qualia_core_db/inference/runtime/kv/paged/mod.rs
1//! Fixed-capacity paged KV planning.
2//!
3//! Construction may allocate bounded vectors. Append, fork, copy-on-write, release, and GPU table
4//! export never grow them, so the decode and scheduling paths do not touch the global allocator.
5
6mod attention;
7mod config;
8mod pool;
9mod table;
10
11pub use config::{PagedKvConfig, DEFAULT_BLOCK_SIZE, INVALID_BLOCK};
12pub use pool::{BlockPool, CopyOnWrite, PoolError};
13pub use table::{fill_identity_block_table, GpuBlockTablePlan, SequenceBlockTable, TableError};
14
15#[cfg(test)]
16mod tests;
17pub use attention::{paged_gqa_attention_into, AttentionError};