List of all items
Structs
- BarrierToken
- BindingManifest
- Buffer
- BufferHandle
- CacheBuster
- DoubleBuffer
- Graph
- HirReflection
- IoBindingEntry
- KernelDispatchConfig
- LocalTransport
- ManifestDiff
- ModelComponent
- ModelVariant
- MoeResidencyStats
- MoeTopkCapture
- Node
- NodeId
- PipelineInspect
- Rank
- RlxEnv
- RuntimeOverrides
- Shape
- SymmetricBuffer
- SymmetricHeap
- SyncCopy
- SyncStream
- Tick
- WeightBlock
- aot_cache::AotCache
- backend::cpu_backend::CpuBackend
- backend::wgpu_backend::WgpuBackend
- compile_cache::BucketedCompileCache
- compile_cache::CacheRunInput
- compile_cache::CompileCache
- compile_cache::DynamicDimCompileCache
- compiled::CompiledGraph
- cost::CpuCostModel
- env::RlxEnv
- env::RuntimeOverrides
- expert_pool::ExpertPool
- expert_pool::ExpertPoolConfig
- expert_pool::ExpertPoolStats
- expert_pool::ExpertRefreshResult
- hwinfo::HwSnapshot
- jacfwd::JacobianBytes
- kv_cache::LayerKvCache
- logit_verify::Diff
- logit_verify::Tolerance
- lora_scheduler::LoraBatch
- lora_scheduler::LoraPayload
- lora_scheduler::LoraRequest
- lora_scheduler::LoraScheduler
- lora_scheduler::UnknownAdapter
- memory_estimate::MemoryDeficit
- memory_estimate::MemoryEstimate
- memory_estimate::MoeOffloadEstimate
- mock_requests::ChatCompletionRequest
- mock_requests::ChatMessage
- mock_requests::EmbeddingRequest
- model_pipeline::ModelCompilePipeline
- moe_expert_store::ExpertStackF32
- moe_expert_store::LayerMoeWeights
- moe_expert_store::MoeExpertStore
- nan_check::NanCheckError
- op_registry::OpRegistry
- op_registry::VjpContext
- options::CompileOptions
- paged_kv::BatchConstructor
- paged_kv::BatchEntry
- paged_kv::KvBlockTable
- paged_kv::KvPageDesc
- paged_kv::KvPageId
- paged_kv::KvPagePool
- perfetto::TraceSpan
- phase::PhaseSchedule
- record_replay::RecordedExchange
- record_replay::RecordingWriter
- record_replay::ReplayReader
- reflect::ModelReflection
- router::OpenAIProtocol
- router::RoutedRequest
- session::Session
- spec_decode::AcceptDecision
- spec_decode::DraftProposal
- spec_decode::SpecDecoder
- spec_decode::VerifyResult
- subgraph::SubgraphCache
- telemetry::Counter
- telemetry::Histogram
- telemetry::HistogramSnapshot
- telemetry::MetricsRegistry
- telemetry::Snapshot
- trace::TracedTensor
- trace::Tracer
- validators::ChannelsAllowed
- validators::ImageContext
- validators::ImageMaxBounds
- validators::MaxBatchSize
- validators::MaxSeqLen
- validators::TextContext
- validators::TokenIdsInVocab
- validators::ValidationError
- weight_registry::WeightEntry
- weight_registry::WeightHandle
- weight_registry::WeightRegistry
- weights::BytesWeightLoader
- worker_pool::InProcessWorker
- worker_pool::WorkerHealth
- worker_pool::WorkerId
- worker_pool::WorkerPool
Enums
- CollectiveError
- CompilationMode
- DType
- Device
- KernelDispatchPolicy
- ModelPhase
- Op
- OpKind
- PrecisionPolicy
- ReduceKind
- aot_cache::AotCacheError
- expert_pool::ExpertRefreshPolicy
- expert_pool::MoEExecMode
- hwinfo::ThermalState
- logit_verify::VerifyError
- mock_requests::Input
- nan_check::BadValue
- op::Activation
- op::AttentionBwdWrt
- op::BinaryOp
- op::ChainOperand
- op::ChainStep
- op::CmpOp
- op::MaskKind
- op::Op
- op::OpKind
- op::ReduceOp
- op::ScaleMode
- op::SteKind
- paged_kv::BatchKind
- phase::Phase
- precision::Precision
- router::OpenAIRequest
- router::RequestKind
- router::RouteError
- weight_registry::WeightKind
- worker_pool::DispatchPolicy
- worker_pool::WorkerError
Traits
- AsyncCopy
- CommandStream
- DeviceArena
- SymmetricTransport
- backend::Backend
- backend::ExecutableGraph
- cost::BackendCostModel
- op_registry::OpExtension
- router::WireProtocol
- spec_decode::Speculator
- validators::Validator
- weights::WeightLoader
- worker_pool::Worker
Macros
Attribute Macros
Functions
- all_gather
- all_reduce
- apply_hir_extensions
- attn_mask::bucket_decode_mask
- backend::compile
- backend::compile_hir
- backend::compile_module
- backend::compile_with_precision
- compile_cache::pad_rows
- compile_cache::slice_rows
- cost::estimate_graph_cost
- cost::pick_best_device
- custom_ops::execute
- custom_ops::register
- custom_ops::registered
- device_ext::available_devices
- device_ext::dispatch_report_for_device
- device_ext::dispatch_report_for_device_with_options
- device_ext::first_unsupported_op
- device_ext::first_unsupported_op_with_options
- device_ext::full_name
- device_ext::is_available
- device_ext::legalize_graph_for_device
- device_ext::legalize_graph_for_device_with_options
- device_ext::legalize_graph_for_device_with_report
- device_ext::supports
- device_ext::supports_graph
- device_ext::supports_graph_with_options
- env::clear_overrides
- env::flag
- env::is_unset
- env::normalize_key
- env::parse_or
- env::set
- env::unset
- env::var
- env::var_os
- expert_pool::gpu_expert_budget_from_vram
- expert_pool::merged_resident_mask
- expert_pool::per_layer_resident_masks
- inspect_graph
- inspect_hir
- inspect_hir_stats
- inspect_lir
- inspect_mir
- inspect_mir_stats
- inspect_pipeline
- jacfwd::jacfwd
- logit_verify::compare
- logit_verify::diff
- lora_scheduler::naive_swap_count
- memory_estimate::available_unified_memory
- memory_estimate::estimate
- memory_estimate::estimate_moe_offload
- mock_requests::chat_simple
- mock_requests::chat_streaming
- mock_requests::chat_system_user
- mock_requests::embed_batch
- mock_requests::embed_single
- nan_check::scan
- op_registry::global_registry
- op_registry::lookup_op
- op_registry::register_op
- perfetto::emit_complete
- perfetto::enabled
- perfetto::flush_and_finalize
- phase::derive_phases
- reduce_scatter
- reflect::load_hir_template_with_extensions
- reflect::specialize_entry
- register_hir_extension
- registered_hir_extensions
- registry::backend_for
- registry::register_backend
- registry::registered_devices
- spec_decode::speculative_accept
- stages::compile_graph_stages
- stages::compile_graph_stages_for_backend
- stages::compile_hir_stages
- stages::compile_module_stages
- stages::fusion_target_for
- stages::graph_from_lir
- stages::maybe_log_fusion
- stages::options_with_supported_ops
- stages::pipeline_for
- subgraph::run_if
- subgraph::run_while
- time_ns
- trace::trace
- validators::run_chain