MCPcopy Create free account

hub / github.com/KnowingNothing/MatmulTutorial / functions

Functions541 in github.com/KnowingNothing/MatmulTutorial

⨍ Functions 541◇ Types & classes 125
↓ 1 callersMethoddouble_t
(cls)
cutlass.py/dtype.py:147
↓ 1 callersMethode3m4_t
(cls)
cutlass.py/dtype.py:131
↓ 1 callersMethode4m3_t
(cls)
cutlass.py/dtype.py:135
↓ 1 callersFunctionfence_barrier_init
include/barrier.h:11
↓ 1 callersMethodfloat_t
(cls)
cutlass.py/dtype.py:143
↓ 1 callersMethodget
(self, name)
util/simulator/kernel.py:13
↓ 1 callersMethodget_accumulator_matrix
(self)
cutlass.py/mma.py:57
↓ 1 callersMethodget_blocks_m_n
(self, M, N)
cutlass.py/tile_scheduler.py:429
↓ 1 callersMethodget_current_m_n_idx
(self)
cutlass.py/tile_scheduler.py:417
↓ 1 callersMethodget_current_work
(self)
cutlass.py/tile_scheduler.py:295
↓ 1 callersMethodget_current_work_for_linear_idx
(self, linear_idx)
cutlass.py/tile_scheduler.py:298
↓ 1 callersMethodget_current_work_info
(self)
cutlass.py/tile_scheduler.py:408
↓ 1 callersMethodget_work_idx_m_and_n
( dev_coord, blk_per_grid_dim, divmod_cluster_shape_major, divmod_cluster_shap
cutlass.py/tile_scheduler.py:319
↓ 1 callersMethodhalf_t
(cls)
cutlass.py/dtype.py:139
↓ 1 callersMethodinit
(self, M, N)
cutlass.py/tile_scheduler.py:402
↓ 1 callersMethodinitialize
( self, problem_shape: HyperCube, # m, n, k, l tile_shape: HyperCube, cluster
cutlass.py/tile_scheduler.py:29
↓ 1 callersMethodint16_t
(cls)
cutlass.py/dtype.py:56
↓ 1 callersMethodint32_t
(cls)
cutlass.py/dtype.py:60
↓ 1 callersMethodint4_t
(cls)
cutlass.py/dtype.py:48
↓ 1 callersMethodint64_t
(cls)
cutlass.py/dtype.py:64
↓ 1 callersMethodint8_t
(cls)
cutlass.py/dtype.py:52
↓ 1 callersFunctioninteger_log2
(x)
cutlass.py/fast_math.py:5
↓ 1 callersMethodinvalid_work_tile
(cls)
cutlass.py/tile_scheduler.py:249
↓ 1 callersFunctionis_block_size_legal
Check if a (block_m, block_n) configuration is legal. Checks performed (matching DeepGEMM): 1. block_n must be divisible by 16 (layout
examples/matmul/this-sm100/fp8_level6/test.py:53
↓ 1 callersFunctionis_block_size_legal
Check TMEM and basic constraints.
examples/matmul/this-sm100/fp8_level5/test.py:59
↓ 1 callersFunctionis_cuda
()
examples/matmul/triton/triton_matmul_sm90.py:29
↓ 1 callersMethodis_e3m4_t
(self)
cutlass.py/dtype.py:112
↓ 1 callersMethodis_e4m3_t
(self)
cutlass.py/dtype.py:115
↓ 1 callersFunctionleaky_relu
(x)
examples/matmul/triton/triton_matmul.py:259
↓ 1 callersFunctionlinear_attention_bwd
(rq, iq, rk, ik, rv, iv, rgo, igo, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm.py:296
↓ 1 callersFunctionlinear_attention_bwd
(rq, iq, rk, ik, rv, iv, rgo, igo, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm_v2.py:350
↓ 1 callersFunctionlinear_attention_fwd
(rq, iq, rk, ik, rv, iv, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_tile_k_tile_n.py:222
↓ 1 callersFunctionlinear_attention_fwd
(rq, iq, rk, ik, rv, iv, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_tile_k.py:199
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level3/test.py:38
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level2/test.py:38
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level1/test.py:44
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level1/debug_test.py:16
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level1/debug_layout.py:15
↓ 1 callersFunctionload_lib
()
examples/matmul/this-sm100/fp8_level1/test_umma.py:16
↓ 1 callersFunctionload_matmul_kernel
()
examples/matmul/this-sm100/test_this_perf.py:224
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex.py:212
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_tile_k_tile_n.py:250
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm.py:330
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_bwd_bind_bh.py:396
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_tile_k.py:227
↓ 1 callersFunctionmain
(batch_size, num_heads, seq_len, model_k, r_scale, i_scale)
examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm_v2.py:384
↓ 1 callersFunctionmain
(M, N, K, batch_sizes=None)
examples/matmul/triton/complex_matmul.py:116
↓ 1 callersFunctionmain
(M, N, K, trials)
examples/matmul/triton/triton_matmul_sm90.py:190
↓ 1 callersFunctionmain
(M, N, K, trials)
examples/matmul/triton/triton_matmul.py:299
↓ 1 callersFunctionmain
(M, N, K)
examples/matmul/cutlass/test_cutlass_perf.py:24
↓ 1 callersFunctionmain
(M, N, K)
examples/matmul/cutlass-sm90/test_cutlass_perf.py:24
↓ 1 callersFunctionmain
(M, N, K, in_dtype, acc_dtype, only_once)
examples/matmul/tvm/relay_matmul_cublas.py:34
↓ 1 callersFunctionmain
(M, N, K, in_dtype, out_dtype, only_once, trials)
