mlx-bun 0.4.0 → 1.0.0

package/src/backends/mlx/batch-group.ts

@@ -1,1790 +0,0 @@
-import type { OrdinaryContinuation } from "./continuation";
-import type { MlxRequestStatePolicy } from "./request-state-policy";
-import { AdmissionRejected } from "../../engine/admission";
-import type { ExecutionContext, ExecutionGroup } from "../../contracts/scheduling";
-import { driveExecutionGroup } from "../../engine/scheduler";
-import { disposeResources, withResource } from "../../engine/resources";
-import type { DisposableResource } from "../../contracts/resources";
-import { MlxPrefillCohort, type PrefillState } from "./prefill-cohort";
-// Continuous-batching scheduler for `--batch N` serving (phase S2, the engine
-// loop). Owns ONE running batch and drives it forward one decode step at a
-// time, admitting waiting requests and evicting finished ones between steps —
-// iteration-level (continuous) scheduling, not static batching. See
-// docs/design/batching.md and docs/design/batching.md.
-//
-// The numerically-hard parts are verified primitives:
-//   - the batched FORWARD (per-row RoPE/mask) is bit-parity with mlx-lm B=N
-//     across all 4 models (tests/batched-decode-parity);
-//   - the dynamic-B FULL-attention ops mergeKVRows / filterKVRows match mlx-lm
-//     BatchKVCache (tests/batched-decode-parity), and the SLIDING-window
-//     BatchedRotatingCache (merge/filter/decode/make_mask incl. ring-wrap)
-//     matches mlx-lm BatchRotatingKVCache (tests/batched-rotating).
-// This module is the ORCHESTRATION on top: admission, the step loop, per-row
-// sampling + token accounting, eviction, and assembling each layer's batched
-// cache by type. Gate: tests/parity/batch-scheduler.test.ts (teacher-forced, KL).
-//
-// Per-layer cache types: a model interleaves full-attention layers (plain
-// KVCache, wrapped per step in a BatchedDecodeMaskCache) and sliding-window
-// layers (a persistent BatchedRotatingCache that is itself the batched cache +
-// mask). Full layers share one leftPad/offset (all rows advance together); the
-// rotating caches self-track per-row leftPad/offset as the ring wraps. The
-// per-row absolute position stays consistent across both (full: offset-leftPad;
-// rot: offsetArr) — see docs/design/batching.md. Hybrid gated-DeltaNet
-// models (Qwen3.5) add "ssm" layers: SSMCache state is plain [B,...] with no
-// temporal axis and no padding (rows solo-prefill unpadded, decode feeds one
-// real token per row), so merge/filter are B-axis concat/take and the cache
-// passes through the step unwrapped (no mask, no per-row RoPE — full layers
-// carry positions). Gate: GenerationGateway.place on cache capability.
-//
-// Engine mechanics (the serial decode loop's hygiene, transplanted —
-// batching-v2-plan step 3):
-//   - PIPELINED decode: each step builds the NEXT step's graph from the
-//     still-unread sampled-token array (asyncEval), then reads the PREVIOUS
-//     step's tokens while the new step computes — mlx-lm GenerationBatch._step.
-//     The pipeline is flushed (read + emit) before a join merges, so admission
-//     never has to reconcile an in-flight token array with a new row.
-//   - clearCache every 256 steps (serial's cadence; mlx-lm batched uses 512),
-//     not every step — per-step clears trashed the buffer pool each token.
-//   - CHUNKED, INTERLEAVED admission: a joiner prefills prefillChunkSize
-//     tokens per loop iteration with one batch decode step run in between, so
-//     running rows stall at most one chunk per joiner (mlx-lm interleaves the
-//     same way), and the prefill transient stays bounded.
-//   - Failure containment: one row's onToken throwing evicts THAT row (its
-//     promise rejects); siblings keep decoding (mlx-lm `remove` semantics). A
-//     forward/sampling error still drops the whole batch (can't be attributed
-//     to a row).
-//
-// Bun-async, NO threads: a single detached driver loop owns the GPU for batched
-// mode (an ExclusiveLock keeps the serial fallback off the GPU concurrently).
-// When `admissionHeld` reports a waiting serial-lane request, the loop stops
-// admitting, finishes the running rows, and releases the lock so the serial
-// request runs (mlx-lm's drain_batch) — resumed via kick().
-// Joins re-merge the whole batch; the keep-the-running-batch `extend`
-// optimization is a later refinement (batching-v2-plan item a).
-
-import { MlxArray } from "../../mlx/array";
-import * as ops from "../../mlx/ops";
-import { activeMemory, cacheMemory, clearCache, Dtype, peakMemory } from "../../mlx/ffi";
-import { runtimeConfig, withRuntimeConfig, type RuntimeConfig } from "../../runtime-config";
-import { CompiledDecode } from "../../model/compiled-decode";
-import { legacyCompiledDecodeAvailable } from "./autoregressive";
-import type { Gemma4Model } from "../../model/gemma4";
-import {
-  KVCache, QuantizedKVCache, RotatingKVCache, RotatingQuantizedKVCache,
-  cacheSignature,
-  isBatchableCache,
-  isPlainKvCache,
-  isQuantizedKvCache,
-  isRotatingPlainCache,
-  isRotatingQuantizedCache,
-  isRowBatchCache,
-  type BatchableCache,
-  type Cache,
-} from "../../model/gemma4-base";
-import { BatchedDecodeMaskCache, mergeKVRows, extendKVRows, extractKVRow, filterKVRows } from "../../model/batched-mask";
-import {
-  BatchedQuantDecodeMaskCache, extendQuantRows, extractQuantRow, filterQuantRows, mergeQuantRows,
-  type QuantRow,
-} from "../../model/batched-quant";
-import { BatchedRotatingQuantCache } from "../../model/batched-rotating-quant";
-import { cloneSingleRowState, leaseCacheStates } from "./state-views";
-import { ownedCacheLayout } from "./cache-layout";
-import { createKvMaintenance, type KvMaintenance } from "./kv-maintenance";
-import type { KvScheme } from "../../kv-scheme";
-import { BatchedRotatingCache } from "../../model/batched-rotating";
-import { SSMCache } from "../../model/qwen3-delta";
-import type { RuntimeModel } from "../../model/factory";
-import type { GrammarController } from "../../grammar";
-import { independentGreedySampling } from "../../sampler";
-import { acquireModelWiredLimit } from "../../generate";
-import { batchRowKvBytes } from "../../serve/kv-budget";
-import type { PromptResponseTrace } from "../../serve/prompt-response-trace";
-
-/** Decode-pipeline kill switch (read once at load, like the serial loop's
- *  MLX_BUN_COMPILED_DECODE): 1 ⇒ read each step's tokens synchronously. */
-
-/** Per-step phase timing (MLX_BUN_BATCH_STEP_TRACE=1, debug-only): where a
- *  decode step's wall time goes — graph BUILD+dispatch (host), the pipelined
- *  READ of the previous step's tokens (GPU wait), row EMIT (onToken/SSE), and
- *  the GAP between consecutive steps (drive-loop + everything else). Sums
- *  print via `stepTraceReport()` (the b1 profile experiment calls it). */
-const STEP_T = { t0: 0, lastEnd: 0, build: 0, read: 0, emit: 0, gap: 0, n: 0 };
-export function stepTraceReport(): string {
-  const per = (x: number) => (STEP_T.n ? (x / STEP_T.n).toFixed(3) : "0");
-  const s = `steps=${STEP_T.n} build=${per(STEP_T.build)}ms read=${per(STEP_T.read)}ms emit=${per(STEP_T.emit)}ms gap=${per(STEP_T.gap)}ms (per step)`;
-  STEP_T.t0 = STEP_T.lastEnd = STEP_T.build = STEP_T.read = STEP_T.emit = STEP_T.gap = STEP_T.n = 0;
-  return s;
-}
-
-// NOTE — oMLX-style "adaptive burst decode" (several steps per event-loop
-// yield, omlx/engine_core.py _step_burst) was ported here 2026-07-02 and
-// REFUTED by measurement: on this runtime it REGRESSED cpm5 B=4 aggregate
-// 345→289 tok/s, batch-lane B=1 149→121, and TTFT ~+100 ms (first-token SSE
-// flush waits out the burst budget). Their win exists because each Python
-// step hand-off ping-pongs the GIL with asyncio/uvicorn (~1 ms/token); Bun's
-// setImmediate hop costs microseconds, so bursting here only delays socket
-// flushes. Don't re-add without new evidence — the per-yield step below is
-// the measured optimum (docs/design/batching.md P4 notes).
