List of all items

Structs

• attention::YarnScaling
• bench_profile::Measured
• block_sparse::BlockSparseConfig
• cache::KvBlockPool
• cache::KvCache
• cache::KvPoolExhausted
• cache::PageGroup
• cache::PagedKvCache
• cache::PagedKvStore
• cache::PagedStoreExhausted
• cache::SharedPagedKv
• expert_budget::BudgetTooSmall
• expert_budget::PoolSizes
• expert_cache::BankEntry
• expert_cache::CopyPlan
• expert_cache::EnsurePlan
• expert_cache::ExpertCache
• expert_cache::ExpertCacheStats
• expert_cache::ExpertId
• expert_cache::GatherPlan
• expert_cache::LayerRoutingSkew
• expert_cache::MissRun
• expert_cache::PrefillPlan
• expert_cache::RebuildRejected
• expert_cache::RoutingSkewReport
• expert_slots::Applied
• expert_slots::ExpertSlots
• expert_slots::HostSlotMemory
• expert_slots::SlotGeometry
• expert_slots::SlotStats
• expert_store::ExpertKey
• expert_store::ExpertLease
• expert_store::ExpertStore
• expert_store::ExpertStoreStats
• expert_store::FileRangeSource
• instance::InstanceConflict
• instance::InstanceGuard
• instance::InstanceInfo
• kernel_registry::Entry
• kernel_registry::Key
• kernel_registry::Lookup
• kernel_registry::Registry
• kernel_registry::SealReport
• kernel_registry::StrictKernelError
• kv_block::BlockHash
• kv_block::BlockHasher
• kv_disk::DecodedBlock
• kv_disk::DiskConfig
• kv_disk::DiskKvStore
• kv_disk::DiskStats
• kv_disk::ReadHandle
• kv_signature::CacheSignature
• kv_signature::KvBlock
• kv_signature::UnverifiedBlock
• kv_swa::BlockLayout
• qstar::BandwidthProfile
• qstar::KernelBandwidths
• qstar::ProfileGpu
• qstar::QStarPolicy
• qstar::QStarSplit
• qstar::Workload
• residency::BankResidency
• residency::PinBudgetEnvError
• residency::ResidencyPlan
• tensor::Tensor
• threads::CpuPoolPlan
• threads::CpuTopology

Enums

• bench_profile::NotMeasurable
• expert_cache::FetchOrder
• expert_slots::SlotFault
• host_memory::FitPlan
• instance::InstancePolicy
• kernel_registry::Backend
• kernel_registry::Outcome
• kernel_registry::Phase
• kernel_registry::Severity
• kv_disk::BlockFormatError
• kv_disk::StoreError
• kv_signature::KvDtype
• kv_signature::SignatureError
• kv_swa::BlockLayoutError
• placement::CpuLayerSpecError
• qstar::MoeBackend
• residency::CopyRoute
• residency::HostResidency
• residency::ResidencyError
• residency::SettleAction
• weight_matrix::BatchActs
• weight_matrix::QuantKind
• weight_matrix::WeightBytes
• weight_matrix::WeightMatrix

Traits

• expert_slots::ExpertRows
• expert_slots::SlotDevice
• expert_store::ExpertSource

