List of all items

Structs

• compiled::CompiledNode
• compiled::op_profile::LaunchRow
• compiled::op_profile::PhaseTimer
• decode_flash::DecodeFlashSpec
• decode_flash::FlashArmCapability
• dispatch::CapabilityRegistry
• dispatch::TransferMatrix
• driver::FrontierConvergenceOptimizer
• driver::OptimizationContext
• driver::PassRegistry
• driver::PlacementForkPathfinder
• fkc::AcceptBlock
• fkc::CapsBlock
• fkc::CostBlock
• fkc::CostEvalError
• fkc::CostMemory
• fkc::CpuLinkRegistry
• fkc::FdxSpec
• fkc::FkcFile
• fkc::FkcFrontMatter
• fkc::FkcKernel
• fkc::FkcProvider
• fkc::GatherSpec
• fkc::ImportWarning
• fkc::ImportedProvider
• fkc::LayoutSpec
• fkc::OpParamFieldSpec
• fkc::OpParamsSchema
• fkc::OutputDesc
• fkc::PrecisionBlock
• fkc::QuantSpec
• fkc::ResolvedFused
• fkc::ResolvedLayout
• fkc::ResolvedPrimitive
• fkc::ReturnBlock
• fkc::ShapeConstraint
• fkc::TensorDesc
• fused::BackendImpl
• fused::CostEstimate
• fused::FusedKernelRegistry
• fused::KernelRevisionHash
• fused::PrecisionGuarantee
• judge_cache::DispatchOptions
• judge_cache::DispatchTable
• judge_cache::Pick
• judge_cache::ProfileEntry
• judge_cache::ProfileReport
• judge_cache::SizeClass
• kernel::BindingEntry
• kernel::KernelBindingTable
• kernel::KernelCaps
• optimize::OptimizedGraph
• pipelined::PersistentOutputs
• pipelined::PipelinedExecutor
• plan::ExecutionPlan
• plan::PlanOptions
• ranker::alternative_set::AlternativeSet
• ranker::alternative_set::DecisionContext
• ranker::candidate::Candidate
• ranker::candidate::CouplingAdjustment
• ranker::chained_selector::ChainedSelector
• ranker::cost_vector::CostVector
• ranker::cost_vector::MemoryTiers
• ranker::device_load::DeviceLoadSelector
• ranker::device_load::MockBackendStreams
• ranker::device_load::MockCombinedRuntime
• ranker::filter::FilterContext
• ranker::filters::accuracy_pref::AccuracyBound
• ranker::filters::accuracy_pref::AccuracyPreferenceFilter
• ranker::filters::precision_floor::PrecisionFloorFilter
• ranker::filters::precision_floor::PrecisionRequirement
• ranker::filters::strided_input_pref::StridedInputPreferenceFilter
• ranker::judge::HashMapJudge
• ranker::judge_aware_selector::JudgeAwareSelector
• ranker::placement_dp::ChainInput
• ranker::placement_dp::PlacementDp
• ranker::profile_oracle::ProfileJudgeOracle
• ranker::route_picker::RouteCache
• ranker::route_picker::TelemetryFingerprint
• ranker::runtime_selector::WinnerSelector
• ranker::vram_pressure_selector::VramPressureSelector
• residency::EvictReload
• residency::LiveRange
• residency::ResidencyPlanner
• residency::ResidencyReport
• runtime_fused_arm::RuntimeFusedSpec
• runtime_fused_kernels::RuntimeKernelBinding
• runtime_fused_pathfinder::RuntimeFusedArmPathfinder
• topology::SystemTopology

Enums

• compiled::CompletionHandle
• fkc::AxisIndex
• fkc::CompiledCostExpr
• fkc::CostNode
• fkc::FkcError
• fkc::RankSpec
• fkc::Resolved
• fkc::ShapeAtom
• fkc::Tri
• fkc::shape_expr::Dim
• fkc::shape_expr::DimValue
• fkc::shape_expr::ShapeExpr
• fkc::shape_expr::ShapeExprError
• fkc::shape_expr::ShapeValue
• judge_cache::Criterion
• judge_cache::OpKind
• kernel::BindingKey
• kernel::MatmulM
• kernel::OpParams
• kernel::RuntimeFusedLookup
• ranker::cost_vector::AccuracyClass
• ranker::filter::FilterClass
• runtime_fused_kernels::FusedLookupMiss

