List of all items

Structs

• analysis::cc_overlap::OverlapStats
• environment::device::Devices
• environment::device::ReturnDevices
• input::torch_graph::TorchGraphImporter
• model::model::Layer
• model::model::Model
• model::model_perf::ModelPerf
• model::model_perf::ModelState
• orchestration::orchestrate_async::AsyncOrchestrate
• orchestration::orchestrate_async::AsyncOrchestrateResult
• orchestration::orchestrate_async::MatrixCell
• orchestration::orchestrate_hierarchical::HierarchicalConductor
• orchestration::orchestrate_sync::MatrixCell
• orchestration::orchestrate_sync::SyncOrchestrate
• orchestration::orchestrate_sync::SyncOrchestrateResult

Enums

• environment::device::AllocationStrategy

Traits

• input::ModelImporter
• orchestration::Orchestrate
• orchestration::OrchestrationResult

Functions

• analysis::cc_overlap::cc_overlap_partial
• analysis::cc_overlap::p3
• analysis::gpu_memory::max_single_gpu_batch_size
• api::pylib::PyInit_HPGO
• parallelism::data_parallel::all_reduce_time
• parallelism::data_parallel::dp_p3_speedup
• parallelism::data_parallel::dp_speedup
• parallelism::data_parallel::dp_speedup_strong
• parallelism::data_parallel::dp_speedup_weak
• parallelism::gpipe::gpipe_pipeline_speedup
• parallelism::gradient_accumulation::current_ga_iter_size
• parallelism::gradient_accumulation::dp_cur_ga_inner_overlap_speedup
• parallelism::gradient_accumulation::dp_cur_ga_p3_speedup
• parallelism::gradient_accumulation::dp_ga_overlap_speedup
• parallelism::gradient_accumulation::dp_ga_speedup
• parallelism::gradient_accumulation::dp_opt_ga_p3_speedup
• parallelism::gradient_accumulation::optimal_ga_iter_size
• parallelism::split_concat::split_concat_all2all_time
• parallelism::split_concat::split_concat_time
• parallelism::sync_pipeline::sync_pipeline_length_intermediate
• parallelism::sync_pipeline::sync_pipeline_speedup_analytical
• parallelism::sync_pipeline::sync_pipeline_speedup_recursive

Typedefs

• model::model_perf::ModelStates
• orchestration::orchestrate_async::Matrix
• orchestration::orchestrate_async::bitset
• orchestration::orchestrate_hierarchical::Matrix
• orchestration::orchestrate_sync::Matrix
• orchestration::orchestrate_sync::bitset

Constants

• environment::device::GPU_MEMORY
• environment::ethernet::BANDWIDTH_ETHERNET_NCCL
• environment::ethernet::BANDWIDTH_ETHERNET_P2P
• environment::network::GIGABYTE
• environment::nvlink::BANDWIDTH_NVLINK_P2P