-
-/** A token sampler for one row: (logits [1,V], step) → token array [1] on
- *  device. Greedy is `(l) => ops.argmaxAxis(l, -1)`; richer closures fold in
- *  temperature / top-p / logits processors + history (built per request from
- *  its sampler options, exactly like generate()'s sampleStep). Called once per
- *  step 0..maxTokens-1 for this row (never beyond maxTokens). */
-export type RowSampler = (logits1V: MlxArray, step: number) => MlxArray;
-
-/** A method prepares and advances rows; the shared executor owns queueing. */
-export interface MlxGroupPreparation {
-  readonly rows: readonly Row[];
-  readonly tokenWeight?: number;
-  readonly canAdmit?: boolean;
-  admit?(row: Row): void;
-  advance(): Promise<boolean>;
-  dispose(): void;
-}
-export interface MlxGroupMethodHost {
-  readonly rows: readonly Row[];
-  readonly runtime: RuntimeConfig;
-  readonly prefillChunkSize: number;
-  readonly promptCache: RowPromptCache | undefined;
-  join(row: Row): void;
-  filterRows(keep: number[]): void;
-  publish(row: Row, token: number, logprobs?: import("../../generate").TokenLogprobs): Promise<void | boolean>;
-  finish(row: Row, reason: "stop" | "length"): void;
-}
-export interface MlxGroupedMethod {
-  prepare(row: Row): MlxGroupPreparation;
-  advance(): Promise<void>;
-  filterRows(keep: readonly number[], discard: boolean): void;
-  dispose(): void;
-}
-export interface MlxGroupMethodRequest {
-  /** Equal keys declare compatible method state, independent of queue size. */
-  readonly key: string;
-  readonly data: unknown;
-  open(host: MlxGroupMethodHost): MlxGroupedMethod;
-}
-
-export type BatchRequest = BatchRequestFields & (
-  | { method: MlxGroupMethodRequest; sample?: RowSampler }
-  | { method?: undefined; sample: RowSampler }
-);
-
-interface BatchRequestFields {
-  /** Optional ordinary continuation policy; owns persistence and sampler recovery. */
-  continuation?: OrdinaryContinuation;
-  statePolicy?: MlxRequestStatePolicy;
-  context?: ExecutionContext;
-  /** Resolve loaded-state identity under the execution lease, after any
-   * queued adapter replacement has finished. */
-  cacheNamespace?: string | (() => string);
-  /** Resolved replay permission; standalone group callers retain auto mode. */
-  compiledDecode?: boolean;
-  promptIds: number[];
-  /** Request policy overrides the execution group's captured prefill default. */
-  prefillChunkSize?: number;
-  maxTokens: number;
-  eosTokenIds: number[];
-  /** Called per emitted (non-EOS) token, in order. Returning `false` halts this
-   *  row (a decoded-text stop sequence fired) — matches generate()'s onToken
-   *  contract. EOS terminates the row WITHOUT an onToken call. Throwing evicts
-   *  THIS row only (its submit promise rejects; siblings continue). May be
-   *  async; keep it cheap — it runs inline in the step loop. */
-  onToken: (token: number, logprobs?: import("../../generate").TokenLogprobs) => void | boolean | Promise<void | boolean>;
-  /** Diagnostic-only request-local trace. */
-  trace?: PromptResponseTrace;
-  /** Closes the gateway's queue/admission span when this row leaves pending. */
-  onAdmitted?: () => void;
-  /** Client/request lifetime. An aborted pending row is removed before
-   *  admission; a row in prefill or decode is evicted at the next safe
-   *  scheduler boundary. */
-  signal?: AbortSignal;
-  /** The row's sampler is PLAIN GREEDY (temperature 0, no curve) with no
-   *  logits processors and no grammar — set by the gateway when true. Lets
-   *  the scheduler take the vectorized sampling fast path (ONE
-   *  log-softmax+argmax over [B,V] instead of B slice/sample/concat graphs)
-   *  when EVERY live row qualifies. Numerically identical per row
-   *  (row-independent ops, same per-row shapes); the per-row closure path
-   *  is the fallback and the MLX_BUN_BATCH_VEC_SAMPLE=0 kill switch. */
-  plainGreedy?: boolean;
-  /** Grammar controller for this row (B1: per-row matchers). When set, the
-   *  scheduler drives it: `accept()` after each emitted token (fires the async
-   *  bitmask fill), `await ready()` before the row's next sample (the sample
-   *  closure applies `applyMask`). Termination (grammar satisfied) is an
-   *  additional per-row stop source. The gateway OWNS disposal (finally around
-   *  submit()); the scheduler uses but never owns. Null on the degrade path. */
-  grammar?: GrammarController;
-  /** Stable cache boundary from the server's template probe (the serial
-   *  lane's options.snapshotAt): the prompt prefix the NEXT turn's render
-   *  preserves. The scheduler snapshots a trim-free prompt-cache entry at
-   *  min(this, promptIds.length-1) during the solo prefill — the oracle
-   *  invariant (mlx-lm insert_segments) that gives wrapped-ring models a
-   *  reuse path. Absent = still capped at promptIds.length-1. */
-  snapshotAt?: number;
-}
-
-export interface BatchStats {
-  spec?: import("../../generate").GenerateStats["spec"];
-  promptTokens: number;
-  generatedTokens: number;
-  /** Prompt tokens served from the prompt cache (Phase 3.2): a joiner's solo
-   *  prefill starts from the longest usable cached prefix, exactly like the
-   *  serial lane's runGeneration take(). 0 on a cold prefill. */
-  cachedTokens: number;
-  finishReason: "stop" | "length";
-  /** Wall-clock from admission (leaving the pending queue) to the first
-   *  emitted token — the row's prefill span on the batch lane. */
-  prefillMs: number;
-  /** Wall-clock from the first emitted token to finish. */
-  decodeMs: number;
-}
-
-/** The slice of PromptCache the scheduler drives (structural — the server's
- *  PromptCache satisfies it). take() on admission (any joiner: a restored
- *  prefix + suffix prefill is byte-safe whether the row later merges or
- *  not). put() paths: rows that finish NEVER-MERGED (adopted lone rows,
- *  Phase 3.2) hand their pristine serial caches over zero-copy; rows that
- *  finish INSIDE a multi-row batch get their KV EXTRACTED per row into
- *  fresh serial caches first (mlx-lm server.py:872 extract_cache — the
- *  cross-request reuse concurrent agents live on). */
-export type RowPromptCache = import("./checkpoint-state").MlxPrefixCache;
-
-export interface Row {
-  spec?: import("../../generate").GenerateStats["spec"];
-  req: BatchRequest;
-  resolve: (s: BatchStats) => void;
-  reject: (e: unknown) => void;
-  current: number; // last emitted token, fed at the next step (pipeline cold)
-  generated: number; // tokens emitted so far (incl. a terminating EOS)
-  sampled: number; // sample() calls so far (leads `generated` by 1 in-pipeline)
-  promptTokens: number;
-  /** Prompt tokens restored from the prompt cache at admission (stats). */
-  cachedTokens: number;
-  /** performance.now() marks for BatchStats timing (0 = not reached). */
-  admittedAt: number;
-  firstTokenAt: number;
-  closeTokenZero?: () => void;
-  cacheNamespace?: string;
-  /** Generated tokens whose KV actually entered the cache (fed as a step
-   *  input) — serial generate()'s `forwarded` list. promptIds+fed is the
-   *  exact token coverage of the row's caches, the put() entry key.
-   *  PER-ROW EXACT for every row: pushes are gated on the pending
-   *  register's per-slot real/placeholder flags (#pendingReal), so a
-   *  doomed slot's placeholder is never recorded — and if one ever DID
-   *  feed a forward, `fedTainted` flips instead (see #step's read). */
-  fed: number[];
-  /** A placeholder slot value fed this row's KV (junk token in the cache):
-   *  the coverage key is no longer derivable — extraction refuses.
