Struct PagedKvStore 

pub struct PagedKvStore { /* private fields */ }

The other half of PagedAttention that KvBlockPool/KvCache::with_pool deliberately don’t implement (see this module’s doc comment): real, shared physical block storage that many sequences’ block tables can address into, instead of each KvCache still owning its own private, contiguous Vec. KvBlockPool only ever bounds a count of blocks each cache may grow to; PagedKvStore is the actual backing memory, and a sequence’s PagedKvCache holds a block table (an ordered list of block IDs into this shared store) instead of owning K/V data directly. This is what makes non-contiguous-block reads during attention (causal_gqa_attention_paged, in attention.rs) possible at all – causal_gqa_attention’s existing contiguous-slice read pattern has no way to express “position 37 lives in block 12, cached out of order relative to block 5.”

Implementations

impl PagedKvStore

pub fn new( block_size: usize, total_blocks: usize, n_kv_heads: usize, head_dim: usize, ) -> Self

pub fn block_size(&self) -> usize

pub fn free_block_count(&self) -> usize

pub fn n_kv_heads(&self) -> usize

pub fn head_dim(&self) -> usize

pub fn k_row(&self, id: usize, offset: usize) -> &[f32]

One position’s K (or V) row within block id at offset (0-based within the block) – [n_kv_heads * head_dim] long. Used by causal_gqa_attention_paged to read attention inputs directly out of shared physical storage via a block table, and by PagedKvCache::push to write a new position into it.

pub fn v_row(&self, id: usize, offset: usize) -> &[f32]