examples/matmul/tvm/tensorir_meta_schedule.py:95
↓ 1 callersFunctionmain
(M, N, K, in_dtype, acc_dtype, only_once)
examples/matmul/tvm/relay_matmul_cutlass.py:68
↓ 1 callersFunctionmain
(M, N, K)
examples/matmul/mlir-gen/test_mlir_perf.py:34
↓ 1 callersFunctionmain
(M, N, K)
examples/matmul/cublas/test_cublas_perf.py:24
↓ 1 callersFunctionmain
()
examples/matmul/this-sm90/test_v2.py:195
↓ 1 callersFunctionmain
()
examples/matmul/this-sm90/test_v3.py:182
↓ 1 callersFunctionmain
()
examples/matmul/this-sm90/test_v4.py:194
↓ 1 callersFunctionmain
(M, N, K, version)
examples/matmul/this-sm90/test_this_perf.py:24
↓ 1 callersFunctionmain
(M, N, K)
examples/matmul/this-sm80/test_this_perf.py:24
↓ 1 callersFunctionmain
(M_list, N, K, trials)
examples/moe_group_gemm/triton/grouped_gemm.py:220
↓ 1 callersFunctionmain
(M_list, N, K, trials)
examples/moe_group_gemm/triton/blocked_grouped_gemm.py:136
↓ 1 callersFunctionmain
(M_list, N, K, trials)
examples/moe_group_gemm/cutlass/cutlass_grouped_gemm.py:16
↓ 1 callersFunctionmatmul_fp16
(N: int, M: int, K: int, in_dtype: str, out_dtype: str)
examples/matmul/tvm/tensorir_meta_schedule.py:28
↓ 1 callersMethodmodulus
(self, dividend)
cutlass.py/fast_math.py:22
↓ 1 callersFunctionmulti_level_tiling_mma
(out_dtype)
examples/matmul/tvm/tensorir_meta_schedule.py:64
↓ 1 callersFunctionprint_matrix
(mtx, rows, cols, func=lambda x: x, prompt="")
cutlass.py/mma.py:171
↓ 1 callersFunctionproducer_acquire
include/pipeline.h:177
↓ 1 callersFunctionprofile_and_build
(mod, params, sm, tmp_dir="./tmp", lib_path="compile.so", use_fast_math=False)
examples/matmul/tvm/relay_matmul_cutlass.py:29
↓ 1 callersFunctionreference_matmul
Compute reference: float(A) @ float(B)^T
examples/matmul/this-sm100/fp8_level1/test.py:90
↓ 1 callersFunctionrelay_matmul
( M, N, K, in_dtype="float16", acc_dtype="float32", target="cuda" )
examples/matmul/tvm/relay_matmul_cublas.py:9
↓ 1 callersFunctionrelay_matmul
( M, N, K, in_dtype="float16", acc_dtype="float32", target="cuda" )
examples/matmul/tvm/relay_matmul_cutlass.py:16
↓ 1 callersFunctionreplace_component_fused
(model: nn.Module)
examples/attention/triton/model.py:212
↓ 1 callersFunctionrun_benchmarks
(lib)
examples/matmul/this-sm100/fp8_level3/test.py:95
↓ 1 callersFunctionrun_benchmarks
(lib)
examples/matmul/this-sm100/fp8_level2/test.py:95
↓ 1 callersFunctionrun_benchmarks
(lib)
examples/matmul/this-sm100/fp8_level1/test.py:166
↓ 1 callersFunctionrun_gemm
Launch GEMM kernel.
examples/matmul/this-sm100/fp8_level6/test.py:294
↓ 1 callersFunctionrun_gemm
Run GEMM with the given library variant.
examples/matmul/this-sm100/fp8_level5/test.py:115
↓ 1 callersFunctionrun_random_test
(lib, M, N, K)
examples/matmul/this-sm100/fp8_level1/debug_test.py:41
↓ 1 callersFunctionrun_test
(lib, M, N, K)
examples/matmul/this-sm100/fp8_level1/debug_test.py:24
↓ 1 callersFunctionselect_best_block_n
(M, N, K, block_m=128)
examples/matmul/this-sm100/fp8_level5/bench_compare.py:44
↓ 1 callersFunctionselect_best_block_n
Select BLOCK_N that minimizes waves, then maximizes last-wave utilization. Matches DeepGEMM's get_best_config heuristic.
examples/matmul/this-sm100/fp8_level5/test.py:71
↓ 1 callersMethodset_blockIdx
(self, x, y, z)
cutlass.py/hw_info.py:35
↓ 1 callersFunctionset_block_rank
Set the destination block-ID in cluster for a given SMEM Address
include/common.h:426
↓ 1 callersFunctionshiftr
include/common.h:88
↓ 1 callersFunctiontest
(lib, size)
examples/matmul/this-sm90/test_v6.py:58
↓ 1 callersFunctiontest
Test correctness for a given size
examples/matmul/this-sm90/test_v9.py:46
↓ 1 callersFunctiontest
Test correctness for a given size
examples/matmul/this-sm90/test_v8.py:48
↓ 1 callersFunctiontest
(lib, size)
examples/matmul/this-sm90/test_v5.py:28
↓ 1 callersFunctiontest
(lib, size)
examples/matmul/this-sm90/test_v7.py:60
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level6/test.py:38
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level7/test.py:38
↓ 1 callersFunctiontest_correctness
(M=4096, N=4096, K=4096, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level9/test.py:35
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level5/test.py:38
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level4/test.py:38
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level2/test.py:38
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level3/test.py:38
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level1/test.py:44
↓ 1 callersFunctiontest_correctness
(M=TEST_SIZE, N=TEST_SIZE, K=TEST_SIZE, atol=1e-2, rtol=1e-2)
examples/matmul/this-sm100/level8/test.py:41
← previousnext →201–300 of 541, ranked by callers