Traits

• compiled::Completion
• driver::Optimizer
• driver::Pathfinder
• fkc::LinkRegistry
• ranker::cost::TransferEstimator
• ranker::filter::AlternativeFilter
• ranker::judge::JudgeOracle
• ranker::runtime_selector::RuntimeSelector

Macros

• register_fused

Functions

• cast_fusion::cast_fusion_predicate
• compiled::compile_node
• compiled::dispatched_kernel_ident
• compiled::execute_compiled
• compiled::execute_compiled_with_wait_hint
• compiled::op_profile::enabled
• compiled::op_profile::set_enabled
• compiled::op_profile::set_trace
• compiled::op_profile::take
• compiled::op_profile::take_phases
• compiled::op_profile::take_trace
• cost::cost_argindex_cpu
• cost::cost_cast_cpu
• cost::cost_causal_conv1d_primitive_cpu
• cost::cost_comparison_cpu
• cost::cost_concat_cpu
• cost::cost_conv2d_primitive_cpu
• cost::cost_conv_transpose2d_primitive_cpu
• cost::cost_elementwise_binary_cpu
• cost::cost_elementwise_unary_cpu
• cost::cost_elementwise_unary_transcendental_cpu
• cost::cost_flash_attn_primitive_cpu
• cost::cost_flash_decoding_cuda
• cost::cost_fused_linear_primitive_cpu
• cost::cost_fused_softmax_cross_entropy_primitive_cpu
• cost::cost_indexing_cpu
• cost::cost_masked_fill_cpu
• cost::cost_matmul_cpu
• cost::cost_nf4_matmul_primitive_cpu
• cost::cost_norm_last_dim_primitive_cpu
• cost::cost_paged_attn_primitive_cpu
• cost::cost_qmatmul_primitive_cpu
• cost::cost_reduce_max_to_backward_primitive_cpu
• cost::cost_reduce_to_cpu
• cost::cost_reduction_cpu
• cost::cost_rope_primitive_cpu
• cost::cost_scalar_op_cpu
• cost::cost_selective_scan_primitive_cpu
• cost::cost_shape_op_cpu
• cost::cost_softmax_last_dim_primitive_cpu
• cost::cost_ssd_chunk_scan_primitive_cpu
• cost::cost_where_cpu
• cost::default_cost_for_op_kind
• decode_flash::flash_decode_admissible
• decode_flash::offer_decode_flash_arm
• dispatch::bump_topology_generation
• dispatch::cpu_input
• dispatch::cpu_output
• dispatch::derive_backend_caps
• dispatch::extend_global_bindings
• dispatch::global_bindings
• dispatch::global_registry
• dispatch::inflight_count
• dispatch::inflight_dec
• dispatch::inflight_inc
• dispatch::read_storage
• dispatch::register_backend_capabilities
• dispatch::register_cpu_kernels
• dispatch::register_default_fused_kernels
• dispatch::resolve_target_backend
• dispatch::resolve_target_backend_residency_aware
• dispatch::topology_generation
• dispatch::write_storage
• fkc::bind_cost_symbols
• fkc::compute_revision
• fkc::cost_estimate
• fkc::eval_cost
• fkc::fused_unknown_cost
• fkc::import_bundle
• fkc::import_bundle_str
• fkc::import_glob
• fkc::is_generic_contract
• fkc::lower_file