-   *  Structurally unreachable today (a doomed row always evicts at the
-   *  emit that follows its placeholder, filtering the slot before the
-   *  next forward; cold steps never contain doomed rows because a flush
-   *  emits every in-flight token) — this flag makes `fed` provably exact
-   *  instead of resting on that timing chain. */
-  fedTainted: boolean;
-  /** Ever merged into a multi-row batch (or poisoned by a reject): its KV
-   *  is interleaved in batched buffers — a finish put() must EXTRACT the
-   *  row (#extractRowCaches) rather than adopt the inners. */
-  merged: boolean;
-}
-
-/** Held while the batch is active; the serial fallback acquires the same lock. */
-export interface ExclusiveLock {
-  acquire(): Promise<() => void>;
-}
-
-export interface MlxBatchExecutionGroupOptions {
-  /** Scheduling work budget for extending an existing preparation cohort. */
-  prefillBatchTokenLimit?: number;
-  runtime?: RuntimeConfig;
-  maxQueued?: number;
-  stateCodecs?: import("../../kv-store").CacheCodecProvider;
-  /** Max rows in the running batch (mlx-lm `--decode-concurrency`). */
-  maxBatch: number;
-  lock?: ExclusiveLock;
-  /** Drain signal: while true, no NEW rows are admitted (running rows finish,
-   *  the lock is released) so a waiting serial-lane request can run — mlx-lm's
-   *  drain_batch. Pair with kick() when it flips back to false. */
-  admissionHeld?: () => boolean;
-  /** Joiner prefill chunk length (default 2048, the serial loop's constant). */
-  prefillChunkSize?: number;
-  /** Aggregate KV-byte budget across all running rows (batching-perf-path
-   *  P3 admission, via `--kv-budget`). A joiner whose PROJECTED KV
-   *  (kvBytesAt(config, prompt + maxTokens), sliding window already capped)
-   *  would push the batch's projected total over this ceiling WAITS in the
-   *  queue (FIFO, head-of-line — no starvation, no reorder) until rows
-   *  evict. A request that can NEVER fit (over budget alone, empty batch)
-   *  is rejected instead of deadlocking. Unset = unlimited (v1 behavior). */
-  kvBudgetBytes?: number;
-  /** Authoritative server-wide scheme. Each joiner's solo prefill converts at
-   * the same chunk boundaries as the serial path, so a row's quantized bytes
-   * preserve the L2 composition. Unsupported schemes fail at construction. */
-  kvScheme?: KvScheme;
-  /** Model binding owns support for precision transitions. */
-  kvBatchCapabilities?: { delayedAffine?: boolean };
-  /** Prompt-cache hook (Phase 3.2): admission take()s the longest usable
-   *  prefix into the joiner's solo caches (suffix-only prefill — the
-   *  multi-turn chat TTFT path); rows that finish never-merged put() their
-   *  caches back. Adapter requests never reach the batch lane, so the
-   *  namespace is always "" here. Runs under the gateway mutex domain, so
-   *  take/put never race the serial lane's use of the same cache. */
-  promptCache?: RowPromptCache;
-}
-
-type LayerInner =
-  | KVCache
-  | QuantizedKVCache
-  | BatchedRotatingCache
-  | BatchedRotatingQuantCache // rot layer under a kv_config scheme (milestone 2)
-  | RotatingKVCache // adopted lone-row state only (see #mergeJoiner adopt)
-  | RotatingQuantizedKVCache // adopted lone-row state, quantized rot layer
-  | SSMCache
-  | BatchableCache;
-type Row1 = { keys: MlxArray; values: MlxArray };
-
-export class MlxBatchExecutionGroup {
-  readonly #runtime: RuntimeConfig;
-  readonly #noPipeline: boolean;
-  readonly #stepTrace: boolean;
-  readonly #stateCodecs: import("../../kv-store").CacheCodecProvider | undefined;
-  readonly #maxQueued: number;
-  #running: Row[] = [];
-  #inners: LayerInner[] | null = null; // per-layer batched KV; null when empty
-  #fullLeftPad: number[] = []; // per-row padding for FULL layers (rot self-tracks)
-  #pending: Row[] = [];
-  #prefill: MlxGroupPreparation | null = null; // the (single) joiner mid-prefill
-  #preparationPublishedOutput = false;
-  /** Sampled-but-unread token array [B], aligned with #running — the decode
-   *  pipeline register. Filtered/disposed alongside the batched KV. */
-  #pendingToks: MlxArray | null = null;
-  /** Per-slot flags for #pendingToks: true = a REAL sampled token of that
-   *  row's stream; false = a doomed/terminated row's placeholder (fed-token
-   *  accounting must skip it — see Row.fed/fedTainted). Filtered/cleared in
-   *  lockstep with the register. */
-  #pendingReal: boolean[] | null = null;
-  #method: MlxGroupedMethod | undefined;
-  #methodKey: string | undefined;
-  #steps = 0; // decode-step counter (clearCache cadence)
-  #looping = false;
-  #closed = false;
-  #driver: Promise<void> | null = null;
-  #wake: (() => void) | null = null;
-  readonly #maxBatch: number;
-  readonly #lock: ExclusiveLock | undefined;
-  readonly #admissionHeld: (() => boolean) | undefined;
-  readonly #prefillChunkSize: number;
-  readonly #prefillBatchTokenLimit: number;
-  readonly #prefillTailSplit: boolean;
-  readonly #kvBudgetBytes: number | undefined;
-  readonly #defaultPromptCache: RowPromptCache | undefined;
-  #statePolicy?: MlxRequestStatePolicy;
-  get #promptCache(): RowPromptCache | undefined {
-    return this.#statePolicy ? this.#statePolicy.promptCache : this.#defaultPromptCache;
-  }
-  /** Retain hook of the currently-ADOPTED row's cache entry (at most one:
-   *  only a lone adopted row holds un-copied entry caches). Runs after the
-   *  adopted caches are disposed, or transfers back on put(). */
-  #adoptedRetain: (() => void) | null = null;
-  #contextKey: string | undefined;
-  #releaseContext: (() => void) | undefined;
-  readonly #kinds: ("full" | "rot" | "ssm" | "owned-batch")[];
-  readonly #rotMaxSize: number[]; // per-layer sliding window (rot layers only)
-  readonly #compressedProjectors: Array<(tokens: number) => number> | null;
-  readonly #batchCacheMaxTokens: number | null;
-  readonly #kvScheme: KvScheme | undefined;
-  readonly #maintainKv: KvMaintenance | null;
-  /** Compiled decode runner for the B=1 serial-class case (Phase 3.2) —
-   *  same eligibility gate as generate.ts (gemma dense, kill switch
-   *  MLX_BUN_COMPILED_DECODE); adapter requests disable replay in their plan. Set
-   *  to null permanently on a failed step (serial disables per
-   *  generation; the scheduler is one long-lived "generation"). */
-  #compiled: CompiledDecode | null;
-
-  constructor(private readonly model: RuntimeModel, opts: MlxBatchExecutionGroupOptions) {
-    this.#runtime = opts.runtime ?? runtimeConfig();
-    this.#noPipeline = this.#runtime.value("MLX_BUN_BATCH_NO_PIPELINE") === "1";
-    this.#stepTrace = this.#runtime.value("MLX_BUN_BATCH_STEP_TRACE") === "1";
-    this.#maxQueued = opts.maxQueued ?? 64;
-    if (!Number.isSafeInteger(this.#maxQueued) || this.#maxQueued < 1) throw new Error("invalid batch queue limit");
-    this.#stateCodecs = opts.stateCodecs;
-    this.#maxBatch = Math.max(1, Math.floor(opts.maxBatch));
-    this.#lock = opts.lock;
-    this.#admissionHeld = opts.admissionHeld;
-    this.#prefillChunkSize = Math.max(1, Math.floor(opts.prefillChunkSize ??
