang/cuda/intrinsics/macro/wgmma_macro_generator.py:167
Method_initialize_mma_prefix
(self, k_dim=16)
tilelang/cuda/intrinsics/macro/mma_macro_generator.py:951
Method_initialize_mma_prefix
(self, k_dim: int = 16)
tilelang/cuda/intrinsics/macro/mma_sm75_macro_generator.py:31
Method_initialize_torch_cuda_arch_flags
Detect target CUDA architecture and set TORCH_CUDA_ARCH_LIST to ensure PyTorch extensions are built for the proper GPU arch.
tilelang/env.py:341
Function_is_cutedsl_available
Return whether the CuTeDSL stack is usable in the current environment.
testing/python/autotune/test_tilelang_autotune_cutedsl_cache.py:62
Function_kernel_min_blocks_per_sm
()
testing/python/language/test_tilelang_language_min_blocks_per_sm.py:9
Function_kernel_rbits_idiv2
rbits = rbits_buf[i // 2] -> shared across 2 lanes, expect x2.
testing/python/language/test_tilelang_cast_rounding.py:245
Function_kernel_rbits_idiv4
rbits = rbits_buf[i // 4] -> shared across 4 lanes, expect x4.
testing/python/language/test_tilelang_cast_rounding.py:222
Function_kernel_rbits_per_element
rbits = rbits_buf[i] (per-element) -> no sharing, expect x1.
testing/python/language/test_tilelang_cast_rounding.py:268
Function_kernel_rbits_rng_rand
rbits = T.rng_rand() -> curand call carries side effects, expect x1. Each lane needs its own random sample (PTX cvt.rs semantics shares one r
testing/python/language/test_tilelang_cast_rounding.py:291
Function_kernel_rbits_scalar
rbits = scalar fragment (loop-invariant) -> max lanes.
testing/python/language/test_tilelang_cast_rounding.py:200
Function_le
(a, b)
tilelang/language/parser/operation.py:104
Method_load_kernel_source
(self, device_kernel_path: str, host_kernel_path: str, verbose: bool = False)
tilelang/cache/kernel_cache.py:782
Function_lt
(a, b)
tilelang/language/parser/operation.py:101
Function_match_metal
(target)
tilelang/metal/op/gemm/__init__.py:8
Function_match_mfma
(target)
tilelang/rocm/op/gemm/__init__.py:9
Function_match_mma
(target)
tilelang/cuda/op/gemm/__init__.py:14
Function_match_mma
(target)
tilelang/cuda/op/gemm_sp/__init__.py:11
Function_match_mma_sm70
(target)
tilelang/cuda/op/gemm/__init__.py:18
Function_match_mma_sm75
(target)
tilelang/cuda/op/gemm/__init__.py:22
Function_match_scalar
(target)
tilelang/cpu/op/gemm/__init__.py:7
Function_match_tcgen05
(target)
tilelang/cuda/op/gemm/__init__.py:30
Function_match_wgmma
(target)
tilelang/cuda/op/gemm/__init__.py:26
Function_match_wgmma
(target)
tilelang/cuda/op/gemm_sp/__init__.py:15
Function_match_wmma
(target)
tilelang/rocm/op/gemm/__init__.py:13
Function_memory_order_to_llvm_load
Convert TileLang memory order ID to LLVM atomic ordering for loads. TileLang memory order mapping: 0: relaxed -> monotonic 1: c
tilelang/contrib/cutedsl/atomic.py:34
Function_memory_order_to_llvm_store
Convert TileLang memory order ID to LLVM atomic ordering for stores. TileLang memory order mapping: 0: relaxed -> monotonic 1:
tilelang/contrib/cutedsl/atomic.py:58
Function_merge_kernel
( o_partial_ptr, lse_partial_ptr, o_ptr, lse_partial_stride_b, lse_partial_stride_h, l
examples/blocksparse_attention/example_triton_sparse_gqa_decode_varlen_mask.py:124
Function_merge_kernel
( o_partial_ptr, lse_partial_ptr, o_ptr, lse_partial_stride_b, lse_partial_stride_h, l
examples/blocksparse_attention/example_triton_sparse_gqa_decode_varlen_indice.py:127
Function_metal_fragment_to_simdgroup
(func: tir.PrimFunc, mod: IRModule, ctx)
tilelang/metal/transform/metal_fragment_to_simdgroup.py:142
Function_mla_attn_kernel
( Q_nope, Q_pe, Kv_c_cache, K_pe_cache, Req_to_tokens, B_seq_len, O, sm_scale,
examples/deepseek_mla/benchmark_mla.py:137
Function_mla_attn_kernel
( Q_nope, Q_pe, Kv_c_cache, K_pe_cache, Req_to_tokens, B_seq_len, O, sm_scale,
examples/deepseek_mla/amd/benchmark_mla_decode_amd_triton.py:59
Function_mla_softmax_reducev_kernel
( Logits, B_seq_len, O, stride_l_b, stride_l_h, stride_l_s, stride_o_b, stride
examples/deepseek_mla/benchmark_mla.py:276
Function_mla_softmax_reducev_kernel
( Logits, B_seq_len, O, stride_l_b, stride_l_h, stride_l_s, stride_o_b, stride
examples/deepseek_mla/amd/benchmark_mla_decode_amd_triton.py:199
Method_model_call
inps: a torch tensor of shape [batch, sequence] the size of sequence may vary from call to call returns: a torch tensor of s
examples/bitnet-1.58b/eval_utils.py:122
Method_model_generate
(self, context, max_length, eos_token_id)
examples/bitnet-1.58b/eval_utils.py:134
Function_ne
(a, b)
tilelang/language/parser/operation.py:98
Method_optimize
(node, rstep)
tilelang/carver/roller/policy/tensorcore.py:133
Method_optimize
(node, rstep)
tilelang/carver/roller/policy/default.py:348
Function_or
(a, b)
tilelang/language/parser/operation.py:42
Function_parallel_vectorize_local_and_var
()
testing/python/language/test_tilelang_language_parallel.py:77
Method_parse_names
(self, target: ast.expr)
tilelang/language/eager/ast.py:290
← previousnext →3,801–3,900 of 7,254, ranked by callers