• fkc::parse_file
• fkc::parse_path
• fkc::parse_rank_spec
• fkc::parse_shape_constraint
• fkc::shape_expr::decode_dim
• fkc::shape_expr::decode_shape
• fkc::shape_expr::eval_dim
• fkc::shape_expr::eval_shape
• fkc::shape_expr::gather_shape
• fkc::shape_expr::matmul_shape
• fkc::shape_expr::reduce_shape
• fkc::shape_expr::resolve_axis
• fkc::shape_expr::shape_consistent
• fkc::solve_probe_shapes
• fkc::validate_file
• fkc::validate_kernel
• fused::cost_attn_backward_cpu
• fused::cost_attn_cpu
• fused::cost_causal_conv1d_cpu
• fused::cost_conv2d_cpu
• fused::cost_conv_transpose2d_cpu
• fused::cost_fused_linear_cpu
• fused::cost_fused_softmax_cross_entropy_cpu
• fused::cost_inplace_affine_cpu
• fused::cost_nf4_matmul_cpu
• fused::cost_norm_family_cpu
• fused::cost_powi_backward_cpu
• fused::cost_qmatmul_cpu
• fused::cost_reduce_max_to_backward_cpu
• fused::cost_rope_cpu
• fused::cost_selective_scan_cpu
• fused::cost_ssd_chunk_scan_cpu
• fused::default_kernel_registry
• fused::register_default_kernels
• fused_cost::cost_from_decompose
• fused_cost::fused_layer1_cost
• fused_cost::is_fused_cost_sentinel
• judge_cache::cached
• judge_cache::cached_oracle
• judge_cache::invalidate
• judge_cache::store
• judge_report_path::default_report_path
• kernel::unknown_cost
• optimize::optimize_graph
• optimize::optimize_graph_with_runtime_fusion
• plan::backend_for_device
• plan::compile_plan
• plan::default_device_for
• plan::node_needs_plan_entry
• ranker::chain::apply_filter_chain
• ranker::cost::apply_inbound_transfer_costs
• ranker::cost::composite_ns
• ranker::cost::compute_static_costs
• ranker::cost::default_backend_rates
• ranker::device_load::load_tier
• ranker::device_load::load_tier_for
• ranker::device_load::mock_combined_lookup
• ranker::device_load::mock_load_lookup
• ranker::enumerate::enumerate_candidates
• ranker::filters::accuracy_pref::meets_accuracy_bound
• ranker::filters::default_chain
• ranker::route_picker::pick_route
• ranker::route_picker::resolve_branch
• ranker::vram_pressure_selector::default_estimate_output_bytes
• residency::insert_residency_evictions
• runtime_fused_arm::offer_runtime_fused_arm
• runtime_fused_kernels::adopt_runtime_fused
• runtime_fused_kernels::fused_kernel_available
• runtime_fused_kernels::fused_kernel_available_in
• runtime_fused_kernels::lookup_runtime_kernel
• runtime_fused_kernels::register_runtime_kernel
• runtime_fused_pathfinder::emit_runtime_fused_arms
• variant_bake::bake_variant_branches
• variant_bake::decode_arm_composite_ns
• variant_bake::decode_arm_composite_ns_judged
• variant_bake::variant_bakes_thread_local