-      this.#runtime.number("MLX_BUN_RD_PREFILL_CHUNK", 2048)));
-    this.#prefillBatchTokenLimit = opts.prefillBatchTokenLimit ?? 2048;
-    this.#prefillTailSplit = this.#runtime.flag("MLX_BUN_PREFILL_TAIL_SPLIT", true);
-    this.#kvBudgetBytes = opts.kvBudgetBytes;
-    this.#defaultPromptCache = opts.promptCache;
-    this.#kvScheme = opts.kvScheme;
-    const proto = withRuntimeConfig(this.#runtime, () => model.makeCache()); // fresh caches hold no buffers
-    if (this.#kvScheme && !this.#kvScheme.batchable(
-      model.config,
-      (layerIdx) =>
-        isPlainKvCache(proto[layerIdx]) || isRotatingPlainCache(proto[layerIdx]),
-      proto.length, opts.kvBatchCapabilities,
-    )) {
-      for (const cache of proto) cache.dispose();
-      throw new Error(`unsupported KV scheme for batch scheduler: ${this.#kvScheme.kind}`);
-    }
-    this.#kinds = proto.map((c) =>
-      isBatchableCache(c)
-        ? "owned-batch"
-        : isRotatingPlainCache(c)
-          ? "rot"
-          : cacheSignature(c) === "ssm"
-            ? "ssm"
-            : "full",
-    );
-    this.#compressedProjectors = proto.every(isBatchableCache)
-      ? proto.map((cache) => (tokens: number) => cache.projectedBytes(tokens))
-      : null;
-    this.#batchCacheMaxTokens = proto.every(isBatchableCache)
-      ? Math.min(...proto.map((cache) => cache.maxTokens ?? Infinity))
-      : null;
-    this.#maintainKv = this.#kvScheme?.quantized
-      ? createKvMaintenance(this.#kvScheme.options)
-      : null;
-    this.#rotMaxSize = proto.map((c) => (isRotatingPlainCache(c) ? c.maxSize : 0));
-    for (const c of proto) c.dispose();
-    this.#compiled =
-      this.#runtime.flag("MLX_BUN_COMPILED_DECODE", true) &&
-      legacyCompiledDecodeAvailable(model)
-        ? withRuntimeConfig(this.#runtime, () => CompiledDecode.for(model as Gemma4Model))
-        : null;
-  }
-
-  get activeRows(): number {
-    return this.#running.length;
-  }
-
-  get pendingRows(): number {
-    return this.#pending.length + (this.#prefill?.rows.length ?? 0);
-  }
-
-  /** Projected KV bytes of one row at its worst case (full prompt + full
-   *  completion; the sliding-window term is window-capped by kvBytesAt). */
-  #rowKvBytes(row: Row): number {
-    return this.#compressedProjectors
-      ? this.#compressedProjectors.reduce(
-          (sum, project) => sum + project(row.promptTokens + row.req.maxTokens),
-          0,
-        )
-      : batchRowKvBytes(
-          this.model.config,
-          row.promptTokens,
-          row.req.maxTokens,
-          this.#kvScheme,
-        );
-  }
-
-  async #forwardHidden(ids: MlxArray, cache: Cache[]): Promise<MlxArray> {
-    const asyncModel = this.model as RuntimeModel & {
-      forwardHiddenAsync?: (ids: MlxArray, cache: Cache[]) => Promise<MlxArray>;
-    };
-    return typeof asyncModel.forwardHiddenAsync === "function"
-      ? await asyncModel.forwardHiddenAsync(ids, cache)
-      : this.model.forwardHidden(ids, cache);
-  }
-
-  /** Projected aggregate KV of everything admitted (running + mid-prefill). */
-  get projectedKvBytes(): number {
-    let total = this.#running.reduce((a, r) => a + this.#rowKvBytes(r), 0);
-    if (this.#prefill) for (const row of this.#prefill.rows) total += this.#rowKvBytes(row);
-    return total;
-  }
-
-  get kvBudgetBytes(): number | undefined {
-    return this.#kvBudgetBytes;
-  }
-
-  /** KV-budget admission for the queue head. True = admit now. A candidate
-   *  that cannot fit even alone is rejected here (never deadlocks the
-   *  queue); one that fits alone but not alongside the current batch waits. */
-  #kvAdmits(candidate: Row): boolean {
-    const requestedTokens = candidate.promptTokens + candidate.req.maxTokens;
-    if (
-      this.#batchCacheMaxTokens !== null &&
-      requestedTokens > this.#batchCacheMaxTokens
-    ) {
-      this.#pending.shift();
-      candidate.reject(new RangeError(
-        `context limit: prompt ${candidate.promptTokens} + max_tokens ` +
-        `${candidate.req.maxTokens} exceeds ${this.#batchCacheMaxTokens}`,
-      ));
-      return false;
-    }
-    if (this.#kvBudgetBytes === undefined) return true;
-    const need = this.#rowKvBytes(candidate);
-    if (need > this.#kvBudgetBytes && this.#running.length === 0 && !this.#prefill) {
-      this.#pending.shift();
-      candidate.reject(
-        new Error(
-          `kv budget: request needs ~${(need / 1e9).toFixed(2)} GB KV ` +
-            `(prompt ${candidate.promptTokens} + max_tokens ${candidate.req.maxTokens}), ` +
-            `over --kv-budget ${(this.#kvBudgetBytes / 1e9).toFixed(2)} GB — ` +
-            `lower max_tokens or raise the budget`,
-        ),
-      );
-      return false;
-    }
-    return this.projectedKvBytes + need <= this.#kvBudgetBytes;
-  }
-
-  /** Submit a request; resolves when its row finishes (EOS, stop, or length). */
-  submit(req: BatchRequest): Promise<BatchStats> {
-    if (this.#closed) return Promise.reject(new Error("scheduler closed"));
-    if (req.signal?.aborted) return Promise.reject(req.signal.reason);
-    if (this.#pending.length >= this.#maxQueued) return Promise.reject(new AdmissionRejected());
-    return new Promise<BatchStats>((resolve, reject) => {
-      let abortListener: (() => void) | null = null;
-      const cleanup = () => {
-        if (abortListener) req.signal?.removeEventListener("abort", abortListener);
-        abortListener = null;
-      };
-      this.#pending.push({
-        req,
-        resolve: (stats) => { cleanup(); resolve(stats); },
-        reject: (error) => { cleanup(); reject(error); },
-        current: 0, generated: 0, sampled: 0, promptTokens: req.promptIds.length,
-        admittedAt: 0, firstTokenAt: 0,
-        cachedTokens: 0, fed: [], fedTainted: false, merged: false,
-      });
-      if (req.signal) {
-        abortListener = () => this.kick();
-        req.signal.addEventListener("abort", abortListener, { once: true });
-      }
-      this.#ensureLoop();
-    });
-  }
-
-  /** Wake the driver loop (e.g. after admissionHeld flips back to false). */
-  kick(): void {
-    this.#ensureLoop();
-  }
-
-  #ensureLoop(): void {
-    if (this.#wake) { this.#wake(); return; }
-    if (this.#looping || this.#closed) return;
-    this.#looping = true;
-    this.#driver = withRuntimeConfig(this.#runtime, () => this.#drive());
-    void this.#driver.catch((error) => console.error(`batch scheduler cleanup failed: ${error}`));
-  }
-
-  /** Stop at the next safe work boundary and release all queued/active runs. */
-  async close(): Promise<void> {
-    this.#closed = true;
-    this.#wake?.();
-    await this.#driver;
-  }
-
-  async #drive(): Promise<void> {
-    const scheduler = this;
-    const group: ExecutionGroup = {
-      get active() { return scheduler.#running.length; },
-      get queued() { return scheduler.#pending.length; },
-      get preparing() { return scheduler.#prefill !== null; },
-      get preparingRows() { return scheduler.#prefill?.rows.length ?? 0; },
-      get canPrepareMore() { return scheduler.#prefill?.canAdmit === true; },
-      get preparingTokens() { return scheduler.#prefill?.tokenWeight ?? 0; },
-      // Cache restoration belongs to preparation. A queued request's prompt
-      // length is an upper bound until its cache owner resolves the prefix.
-      get nextPreparationTokens() { return scheduler.#pending[0]?.promptTokens ?? 0; },
-      get maxPreparationTokens() { return scheduler.#prefillBatchTokenLimit; },
-      get preparationPublishedOutput() { return scheduler.#preparationPublishedOutput; },
-      get maxActive() { return scheduler.#maxBatch; },
-      get admissionHeld() { return scheduler.#admissionHeld?.() === true; },
-      get closed() { return scheduler.#closed; },
-      pruneCancelled() {
-        for (let i = scheduler.#pending.length - 1; i >= 0; i--) {
-          const row = scheduler.#pending[i]!;
-          if (!row.req.signal?.aborted) continue;
-          scheduler.#pending.splice(i, 1);
-          row.reject(row.req.signal.reason);
-        }
-      },
-      admitNext: () => this.#admitNext(),
-      canBurst: () => this.#contextCompatible(this.#pending[0]!) &&
-        (this.#kvBudgetBytes === undefined ||
-          this.projectedKvBytes + this.#rowKvBytes(this.#pending[0]!) <= this.#kvBudgetBytes),
-      advancePreparation: () => this.#advancePreparation(),
-      advance: () => this.#method ? this.#method.advance() : this.#step(),
-      failActive: (error) => {
-        for (const row of this.#running) row.reject(error);
-        this.#applyFilter([], true); // failed state never enters the prefix store
-      },
-      failAll: (error) => this.#failAll(error),
-      reserveResidency: () => {
-        const release = acquireModelWiredLimit(this.model);
-        return () => disposeResources([
-          { dispose: () => this.#leaveContext() }, { dispose: release },
-        ]);
-      },
-      ...(this.#lock ? { acquireExecution: () => this.#lock!.acquire() } : {}),
-      waitForWork: async () => {
-        await new Promise<void>((resolve) => { this.#wake = resolve; });
-        this.#wake = null;
-      },
-    };
-    try {
-      await driveExecutionGroup(group, {
-        now: () => performance.now(),
-        yield: () => new Promise<void>((resolve) => setImmediate(resolve)),
-      }, this.#runtime.value("MLX_BUN_EARLY_FIRST_TOKEN") === "1");
-    } finally { this.#looping = false; }
-  }
-
