Types & classes — github.com/bytedance/flux

CodeHub MCPcopy Create free account

hub / github.com/bytedance/flux / types & classes

Types & classes309 in github.com/bytedance/flux

⨍ Functions 784◇ Types & classes 309

↓ 13 callersClassNotCompiled

python/flux/cpp_mod.py:52

↓ 4 callersClassElementD

include/flux/cuda/cutlass_v3_builder.hpp:209

↓ 2 callersClassPerfResult

test/test_ag_kernel_pyshmem.py:58

↓ 2 callersClassPerfResult

test/test_gemm_rs.py:61

↓ 2 callersClassPerfResult

test/test_ag_kernel.py:59

↓ 2 callersClassPerfResult

test/test_ag_kernel_crossnode.py:60

↓ 1 callersEnumArchEnum

include/flux/flux.h:455

↓ 1 callersClassClangFunction

pynvshmem/tools/autogen_pybind.py:119

↓ 1 callersClassConsumerStoreCallbacks

src/reduce_scatter/sm90_epilogue_evt.hpp:190

↓ 1 callersClassDistEnv

python/flux/dist_utils.py:26

↓ 1 callersEnumDmaWarpRole

src/reduce_scatter/sm90_gemm_tma_warpspecialized_cooperative_reduce_scatter.hpp:375

↓ 1 callersEnumGemmKernelScheduleEnum

include/flux/flux.h:476

↓ 1 callersClassPerfResult

test/test_gemm_only.py:40

↓ 1 callersEnumProducerWarpRole

src/all_gather/sm90_all_gather_gemm_tma_warpspecialized_cooperative.hpp:426

↓ 1 callersClassPyTuningRecord

Wraps c++ types in class holder, in order to communicate with python

include/flux/ths_op/ths_op.h:47

↓ 1 callersClassTuningConfig

tools/tune_ag_gemm_kernel.py:58

↓ 1 callersClassTuningConfig

tools/tune_gemm_rs.py:57

↓ 1 callersEnumWarpGroupRole

In the Cooperative kernel, Consumer0 and Consumer1 collaborate on the same tile */

src/all_gather/sm90_all_gather_gemm_tma_warpspecialized_cooperative.hpp:425

↓ 1 callersEnumWarpGroupRole

In the Cooperative kernel, Mma0 and Mma1 collaborate on the same tile */

src/reduce_scatter/sm90_gemm_tma_warpspecialized_cooperative_reduce_scatter.hpp:374

ClassAGFP8KernelArguments

include/flux/args/all_gather.h:38

python/flux/ag_gemm.py:68

All Gather GEMM Kernel OP

src/all_gather/ths_op/all_gather_gemm_kernel.cc:78

ClassAGKernelArguments

include/flux/args/all_gather.h:21

ClassAGKernelCrossNode

All Gather GEMM Kernel OP

src/all_gather/ths_op/all_gather_gemm_kernel_crossnode.cc:105

ClassAGKernelTileScheduler

src/all_gather/gemm_v3_ag_kernel.hpp:33

ClassAGKernelXNode

python/flux/ag_kernel_crossnode.py:85

All2All for nvlink mode. for NVLINK machine, default is 0 Ring1D for 1d-ring. for PCI-e machine without GPUs cross NUMA nodes use ring 1d Ring2D for 2

src/all_gather/ths_op/all_gather_types.h:11

ClassAGThreadblockSwizzleStreamKRankOffset

src/all_gather/sm80_all_gather_gemm_threadblock_swizzle.hpp:33

ClassAppendCloseIpcMemHandleCallbackRAII

src/ths_op/ths_op.cc:95

src/all_gather/sm90_all_gather_gemm_tile_scheduler.hpp:50

src/all_gather/sm80_all_gather_gemm.hpp:105

src/all_gather/sm90_all_gather_gemm_tma_warpspecialized_cooperative.hpp:176

src/reduce_scatter/epilogue_reduce_scatter.hpp:109

src/reduce_scatter/reduce_scatter_kernel.hpp:1917

src/reduce_scatter/epilogue_evt.hpp:103

src/reduce_scatter/epilogue_vectorized_reduce_scatter.hpp:118

src/reduce_scatter/epilogue_evt_nvshmem.hpp:84

src/reduce_scatter/epilogue_nvshmem_reduce_scatter.hpp:130

src/reduce_scatter/gemmk_universal_with_visitor_streamk.h:133

src/reduce_scatter/sm90_epilogue_evt.hpp:123

src/reduce_scatter/sm90_gemm_tma_warpspecialized_cooperative_reduce_scatter.hpp:143

