Functions — github.com/KnowingNothing/MatmulTutorial

CodeHub MCPcopy Create free account

hub / github.com/KnowingNothing/MatmulTutorial / functions

Functions541 in github.com/KnowingNothing/MatmulTutorial

⨍ Functions 541◇ Types & classes 125

util/simulator/kernel.py:103

(self, other: Any)

util/simulator/data_math.py:46

Method__rtrueediv__

(self, other: Any)

util/simulator/data_math.py:94

(self, other: Any)

util/simulator/data_math.py:233

Method__setitem__

(self, key: int, value: Union[int, IntegerType])

cutlass.py/tiling.py:30

Method__setitem__

(self, key: int, value: Union[int, IntegerType])

cutlass.py/tiling.py:112

util/simulator/kernel.py:88

(self, other: Any)

util/simulator/data_math.py:38

Method__truediv__

(self, other: Any)

util/simulator/data_math.py:86

(self, other: Any)

util/simulator/data_math.py:225

Function_matmul_launch_metadata

(grid, kernel, args)

examples/matmul/triton/triton_matmul_sm90.py:37

include/barrier.h:299

include/scheduler.h:48

Functionalloc_cpu_tensor

include/common.h:276

Functionalloc_gpu_tensor

include/common.h:310

Functionarange_fill

include/common.h:292

Signal completion of Stage and move to the next stage (group_id) signals to (group_id+1)

include/barrier.h:293

Methodarrive_and_reset_bytes

DEPRECATED APIs

include/barrier.h:420

util/simulator/data_math.py:16

Functionassert_allclose

include/common.h:368

Functionasync_build

(mod, target, _params)

examples/matmul/tvm/tensorir_meta_schedule.py:49

cutlass.py/mapping.py:10

cutlass.py/dtype.py:102

Methodcheck_phase

include/barrier.h:284

Functioncluster_arrive_relaxed

include/barrier.h:18

Functioncluster_wait

include/barrier.h:22

include/barrier.h:432

Methodcomplete_transaction

Performs an expected transaction bytes decrement without doing an arrive operation

include/barrier.h:334

Functioncomplex_matmul_bf16_kernel

( ar_ptr, ai_ptr, br_ptr, bi_ptr, cr_ptr, ci_ptr, batch_size, M, N, K, stride_ab, stride_am, strid

examples/matmul/triton/complex_matmul.py:22

Functionconstant_fill

include/common.h:301

Functionconsumer_release

include/pipeline.h:217

Functionconsumer_wait

include/pipeline.h:208

include/tma.h:16

include/tma.h:36

include/tma.h:56

include/tma.h:77

Functioncopy_to_cpu

include/common.h:349

Functioncopy_to_cpu_async

include/common.h:358

Functioncopy_to_gpu

include/common.h:330

Functioncopy_to_gpu_async

include/common.h:339

util/simulator/kernel.py:44

Functiondouble_t

cutlass.py/dtype.py:215

cutlass.py/dtype.py:199

cutlass.py/dtype.py:203

Methodelapsed_millis

Return the elapsed time (in milliseconds)

examples/matmul/cutlass/helper.h:101

Methodexpect_transaction

Performs an expected transaction bytes increment without doing an arrive operation

include/barrier.h:327

(qr, qi, kr, ki, vr, vi)

examples/attention/triton/model.py:56

( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None,

examples/attention/triton/model.py:140

cutlass.py/mapping.py:7

Functionfree_cpu_tensor

include/common.h:320

Functionfree_gpu_tensor

include/common.h:325

Functionget_best_kernel

Adaptive kernel selection: compile both JIT and non-JIT variants, auto-tune by running both and picking the faster one. This handles the

examples/matmul/this-sm100/fp8_level6/test.py:254

Functionget_col_major_ind

(x, y, rows, cols)

cutlass.py/swizzle.py:41

Functionget_col_major_tuple

(xy, rows, cols)

cutlass.py/swizzle.py:44

Methodget_current_tile

include/scheduler.h:34

Methodget_current_tile_for_idx

include/scheduler.h:38

Methodget_grid_shape

( problem_shape_mnk, cta_shape, cluster_shape, hw_info, arguments,

cutlass.py/tile_scheduler.py:365

Functionget_sm_count

include/common.h:43

Methodget_static_dims_with_keys

cutlass.py/tiling.py:57

Methodget_static_dims_with_keys

cutlass.py/tiling.py:128

Methodget_tiled_cta_shape_mnl

(problem_shape_mnkl, cta_shape, cluster_shape)