-  #contextCompatible(row: Row): boolean {
-    return (!this.#running.length && !this.#prefill) || ((row.req.context?.key ?? "") === this.#contextKey && row.req.method?.key === this.#methodKey && row.req.statePolicy?.key === this.#statePolicy?.key);
-  }
-
-  #leaveContext(): void {
-    const release = this.#releaseContext;
-    this.#releaseContext = undefined;
-    this.#contextKey = undefined;
-    release?.();
-  }
-
-  #admitNext(): boolean {
-    if (!this.#contextCompatible(this.#pending[0]!) || !this.#kvAdmits(this.#pending[0]!)) return false;
-    const row = this.#pending.shift()!;
-    try {
-      this.#statePolicy = row.req.statePolicy;
-      row.cacheNamespace = typeof row.req.cacheNamespace === "function"
-        ? row.req.cacheNamespace() : row.req.cacheNamespace;
-      const key = row.req.context?.key ?? "";
-      if (key !== this.#contextKey) {
-        this.#leaveContext();
-        this.#releaseContext = row.req.context?.enter();
-        this.#contextKey = key;
-      }
-      row.req.onAdmitted?.();
-      row.admittedAt = performance.now();
-      if (row.req.method?.key !== this.#methodKey) {
-        this.#method?.dispose(); this.#method = undefined; this.#methodKey = undefined;
-        if (row.req.method) {
-          const executor = this;
-          this.#method = row.req.method.open({
-            get rows() { return executor.#running; }, runtime: this.#runtime,
-            prefillChunkSize: this.#prefillChunkSize, promptCache: this.#promptCache,
-            join: row => { this.#running.push(row); },
-            filterRows: keep => this.#applyFilter(keep),
-            publish: async (row, token, logprobs) => {
-              row.req.signal?.throwIfAborted();
-              this.#firstOutput(row);
-              return row.req.onToken(token, logprobs);
-            },
-            finish: (row, reason) => this.#finish(row, reason),
-          });
-          this.#methodKey = row.req.method.key;
-        }
-      }
-      if (this.#prefill) {
-        this.#prefill.admit!(row);
-      } else if (this.#method) {
-        this.#prefill = this.#method.prepare(row);
-      } else {
-        const preparation = new MlxPrefillCohort({
-          model: this.model, chunkSize: this.#prefillChunkSize, tailSplit: this.#prefillTailSplit,
-          promptCache: this.#promptCache, stateCodecs: this.#stateCodecs, maintain: this.#maintainKv ?? undefined,
-          forward: (ids, caches) => this.#forwardHidden(ids, caches),
-          project: (hidden, caches, completed) => this.#projectPrefill(hidden, caches, completed),
-          complete: (state, logits) => this.#completePrefill(state, logits),
-          resume: state => this.#completePrefill(state, null),
-          reject: (failed, error) => this.#rejectPreparingRows([failed], error),
-        });
-        preparation.admit(row); this.#prefill = preparation;
-      }
-    } catch (error) { row.reject(error); }
-    return true;
-  }
-
-  #rejectPreparingRows(rows: readonly Row[], error: unknown): void {
-    const retiring = new Set(rows);
-    const keep = this.#running.flatMap((row, index) => retiring.has(row) ? [] : [index]);
-    if (keep.length !== this.#running.length) this.#applyFilter(keep, true);
-    for (const row of rows) row.reject(error);
-  }
-
-  async #advancePreparation(): Promise<void> {
-    const p = this.#prefill!, rows = p.rows;
-    this.#preparationPublishedOutput = false;
-    try {
-      if (await p.advance()) this.#prefill = null;
-    } catch (error) {
-      this.#prefill = null;
-      let failure = error;
-      try { p.dispose(); }
-      catch (cleanupError) { failure = new AggregateError([error, cleanupError], "batch prefill and cleanup failed"); }
-      try { this.#rejectPreparingRows(rows, failure); }
-      catch (cleanupError) {
-        failure = new AggregateError([failure, cleanupError], "batch admission retirement failed");
-        for (const row of rows) row.reject(failure);
-        this.#failAll(failure);
-      }
-    }
-  }
-
-  #failAll(error: unknown): void {
-    const p = this.#prefill;
-    const rows = new Set([...this.#running, ...this.#pending, ...(p?.rows ?? [])]);
-    const retain = this.#adoptedRetain;
-    const resources = [p, this.#method, ...(this.#inners ?? []), this.#pendingToks,
-      { dispose: () => retain?.() }]
-      .filter((resource): resource is DisposableResource => resource != null);
-    this.#pending = []; this.#running = []; this.#prefill = null;
-    this.#method = undefined; this.#methodKey = undefined;
-    this.#inners = null; this.#pendingToks = null; this.#pendingReal = null;
-    this.#fullLeftPad = []; this.#adoptedRetain = null;
-    for (const row of rows) row.reject(error);
-    disposeResources(resources);
-  }
-
-  /** Per-layer mixed-precision conversion of a joiner's SOLO caches — the
-   *  scheduler-side mirror of the serial maybeQuantizeKv (generate.ts): same
-   *  per-layer map, same skip rules (empty cache, already quantized), same
-   *  streaming discipline (evalAll the converted layer's state so the bf16
-   *  source frees before the next layer converts). Called at every prefill
-   *  chunk boundary AND once before merge, exactly where the serial loop
-   *  calls maybeQuantizeKv — that placement is what makes a row's quantized
-   *  bytes bit-exact vs serial `--kv-quant config`. Gateway placement and the
-   *  constructor both guarantee every named cache can convert. */
-  #quantizeSolo(solo: Cache[], trace?: PromptResponseTrace): void {
-    if (!this.#maintainKv) return;
-    const close = trace?.begin("prefill.kv_maintenance", {
-      mechanism: "continuous",
-    });
-    try {
-      this.#maintainKv(solo);
-    } finally {
-      close?.();
-    }
-  }
-
-  /** The target owns projection geometry; samplers receive individual rows. */
-  #projectPrefill(hidden: MlxArray, caches: Cache[], completed: readonly PrefillState[]): MlxArray {
-    for (const p of completed) {
-      p.closePrefill?.(); p.closePrefill = undefined;
-      p.row.closeTokenZero = p.row.req.trace?.begin("token_zero.total", { mechanism: "continuous" });
-    }
-    const attributed = this.#runtime.value("MLX_BUN_P2R_SYNC") === "1"
-      ? completed.flatMap(p => p.row.req.trace ? [p.row.req.trace] : []) : [];
-    const begin = (phase: "token_zero.forward" | "token_zero.head") => attributed.map(trace => trace.begin(phase, {
-      mechanism: "continuous", activeBytes: activeMemory(), cacheBytes: cacheMemory(), peakBytes: peakMemory(),
-    }));
-    if (attributed.length) {
-      const closes = begin("token_zero.forward");
-      try { withResource(leaseCacheStates(caches), state => ops.evalAll([hidden, ...state])); }
-      finally { for (const close of closes) close(); }
-    }
-    // Project before selecting rows: M4 quantized projection can use different
-    // arithmetic at M=1. Tail-split preparation reaches this with N=1.
-    const logits = this.model.logitsFromHidden(hidden);
-    try {
-      if (attributed.length) {
-        const closes = begin("token_zero.head");
-        try { ops.evalAll([logits]); } finally { for (const close of closes) close(); }
-      }
-      return logits;
-    } catch (error) { logits.dispose(); throw error; }
-  }
-
-  /** Sample borrowed logits from the shared target projection. */
-  async #completePrefill(p: PrefillState, logits: MlxArray | null): Promise<void> {
-    if (p.continuation) {
-      const saved = p.continuation, row = p.row;
-      row.req.continuation!.resumeSampling(saved);
-      const interval = row.req.continuation!.interval;
-      this.#checkpointAt.set(row, (Math.floor(saved.generatedTokens / interval) + 1) * interval);
-      row.firstTokenAt = performance.now();
-      row.closeTokenZero?.(); row.closeTokenZero = undefined;
-      this.#preparationPublishedOutput = true;
-      for (const token of saved.cacheTokens.slice(row.req.promptIds.length)) {
-        row.req.signal?.throwIfAborted();
-        if (await row.req.onToken(token) === false)
-          throw new Error("saved generation prefix triggered a terminal stop while replaying");
-      }
-      row.fed = saved.cacheTokens.slice(row.req.promptIds.length);
-      row.generated = saved.generatedTokens + 1;
-      row.sampled = saved.generatedTokens + 1;
-      row.cachedTokens = Math.min(saved.cacheTokens.length, row.promptTokens);
-      const stop = await this.#emit(row, saved.pendingToken);
-      if (stop !== "continue") {
-        this.#putOrDispose(p.solo, saved.cacheTokens, p.retain, row.cacheNamespace);
-        p.solo = [];
-        this.#finish(row, stop);
-        return;
-      }
-      await this.#mergeJoiner(p);
-      return;
-    }
-    if (!logits) throw new Error("ordinary prefill completed without logits");
-    const forceAttribution = !!p.row.req.trace && this.#runtime.value("MLX_BUN_P2R_SYNC") === "1";
-    const V = logits.shape[2]!;
-    const last2 = ops.reshape(logits, [1, V]);
-    const closeSample = forceAttribution
-      ? p.row.req.trace!.begin("token_zero.sample", { mechanism: "continuous" })
-      : undefined;
-    let sampled: MlxArray;
-    try { sampled = p.row.req.sample!(last2, 0); } finally { last2.dispose(); }
-    // The first admission can seed the existing device-token register. Its
-    // first decode forward then overlaps token-zero readback, just like every
-    // later pipeline step. Grammar and explicit eager-output modes retain
-    // their read-before-advance order.