src/reduce_scatter/sm90_reduce_scatter_utils.hpp:95

src/reduce_scatter/gemmk_visitor_load.hpp:64

src/reduce_scatter/tile_scheduler/sm90_tile_scheduler_reduce_scatter.hpp:38

src/reduce_scatter/tile_scheduler/sm90_tile_scheduler_reduce_scatter.hpp:112

include/flux/flux.h:530

ClassBarrierWorkQeueuFlags

src/reduce_scatter/reduce_scatter_barrier_struct.hpp:82

ClassBarrierWorkQeueuFlagsAoSWrapper

src/reduce_scatter/reduce_scatter_barrier_struct.hpp:198

ClassBaseEpilogueParams

include/flux/cuda/cutlass_v3_builder.hpp:212

ClassBaseParams

include/flux/cuda/cutlass_v3_builder.hpp:162

ClassBaseParams

include/flux/cuda/cutlass_v3_builder.hpp:252

ClassBaseParams

include/flux/cuda/cutlass_v3_builder.hpp:273

include/flux/cuda/cuda_stub.h:41

ClassCachedWheelsCommand

The CachedWheelsCommand plugs into the default bdist wheel, which is ran by pip when it cannot find an existing wheel (which is currently the

src/reduce_scatter/epilogue_evt.hpp:186

src/reduce_scatter/epilogue_evt_nvshmem.hpp:163

src/reduce_scatter/gemmk_visitor_load.hpp:107

include/flux/flux.h:90

include/flux/gemm_meta.h:223

EnumCommKindEnum

include/flux/flux.h:471

include/flux/flux.h:456

ClassCommSpecifics

include/flux/gemm_hparams.h:240

ClassCopyAtomR2G

include/flux/cuda/cutlass_v3_builder.hpp:257

ClassCopyAtomR2S

include/flux/cuda/cutlass_v3_builder.hpp:255

include/flux/cuda/cutlass_v3_builder.hpp:278

include/flux/cuda/cutlass_v3_builder.hpp:283

include/flux/cuda/cutlass_v3_builder.hpp:281

include/flux/cuda/cutlass_v3_builder.hpp:280

ClassCopy_Traits<SM90_TMA_STORE_ADD, NumBitsPerTMA, AuxParams_>

include/flux/cuda/memory_utils.hpp:371

ClassCustomizedGenericBarrier

include/flux/cuda/system_barrier.hpp:140

include/flux/gemm_meta.h:80

include/flux/gemm_meta.h:80

include/flux/gemm_meta.h:80

include/flux/gemm_meta.h:80

include/flux/gemm_meta.h:80

EnumDataTypeEnum

//////////////////////////////////////////////// Enum classes ////////////////////////////////////////////////

include/flux/flux.h:454

ClassDispatchPolicy

include/flux/cuda/cutlass_v3_builder.hpp:163

ClassDispatchPolicy

include/flux/cuda/cutlass_v3_builder.hpp:274

include/flux/utils.h:36

include/flux/ths_op/ths_op.h:125

ClassDistEnvTPWithEP

include/flux/ths_op/ths_op.h:131

ClassDynamicLibrary

src/cuda/nvml_stub.cc:72

ClassDynamicLibrary

src/cuda/cuda_stub.cc:64

ClassElementAccumulator

include/flux/cuda/cutlass_v3_builder.hpp:211

ClassEpilogueReduceScatter

src/reduce_scatter/epilogue_reduce_scatter.hpp:79

ClassEpilogueReduceScatterVectorized

src/reduce_scatter/epilogue_vectorized_reduce_scatter.hpp:85

ClassEpilogueReduceScatterVectorizedNvshmemLocalReduce

src/reduce_scatter/epilogue_nvshmem_reduce_scatter.hpp:97

ClassEpilogueTileMN

include/flux/cuda/cutlass_v3_builder.hpp:276

ClassFluxGemmKernel

include/flux/gemm_operator_base.h:44

EnumFluxNamedBarriers

include/flux/cuda/memory_utils.hpp:33

ClassFluxNamedTupleBase

include/flux/flux.h:935

ClassFuseReduction

include/flux/gemm_meta.h:223

ClassFusionCallbacks

include/flux/cuda/cutlass_v3_builder.hpp:277

ClassGatherRSCTAs

include/flux/gemm_hparams.h:130

ClassGatherRSHParams

include/flux/gemm_hparams.h:113

ClassGatherRSMeta

include/flux/gemm_meta.h:236

ClassGemmDTypeConfig

include/flux/gemm_meta.h:27

ClassGemmFP8Arguments

FP8 GEMM Aux = ((alpha * scale_a * scale_b) * accumulator) + ((beta * scale_c) * source) + bias D = activation(Aux) if Aux is fp8: abs_max_output = ma

include/flux/args/comm_none.h:44

ClassGemmGroupedImpl

include/flux/cuda/gemm_impls/gemm_grouped_impl.hpp:54

ClassGemmGroupedOpArguments

include/flux/args/comm_none.h:66

next →1–100 of 309, ranked by callers