Type Aliases

• fkc::ProbeCombo
• kernel::CostFn
• kernel::CoverageBinding
• kernel::KernelDTypes
• kernel::KernelRef
• pipelined::StorageCache
• ranker::cost::CapabilitiesLookup
• ranker::cost_vector::PrecisionDigits
• ranker::device_load::LoadLookup
• ranker::device_load::MockCombinedEntry
• ranker::vram_pressure_selector::BackendRuntimeHandle
• ranker::vram_pressure_selector::BackendRuntimeLookup
• ranker::vram_pressure_selector::OutputBytesEstimator
• variant_bake::ArmCostFn

Statics

• fkc::CPU_AFFINE_CLAMP_POWI_ENTRY_POINTS
• fkc::CPU_ATTENTION_ENTRY_POINTS
• fkc::CPU_BINARY_ENTRY_POINTS
• fkc::CPU_COMPARE_ENTRY_POINTS
• fkc::CPU_CONV_ENTRY_POINTS
• fkc::CPU_FUSED_CONV_ROPE_ENTRY_POINTS
• fkc::CPU_FUSED_LINEAR_QUANT_ENTRY_POINTS
• fkc::CPU_FUSED_NORM_ENTRY_POINTS
• fkc::CPU_INPLACE_ENTRY_POINTS
• fkc::CPU_MATMUL_ENTRY_POINTS
• fkc::CPU_NORM_BACKWARD_ENTRY_POINTS
• fkc::CPU_NORM_ENTRY_POINTS
• fkc::CPU_PADDING_ENTRY_POINTS
• fkc::CPU_REDUCE_ENTRY_POINTS
• fkc::CPU_REDUCE_TO_ENTRY_POINTS
• fkc::CPU_ROPE_ENTRY_POINTS
• fkc::CPU_SHAPE_OPS_ENTRY_POINTS
• fkc::CPU_SSM_ENTRY_POINTS
• fkc::CPU_UNARY_ENTRY_POINTS
• fkc::CPU_WHERE_ENTRY_POINTS

Constants

• fkc::FKC_SUPPORTED_VERSIONS
• fkc::shape_expr::LAST
• fkc::shape_expr::SYMBOLIC
• fkc::shape_expr::TAG_ADD
• fkc::shape_expr::TAG_CONST
• fkc::shape_expr::TAG_DIMS
• fkc::shape_expr::TAG_DIV
• fkc::shape_expr::TAG_EXTENT
• fkc::shape_expr::TAG_MUL
• fkc::shape_expr::TAG_PARAM
• fkc::shape_expr::TAG_REDUCE
• fkc::shape_expr::TAG_SAME_AS
• fkc::shape_expr::TAG_SUB
• fkc::shape_expr::TAG_WITH_DIM
• fused::ATTN_BACKWARD_CPU_PRECISION
• fused::ATTN_CPU_PRECISION
• fused::CAUSAL_CONV1D_CPU_PRECISION
• fused::CONV2D_CPU_PRECISION
• fused::CONV_TRANSPOSE2D_CPU_PRECISION
• fused::FUSED_LINEAR_CPU_PRECISION
• fused::FUSED_SOFTMAX_CROSS_ENTROPY_CPU_PRECISION
• fused::INPLACE_AFFINE_CPU_PRECISION
• fused::NF4_MATMUL_CPU_PRECISION
• fused::NORM_FAMILY_CPU_PRECISION
• fused::POWI_BACKWARD_CPU_PRECISION
• fused::QMATMUL_CPU_PRECISION
• fused::REDUCE_MAX_TO_BACKWARD_CPU_PRECISION
• fused::ROPE_CPU_PRECISION
• fused::SELECTIVE_SCAN_CPU_PRECISION
• fused::SSD_CHUNK_SCAN_CPU_PRECISION
• fused::VULKAN_BYTE_LEVEL_PRECISION
• fused::VULKAN_CAST_PRECISION
• fused::VULKAN_FLOAT_POINTWISE_PRECISION
• fused::VULKAN_HALF_POINTWISE_PRECISION
• fused::VULKAN_MATMUL_PRECISION
• fused::VULKAN_MATMUL_TENSORCORE_PRECISION
• fused::VULKAN_QMATMUL_PRECISION
• fused::VULKAN_TRANSCENDENTAL_PRECISION
• judge_cache::DEFAULT_ACCURACY_PENALTY
• judge_report_path::PROFILE_REPORT_FILENAME
• ranker::alternative_set::KEEP_PER_DEVICE
• ranker::cost::CPU_COMPUTE_FLOPS_PER_NS
• ranker::cost::CPU_MEM_BYTES_PER_NS
• ranker::cost::GPU_COMPUTE_FLOPS_PER_NS
• ranker::cost::GPU_MEM_BYTES_PER_NS
• ranker::device_load::LOAD_TIER_IDLE
• ranker::device_load::LOAD_TIER_MODERATE
• ranker::device_load::LOAD_TIER_SATURATED
• ranker::filters::accuracy_pref::DEFAULT_ACCURACY_BOUND
• ranker::route_picker::LOOKAHEAD_K