-    if (!this.#inners && !p.row.req.grammar && !this.#noPipeline &&
-        this.#runtime.value("MLX_BUN_EARLY_FIRST_TOKEN") !== "1" && !forceAttribution) {
-      let owned: MlxArray | null = sampled;
-      try {
-        ops.asyncEvalAll([sampled]);
-        p.row.sampled = 1;
-        this.#quantizeSolo(p.solo, p.row.req.trace);
-        await this.#mergeJoiner(p);
-        this.#pendingToks = sampled;
-        this.#pendingReal = [true];
-        owned = null;
-        return;
-      } finally { owned?.dispose(); }
-    }
-    const tok = this.#readToken(sampled);
-    closeSample?.();
-    p.row.sampled = 1;
-    p.row.generated = 1;
-
-    // Grammar (B1): the mask0 was applied inside the sample closure (the
-    // controller is primed at compile). After reading token 0, advance the
-    // matcher — fires the async fill for the NEXT step, overlapping the merge
-    // below. If the grammar is already satisfied at token 0 (a 1-token grammar,
-    // e.g. guided_choice landing on a single-token option), emit + finish("stop")
-    // WITHOUT merging into the batch — the row never joins #running.
-    if (p.row.req.grammar) {
-      p.row.req.grammar.accept(tok);
-      if (p.row.req.grammar.isTerminated) {
-        const stop = await this.#emit(p.row, tok);
-        // Token 0 was sampled but never fed — the caches cover exactly the
-        // prompt, a clean prompt-only entry (put-or-dispose).
-        this.#putOrDispose(p.solo, p.row.req.promptIds, p.retain, p.row.cacheNamespace);
-        this.#finish(p.row, stop === "continue" ? "stop" : stop);
-        return;
-      }
-    }
-
-    const stop = await this.#emit(p.row, tok);
-    if (stop !== "continue") {
-      this.#putOrDispose(p.solo, p.row.req.promptIds, p.retain, p.row.cacheNamespace);
-      this.#finish(p.row, stop);
-      return;
-    }
-    // Default tail-split path: the caches were already converted at the head
-    // boundary above (oracle composition: prefill ids[:-1] → convert → L=1
-    // step-0), so this call is an idempotent no-op (converted caches match
-    // neither serial class). It is LOAD-BEARING only under
-    // MLX_BUN_PREFILL_TAIL_SPLIT=0 — the old serial order: token 0 sampled
-    // from the unconverted final-chunk logits, THEN the caches convert
-    // (before decode step 1 == before the merge).
-    this.#quantizeSolo(p.solo, p.row.req.trace);
-    await this.#mergeJoiner(p);
-    return;
-  }
-
-  /** Merge a fully-prefilled joiner with the running batch, layer by layer
-   *  (re-merge; `extend` is the later refinement). Flushes the decode pipeline
-   *  first so the row set is settled and the next step starts cold. */
-  async #mergeJoiner(p: PrefillState): Promise<void> {
-    await this.#flushPipeline();
-    this.#maintainKv?.prepareBatch?.(p.solo);
-
-    // ADOPT, don't copy (unified-engine plan Phase 3.2): a row joining an
-    // EMPTY batch keeps its solo caches as the batch inners — a pointer
-    // handoff, zero bytes moved (the old path ran the full merge machinery
-    // to produce a byte-identical [1,...] copy). The copy now happens only
-    // when a SECOND row joins and a genuinely new layout must exist. The
-    // prize beyond the saved copy: the lone row's caches stay SERIAL-CLASS
-    // (KVCache / RotatingKVCache / QuantizedKVCache), so the B=1 step is
-    // literally the serial graph, and compiled decode + prompt-cache
-    // take/put become possible for it. The rot branch below knows how to
-    // treat an adopted RotatingKVCache as the merge's first row.
-    if (!this.#inners) {
-      this.#inners = p.solo as LayerInner[]; p.solo = [];
-      this.#fullLeftPad = [0];
-      this.#adoptedRetain = p.retain ?? null; p.retain = undefined;
-      this.#running.push(p.row);
-      return;
-    }
-
-    const prev = this.#inners;
-    const prevPad = this.#fullLeftPad;
-    const B = this.#running.length;
-    const newInners: LayerInner[] = [];
-    let newFullPad = this.#fullLeftPad;
-    for (let layer = 0; layer < this.#kinds.length; layer++) {
-      const solo = p.solo[layer]!;
-      const merged = ownedCacheLayout(solo);
-      if (merged) {
-        const previous = prev?.[layer];
-        merged.mergeRows(previous ? [previous, solo] : [solo]);
-        newFullPad = [...merged.leftPad];
-        newInners.push(merged);
-        continue;
-      }
-      if (this.#kinds[layer] === "owned-batch")
-        throw new Error(`batch-capable layer ${layer} lost its cache capability`);
-      if (this.#kinds[layer] === "ssm") {
-        // No temporal axis, no left-pad: B-axis concat of the state slots.
-        // mergeRows steals the solo arrays when the batch starts cold, so the
-        // unconditional p.solo dispose below stays safe either way.
-        newInners.push(SSMCache.mergeRows(
-          (prev?.[layer] as SSMCache | undefined) ?? null,
-          p.solo[layer] as SSMCache,
-        ));
-        continue;
-      }
-      if (isQuantizedKvCache(p.solo[layer])) {
-        // Phase 3.1 — quantized full layer: same merge/extend shapes as the
-        // bf16 branch below, over (packed, scales, biases) triples. The solo
-        // row was converted by #quantizeSolo with the serial ops, so its
-        // bytes already bit-match serial `--kv-quant config`; this branch
-        // only re-arranges rows along the batch axis.
-        const qSolo = p.solo[layer] as QuantizedKVCache;
-        const [qk, qv] = qSolo.temporalView();
-        const qRow = { keys: qk, values: qv };
-        const dispose3 = (t: { packed: MlxArray; scales: MlxArray; biases: MlxArray }) => {
-          t.packed.dispose(); t.scales.dispose(); t.biases.dispose();
-        };
-        const prevQ = prev?.[layer] as QuantizedKVCache | undefined;
-        if (prevQ && this.#runtime.value("MLX_BUN_BATCH_EXTEND") !== "0") {
-          const [k0, v0] = prevQ.temporalView();
-          const ext = extendQuantRows(k0, v0, prevPad, qRow);
-          dispose3(k0); dispose3(v0);
-          newFullPad = ext.leftPad;
-          const c = new QuantizedKVCache(qSolo.groupSize, qSolo.bits);
-          c.restoreState(ext.keys, ext.values, ext.width);
-          newInners.push(c);
-        } else {
-          const rows: { keys: typeof qk; values: typeof qv }[] = [];
-          if (prevQ) {
-            const [k0, v0] = prevQ.temporalView(); // [B,H,off,*]
-            const S = k0.packed.shape[2]!;
-            for (let b = 0; b < B; b++) {
-              const pad = prevPad[b]!;
-              const cutRow = (t: typeof k0): typeof k0 => ({
-                packed: t.packed.slice([b, 0, pad, 0], [b + 1, t.packed.shape[1]!, S, t.packed.shape[3]!]),
-                scales: t.scales.slice([b, 0, pad, 0], [b + 1, t.scales.shape[1]!, S, t.scales.shape[3]!]),
-                biases: t.biases.slice([b, 0, pad, 0], [b + 1, t.biases.shape[1]!, S, t.biases.shape[3]!]),
-              });
-              rows.push({ keys: cutRow(k0), values: cutRow(v0) });
-            }
-            dispose3(k0); dispose3(v0);
-          }
-          rows.push(qRow);
-          const merged = mergeQuantRows(rows);
-          newFullPad = merged.leftPad;
-          const c = new QuantizedKVCache(qSolo.groupSize, qSolo.bits);
-          c.restoreState(merged.keys, merged.values, merged.width);
-          newInners.push(c);
-          for (const r of rows) { dispose3(r.keys); dispose3(r.values); }
-        }
-        // qRow views are disposed via the rows loop above or here for extend
-        if (prevQ && this.#runtime.value("MLX_BUN_BATCH_EXTEND") !== "0") { dispose3(qRow.keys); dispose3(qRow.values); }
-        else if (!prevQ) { /* disposed in the rows loop */ }
-        continue;
-      }
-      // NOTE: the quantized-rotating branch below never uses this layer's
-      // bf16 temporalView — calling it before the branch leaked the pair
-      // (six arrays per second-row join on quantized rotating layers,
-      // introduced in 859572d; 2026-07-07 review fix). The view is taken
-      // AFTER the branch, on the paths that actually consume it.
-      const soloC = p.solo[layer] as KVCache | RotatingKVCache;
-      if (this.#kinds[layer] === "rot" && isRotatingQuantizedCache(p.solo[layer])) {
-        // Milestone 2 — QUANTIZED rotating layer: the solo row converted at
-        // the serial boundaries (#quantizeSolo), so its ring bytes are the
-        // serial oracle's; this branch re-arranges temporal triples across
-        // the batch axis (the bf16 rot merge over triples).