cutlass.py/tile_scheduler.py:359

Functiongmma_selector

( A_dtype: DType, B_dtype: DType, C_dtype: DType, MNK_tiling: HyperCube, A_major: GmmaMajo

cutlass.py/mma.py:107

Functiongroup_gemm_fn

(group_A, group_B)

examples/moe_group_gemm/triton/grouped_gemm.py:162

Functiongrouped_matmul_kernel

( # device tensor of matrices pointers group_a_ptrs, group_b_ptrs, group_c_ptrs, # device

examples/moe_group_gemm/triton/grouped_gemm.py:83

Functiongrouped_matmul_kernel

( A, B, C, scatter_idx, expert_idx, M, N, K, E, num_valid_tokens,

examples/moe_group_gemm/triton/blocked_grouped_gemm.py:42

Methodhas_dynamic

cutlass.py/tiling.py:125

include/pipeline.h:95

Functioninitializer

examples/matmul/tvm/tensorir_meta_schedule.py:47

Functionint16_t

cutlass.py/dtype.py:167

Functionint32_t

cutlass.py/dtype.py:171

cutlass.py/dtype.py:159

Functionint64_t

cutlass.py/dtype.py:175

cutlass.py/dtype.py:163

Methodinvalidate

include/barrier.h:155

Methodis_double_t

cutlass.py/dtype.py:127

Methodis_final_split

(self, k_tiles_per_output_tile)

cutlass.py/tile_scheduler.py:252

Methodis_float_t

cutlass.py/dtype.py:124

cutlass.py/dtype.py:118

Methodis_half_t

cutlass.py/dtype.py:121

Methodis_int16_t

cutlass.py/dtype.py:23

Methodis_int32_t

cutlass.py/dtype.py:26

Methodis_int4_t

cutlass.py/dtype.py:17

Methodis_int64_t

cutlass.py/dtype.py:29

Methodis_int8_t

cutlass.py/dtype.py:20

Functionis_multicast_legal

Check if 2-CTA multicast is legal for given M. Requirements (matching DeepGEMM): 1. M ≥ 512 (enough work to benefit from multicast)

examples/matmul/this-sm100/fp8_level6/test.py:86

(self, another_type: "IntegerType")

cutlass.py/dtype.py:87

(self, another_type: "FloatingType")

cutlass.py/dtype.py:150

Functionis_same_row_or_col

include/pipeline.h:59

Functionis_same_type

(A_type: DType, B_type: DType)

cutlass.py/dtype.py:219

Methodis_uint16_t

cutlass.py/dtype.py:38

Methodis_uint32_t

cutlass.py/dtype.py:41

Methodis_uint4_t

cutlass.py/dtype.py:32

Methodis_uint64_t

cutlass.py/dtype.py:44

Methodis_uint8_t

cutlass.py/dtype.py:35

cutlass.py/tile_scheduler.py:245

Functionlinear_attention_bwd_kernel

( realQ, imageQ, realK, imageK, realV, imageV, realgO, imagegO, # realP, imageP, realgQ, imagegQ, real

examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm.py:20

Functionlinear_attention_bwd_kernel

( realQ, imageQ, realK, imageK, realV, imageV, realgO, imagegO, # realP, imageP, realgQ, imagegQ, real

examples/attention/triton/fused_linear_attention_complex_bwd_bind_bh.py:20

Functionlinear_attention_bwd_kernel

( realQ, imageQ, realK, imageK, realV, imageV, realgO, imagegO, # realP, imageP, realgQ, imagegQ, real

examples/attention/triton/fused_linear_attention_complex_bwd_bind_bhm_v2.py:21

Functionlinear_attention_fwd_kernel

( realQ, imageQ, realK, imageK, realV, imageV, realO, imageO, # realP, imageP, real_sm_scale, image_sm

examples/attention/triton/fused_linear_attention_complex.py:19

Functionlinear_attention_fwd_kernel

( realQ, imageQ, realK, imageK, realV, imageV, realO, imageO, # realP, imageP, real_sm_scale, image_sm

examples/attention/triton/fused_linear_attention_complex_tile_k_tile_n.py:20

← previous next →401–500 of 541, ranked by callers