Functions

• attention::apply_rope
• attention::apply_rope_back
• attention::apply_rope_interleaved
• attention::apply_rope_interleaved_back
• attention::apply_rope_interleaved_with_freq_factors
• attention::apply_rope_with_freq_factors
• attention::causal_gqa_attention
• attention::causal_gqa_attention_paged
• attention::causal_gqa_attention_paged_sinks
• attention::causal_gqa_attention_prefill
• attention::causal_gqa_attention_prefill_shared_kv
• attention::causal_gqa_attention_prefill_shared_kv_windowed
• attention::causal_gqa_attention_sinks
• attention::causal_gqa_attention_softcap
• attention::causal_gqa_attention_windowed
• attention::causal_gqa_attention_windowed_softcap
• attention::causal_mla_attention
• attention::causal_mla_attention_sinks
• attention::causal_mla_attention_sparse
• attention::causal_mla_attention_sparse_sinks
• attention::lightning_indexer_topk
• attention::proportional_freq_factors
• attention::yarn_correction_range
• attention::yarn_freq_factors
• bench_profile::bench_format
• bench_profile::cache_dir
• bench_profile::default_profile_path
• bench_profile::default_profile_path_in
• bench_profile::entry_from
• bench_profile::env_profile_path
• bench_profile::latest_profile_path
• bench_profile::latest_profile_path_in
• bench_profile::load_backend_recommendation
• bench_profile::load_hybrid_fetch_fraction
• bench_profile::load_policy
• bench_profile::read_profile
• bench_profile::usable_profile
• bench_profile::usable_profile_in
• bench_profile::write_profile
• block_sparse::block_sparse_select
• block_sparse::positions_of_blocks
• csa_hca_compress::channel_gated_pool
• csa_hca_compress::compress_block
• deepseek_v4_attention::csa_attention
• deepseek_v4_attention::hca_attention
• expert_budget::expert_bytes_per_slot
• expert_budget::net_cache_budget_bytes
• expert_budget::plan_cache_budget
• expert_budget::required_bytes
• expert_budget::startup_kv_budget
• expert_pool::split_pair
• host_memory::available_bytes
• host_memory::parse_vm_stat
• host_memory::plan_for
• instance::current_backend
• instance::live_instances
• instance::register
• instance::registry_dir
• kernel_registry::enabled
• kernel_registry::global
• kernel_registry::hit
• kernel_registry::miss
• kernel_registry::miss_by_design
• kernel_registry::record_build
• kernel_registry::seal
• kernel_registry::seal_or_error
• kernel_registry::strict
• kernel_registry::verbose
• kv_block::full_blocks
• kv_disk::decode_block
• kv_disk::encode_block
• kv_disk::encoded_len
• kv_swa::aligned_block_size
• matmul::geglu
• matmul::gelu
• matmul::layer_norm
• matmul::matmul_f32
• matmul::rms_norm
• matmul::rms_norm_per_head
• matmul::silu
• matmul::situ_and_mul
• matmul::softcap_inplace
• matmul::swiglu
• placement::auto_cpu_layers
• placement::parse_cpu_layers_spec
• placement::round_half_even
• placement::strided_layers
• qstar::balanced_fetch
• qstar::fetch_fraction_from_bandwidths
• qstar::fetch_fraction_from_overlap
• qstar::recommend_backend
• residency::host_kernel_release
• residency::host_phys_ram_bytes
• residency::host_pin_budget_bytes
• residency::is_pin_capped_host
• residency::parse_pin_budget_gb
• residency::pin_budget_bytes
• residency::requested_labels
• residency::resolve_pin_budget
• summary_stats::mean_of_present
• summary_stats::percentile
• threads::clamp_intra_op_threads
• threads::current_qos_name
• threads::for_each_chunk_init
• threads::for_each_row
• threads::init_cpu_pool
• threads::parse_thread_siblings_list
• threads::perf_core_count
• threads::physical_core_count
• threads::physical_core_cpus
• threads::physical_core_cpus_in
• threads::plan_cpu_pool
• threads::plan_cpu_pool_in
• threads::process_affinity_cpus
• threads::resolve_cpu_threads
• threads::resolve_threads_and_affinity
• threads::resolve_threads_and_affinity_in
• threads::should_parallelize
• turboquant::dequantize_turbo4_groups
• turboquant::fwht_inplace
• turboquant::fwht_orthonormal_inplace
• turboquant::quantize_turbo4_groups
• weight_matrix::active_backend
• weight_matrix::cpu_int_dot_enabled
• weight_matrix::cpu_int_dot_kind_supported
• weight_matrix::cuda_matvec_kind_supported
• weight_matrix::cuda_mul_mm_kind_supported
• weight_matrix::default_cpu_int_dot_on
• weight_matrix::metal_matvec_kind_name
• weight_matrix::metal_mul_mm_kind_supported
• weight_matrix::quant_kind_for

Type Aliases

• kv_disk::FreeSpaceProbe
• kv_disk::ReadOutcome

Constants

• bench_profile::LEGACY_PROFILE_FILE
• bench_profile::PROFILE_PATH_ENV
• bench_profile::PROFILE_SUBDIR
• block_sparse::MINIMAX_BLOCK_SIZE
• expert_cache::SMALL_BANK_FEAT_BYTES
• kernel_registry::op::ENGINE_PREFILL_BATCH
• kernel_registry::op::FFN_SWIGLU
• kernel_registry::op::GEMM_PREFILL
• kernel_registry::op::MATVEC
• kernel_registry::op::MATVEC_MULTI
• kv_disk::BLOCK_FILE_EXT
• kv_signature::BLOCK_FORMAT_VERSION
• kv_signature::READABLE_FORMAT_VERSIONS
• qstar::DEFAULT_RECOMMEND_THRESHOLD
• qstar::FRACTION_ONE
• residency::PIN_BUDGET_ENV
• residency::WSL_KERNEL_TAG
• residency::WSL_PIN_FRACTION
• threads::SYSFS_CPU_ROOT
• vexp::EXP_C0
• vexp::EXP_C1
• vexp::EXP_C2
• vexp::EXP_C3
• vexp::EXP_C4
• vexp::EXP_LN2_HI
• vexp::EXP_LN2_LO
• vexp::EXP_LOG2E
• vexp::EXP_SHIFT