-        const qSolo = p.solo[layer] as RotatingQuantizedKVCache;
-        const [qk, qv] = qSolo.temporalView();
-        const rows: QuantRow[] = [];
-        const offsets: number[] = [];
-        const dispose3 = (t: QuantRow) => {
-          t.keys.packed.dispose(); t.keys.scales.dispose(); t.keys.biases.dispose();
-          t.values.packed.dispose(); t.values.scales.dispose(); t.values.biases.dispose();
-        };
-        const prevC = prev?.[layer];
-        if (prevC && isRowBatchCache(prevC) && isRotatingQuantizedCache(prevC)) {
-          const batched = prevC as BatchedRotatingQuantCache;
-          const [k0, v0] = batched.temporalView(); // [B,H,valid,*] triples, temporal
-          const valid = k0.packed.shape[2]!;
-          for (let b = 0; b < B; b++) {
-            const pad = Math.max(0, batched.leftPad[b]!);
-            const cutRow = (t: typeof k0): typeof k0 => ({
-              packed: t.packed.slice([b, 0, pad, 0], [b + 1, t.packed.shape[1]!, valid, t.packed.shape[3]!]),
-              scales: t.scales.slice([b, 0, pad, 0], [b + 1, t.scales.shape[1]!, valid, t.scales.shape[3]!]),
-              biases: t.biases.slice([b, 0, pad, 0], [b + 1, t.biases.shape[1]!, valid, t.biases.shape[3]!]),
-            });
-            rows.push({ keys: cutRow(k0), values: cutRow(v0) });
-            offsets.push(batched.offsetArr[b]!);
-          }
-          for (const t of [k0, v0]) { t.packed.dispose(); t.scales.dispose(); t.biases.dispose(); }
-        } else if (isRotatingQuantizedCache(prevC)) {
-          // Adopted lone row (Phase 3.2 adopt): its chronological triples
-          // are the merge's first row, pad 0 by definition.
-          const [k0, v0] = prevC.temporalView();
-          rows.push({ keys: k0, values: v0 });
-          offsets.push(prevC.offset);
-        }
-        rows.push({ keys: qk, values: qv });
-        offsets.push(qSolo.offset);
-        newInners.push(BatchedRotatingQuantCache.merge(
-          rows, offsets, this.#rotMaxSize[layer]!, qSolo.groupSize, qSolo.bits,
-        ));
-        for (const r of rows) dispose3(r);
-        continue;
-      }
-      const [sk, sv] = soloC.temporalView();
-      const newRow: Row1 = { keys: sk, values: sv };
-      if (this.#kinds[layer] === "rot") {
-        const rows: Row1[] = [];
-        const offsets: number[] = [];
-        const prevC = prev?.[layer];
-        if (isRotatingPlainCache(prevC) && !isRowBatchCache(prevC)) {
-          // Adopted lone row (Phase 3.2): a plain serial rotating cache —
-          // its chronological view is the merge's first row, same as a
-          // fresh solo (pad 0 by definition).
-          const [k0, v0] = prevC.temporalView();
-          rows.push({ keys: k0, values: v0 });
-          offsets.push(prevC.offset);
-        }
-        // Route by CAPABILITY (isRowBatchCache), never by signature string,
-        // inside the rot branch: the running batch's ring is a row-batch
-        // cache whatever its storage kind. The 2026-08-22 agg×4 regression
-        // (443f333) came from a signature-based conjunct here silently
-        // dropping the running rows; since then Cache.signature() is
-        // REQUIRED (BatchedRotatingCache reports "kv:rotating-plain") and
-        // the quant family is dispatched by its own branch above.
-        const prevRot = prevC && isRowBatchCache(prevC)
-          ? prevC as unknown as BatchedRotatingCache
-          : undefined;
-        if (prevRot) {
-          const [k0, v0] = prevRot.temporalView(); // [B,H,valid,D]
-          const [, H, valid, D] = k0.shape as [number, number, number, number];
-          const vD = v0.shape[3]!;
-          for (let b = 0; b < B; b++) {
-            const pad = Math.max(0, prevRot.leftPad[b]!);
-            rows.push({
-              keys: k0.slice([b, 0, pad, 0], [b + 1, H, valid, D]),
-              values: v0.slice([b, 0, pad, 0], [b + 1, H, valid, vD]),
-            });
-            offsets.push(prevRot.offsetArr[b]!);
-          }
-          k0.dispose(); v0.dispose();
-        }
-        rows.push(newRow);
-        offsets.push(soloC.offset);
-        newInners.push(BatchedRotatingCache.merge(rows, offsets, this.#rotMaxSize[layer]!));
-        for (const r of rows) { r.keys.dispose(); r.values.dispose(); }
-      } else {
-        const prevFull = prev?.[layer] as KVCache | undefined;
-        if (prevFull && this.#runtime.value("MLX_BUN_BATCH_EXTEND") !== "0") {
-          // extend-join (mlx-lm BatchKVCache.extend semantics, P0): append
-          // the new right-justified row to the running buffer in ONE pad +
-          // ONE concat — no per-row extraction. Existing pads grow, never
-          // shrink (the re-merge below re-normalizes them instead; both are
-          // masked, both token-exact vs their mlx-lm protocol twin).
-          // MLX_BUN_BATCH_EXTEND=0 = the O(B·S) re-merge, kill switch/A-B.
-          const [k0, v0] = prevFull.temporalView(); // [B,H,off,D]
-          const ext = extendKVRows(k0, v0, prevPad, newRow);
-          k0.dispose(); v0.dispose();
-          newFullPad = ext.leftPad;
-          const c = new KVCache();
-          c.restoreState(ext.keys, ext.values, ext.width);
-          newInners.push(c);
-          newRow.keys.dispose(); newRow.values.dispose();
-        } else {
-          const rows: Row1[] = [];
-          if (prevFull) {
-            const [k0, v0] = prevFull.temporalView(); // [B,H,off,D]
-            const [, H, off, D] = k0.shape as [number, number, number, number];
-            const vD = v0.shape[3]!;
-            for (let b = 0; b < B; b++) {
-              const pad = prevPad[b]!;
-              rows.push({
-                keys: k0.slice([b, 0, pad, 0], [b + 1, H, off, D]),
-                values: v0.slice([b, 0, pad, 0], [b + 1, H, off, vD]),
-              });
-            }
-            k0.dispose(); v0.dispose();
-          }
-          rows.push(newRow);
-          const merged = mergeKVRows(rows);
-          newFullPad = merged.leftPad;
-          const c = new KVCache();
-          c.restoreState(merged.keys, merged.values, merged.width);
-          newInners.push(c);
-          for (const r of rows) { r.keys.dispose(); r.values.dispose(); }
-        }
-      }
-    }
-    if (prev) for (const c of prev) c.dispose();
-    // An adopted row's entry-backed arrays are gone after the prev dispose;
-    // run its retain now. The joiner's likewise after its solo dispose.
-    this.#adoptedRetain?.();
-    this.#adoptedRetain = null;
-    const completed = p.solo; p.solo = [];
-    const release = p.retain; p.retain = undefined;
-    disposeResources([...completed, { dispose: () => release?.() }]);
-    // Every row in a REAL merge has its KV interleaved in batched buffers —
-    // no longer prompt-cache put() candidates.
-    for (const r of this.#running) r.merged = true;
-    p.row.merged = true;
-    this.#inners = newInners;
-    this.#fullLeftPad = newFullPad;
-    this.#running.push(p.row);
-  }
-
-  /** One PIPELINED batched decode step (mlx-lm GenerationBatch._step):
-   *  1. forward all rows from the UNREAD pending token array (or, pipeline
-   *     cold, from the rows' last emitted tokens), sample each live row's next
-   *     token on its [1,V] slice, asyncEval the new [B] token array;
-   *  2. THEN sync-read the previous step's tokens (overlapping the readback
-   *     with this step's compute), emit them, and evict finished rows.
-   *  Rows that finish get one extra harmless KV write from the already-built
-   *  step; filter drops the row (mlx-lm behaves identically). Length-finished
-   *  rows are known in advance and are NOT sampled (placeholder slot). */
-  async #step(): Promise<void> {
-    if (this.#stepTrace) {
-      const now = performance.now();
-      if (STEP_T.lastEnd) STEP_T.gap += now - STEP_T.lastEnd;
-      STEP_T.t0 = now;
-    }
-    const rows = this.#running;
-    const B = rows.length;
-    const inners = this.#inners!;
-
-    // A row is live if it still needs tokens sampled; a row whose pending
-    // unread token is its last (sampled == maxTokens) only awaits emission.
-    const anyLive = rows.some((r) => r.sampled < r.req.maxTokens);
-    // Grammar (B1): if any row has a LIVE grammar controller (not terminated,
-    // still sampling), take the read-before-build shape — the matcher's
-    // acceptToken needs the token id as a JS number, so the previous step's
-    // [B] token array is read back NOW (before building the next graph),
-    // accept()ed per row (firing async fills that overlap the forward), then
-    // ready() is awaited before this step's sample. This is the serial loop's
-    // grammar resolution transplanted to the batch. Batches with NO live
-    // grammar row keep the pipelined path byte-identical (zero cost when
-    // unused). The trade: while a grammar row is live the batch runs
-    // effectively NO_PIPELINE (readback no longer overlaps GPU compute) —
-    // bounded by the readback (~0.1 ms) + fills (0.004–0.19 ms/row, overlapped
-    // with the graph build). Serial grammar pays the identical trade today.
-    const hasLiveGrammar = anyLive && rows.some(
-      (r) => r.req.grammar && !r.req.grammar.isTerminated,
-    );
-    if (hasLiveGrammar) return this.#stepGrammar();
-
-    let nextToks: MlxArray | null = null;
-    let nextReal: boolean[] | null = null;
-    if (anyLive) {
-      // Fed-token accounting (prompt-cache put): a COLD step feeds each
-      // row's `current` (values known now, always a real emitted token); a
-      // pipelined step feeds the pending array, whose values are pushed at
-      // the read below gated on the per-slot real flags.
-      if (!this.#pendingToks) for (const r of rows) r.fed.push(r.current);
-      // Per-layer forward cache: rot layers use the persistent
-      // BatchedRotatingCache directly; ssm layers are already [B,...] state
-      // with no padding (no mask, no per-row RoPE — pass through); full
-      // layers get a fresh BatchedDecodeMaskCache wrapper — UNLESS no row
-      // has left padding. UNPADDED FAST PATH (unified-engine plan Phase 2,
-      // the B=1 case above all): with every leftPad 0 the wrapper's two
-      // jobs vanish — the padding mask (KVCache.makeMask(1) is the empty
-      // mask, exactly the serial loop's) and the per-row rope positions
-      // (every row sits at the shared scalar offset). The bare cache then
-      // dispatches the SAME per-step graph serial builds; the wrapper
-      // otherwise costs a host mask build + ~8 device nodes PER FULL LAYER
-      // PER TOKEN (the Phase-0 constant ~4–6 ms/step host tax at B=1).
-      const unpadded = this.#fullLeftPad.every((p) => p === 0);
-      // Compiled decode at B=1 (Phase 3.2): after adopt-don't-copy, a lone
-      // row's caches are SERIAL-CLASS, so the serial engine's compiled step
-      // replays the same C++ graph here — closing the batch lane's last
-      // B=1 host-tax gap (e4b's ~7%). Guards: gemma dense (constructor),
-      // serial-class caches (supports — a merged batch's BatchedRotating
-      // layers fail it), unpadded, and a uint32 pipeline register (the
-      // trace signature; per-row int32 samplers take the graph path).
-      // Grammar batches use #stepGrammar and stay on the graph path.
-      let lg: MlxArray | null = null;
-      let evalWith: MlxArray[] = [];
-      if (
-        this.#compiled && B === 1 && unpadded && this.#running[0]!.req.compiledDecode !== false &&
-        (!this.#pendingToks || this.#pendingToks.dtype === Dtype.uint32) &&
-        // A filtered-to-one BATCHED rot-quant cache subclasses the serial
-        // class (so supports() passes) but carries batched ring state —
-        // exclude it; only truly-adopted serial caches replay compiled.
-        !inners.some((c) => isRowBatchCache(c) && isRotatingQuantizedCache(c)) &&
-        CompiledDecode.supports(inners as Cache[])
-      ) {
-        let cur = this.#pendingToks;
-        let owned = false;
-        if (!cur) {
-          const i = ops.fromInt32([rows[0]!.current], [1]);
-          cur = i.astype(Dtype.uint32);
-          i.dispose();
-          owned = true;
-        }
-        try {
-          const r = this.#compiled.step(cur, inners as Cache[]);
-          lg = r.logits; // [1,1,V]
-          evalWith = r.evalWith;
-        } catch (e) {
-          // A failed step is transactional (see generate.ts) — safe to
-          // re-forward the same token on the graph path below.
-          this.#compiled = null;
-          console.warn(`batch lane: compiled decode disabled: ${e}`);
-        } finally {
-          if (owned) cur.dispose();
-        }
-      }
-      let fwd: Cache[] | null = null;
-      try {
-        if (!lg) {
-          fwd = inners.map((c) =>
-            isRowBatchCache(c) || isBatchableCache(c) || unpadded
-              ? c
-              : isQuantizedKvCache(c)
-                ? new BatchedQuantDecodeMaskCache(c, B, this.#fullLeftPad)
-                : new BatchedDecodeMaskCache(c, B, this.#fullLeftPad, null),
-          );
-          const ids = this.#pendingToks
-            ? ops.reshape(this.#pendingToks, [B, 1]) // feed the unread tokens
-            : ops.fromInt32(rows.map((r) => r.current), [B, 1]); // pipeline cold
-          const h = await this.#forwardHidden(ids, fwd);
-          ids.dispose();
-          lg = this.model.logitsFromHidden(h); // [B,1,V]
-          h.dispose();
-        }
-        const V = lg.shape[2]!;
-        // Vectorized homogeneous sampling (batching-perf-path P0): when every
-        // LIVE row is plain greedy, one log-softmax+argmax over [B,V] replaces
-        // B slice/sample/concat graphs. Per-row identical math (row-independent
-        // ops, same per-row shapes — argmax over log-softmax mirrors the
-        // closure's toLogprobs→argmax exactly, tie behavior included). Doomed
-        // rows get a real argmax instead of the placeholder 0 — equally
-        // harmless (the slot is filtered before it is ever emitted; its only
-        // use is one KV write on the row's own, about-to-evict row).
-        const vecOk =
-          this.#runtime.value("MLX_BUN_BATCH_VEC_SAMPLE") !== "0" &&
-          rows.every((r) => r.sampled >= r.req.maxTokens || r.req.plainGreedy);
-        // Doomed slots hold placeholders (vec path: a real argmax value,
-        // equally not part of the row's stream) — flag them so the fed
-        // accounting at the read stays per-row exact.
-        nextReal = rows.map((r) => r.sampled < r.req.maxTokens);
-        if (vecOk) {
-          const flat = ops.reshape(lg, [B, V]);
-          lg.dispose();
-          try { nextToks = independentGreedySampling.sample(flat); }
-          finally { flat.dispose(); }
-          for (const row of rows) if (row.sampled < row.req.maxTokens) row.sampled++;
-        } else {
-          const sampled: MlxArray[] = [];
-          for (let b = 0; b < B; b++) {
-            const row = rows[b]!;
-            if (row.sampled >= row.req.maxTokens) {
-              // Length-doomed row: evicted right after the emission below ever
-              // uses this slot as input — placeholder keeps the [B] alignment.
-              sampled.push(ops.fromInt32([0], [1]));
-              continue;
-            }
-            const rl = lg.slice([b, 0, 0], [b + 1, 1, V]);
-            const rl2 = ops.reshape(rl, [1, V]);
-            rl.dispose();
-            sampled.push(row.req.sample!(rl2, row.sampled));
-            row.sampled++;
-            rl2.dispose();
-          }
-          lg.dispose();
-          nextToks = ops.concatAxis(sampled, 0); // [B]
-          for (const t of sampled) t.dispose();
-        }
-        // dispatch; read NEXT iteration. evalWith: the compiled step's
-        // cache-update nodes must ride the same async_eval (generate.ts).
-        ops.asyncEvalAll([nextToks, ...evalWith]);
-      } finally {
-        // Free the step's RoPE arrays; do NOT dispose (full wrappers would free
-        // their persistent inner; rot caches persist across steps).
-        if (fwd) for (const c of fwd) (c as { releaseRopeArr?: () => void }).releaseRopeArr?.();
-      }
-    }
-    this.#steps++;
-    if (this.#steps % 256 === 0) clearCache(); // serial's cadence, not per-step
-    if (this.#stepTrace) STEP_T.build += performance.now() - STEP_T.t0;
-
-    // Read + emit the PREVIOUS step's tokens while the new step computes.
-    const prev = this.#pendingToks;
-    const prevReal = this.#pendingReal;
-    this.#pendingToks = nextToks;
-    this.#pendingReal = nextReal;
-    // Kill switch / A-B lever (house style, cf. MLX_BUN_COMPILED_DECODE=0):
-    // MLX_BUN_BATCH_NO_PIPELINE=1 reads THIS step's tokens synchronously —
-    // set from process start `prev` is always null, so the flush below IS the
-    // whole phase 2. Same math either way (pipelining is scheduling).
-    if (this.#noPipeline) {
-      if (this.#running.some(row => row.req.continuation)) this.#captureContinuations();
-      await this.#flushPipeline();
-      return;
-    }
-    if (prev) {
-      const tRead = this.#stepTrace ? performance.now() : 0;
-      const toks = prev.toIntTokens();
-      prev.dispose();
-      // These values were the step's forward input (fed) iff a forward ran.
-      // Placeholder slots are excluded from `fed` — and taint their row if
-      // they ever fed (see Row.fedTainted; structurally unreachable today).
-      if (anyLive)
-        for (let b = 0; b < B; b++) {
-          if (!prevReal || prevReal[b]) rows[b]!.fed.push(toks[b]!);
-          else rows[b]!.fedTainted = true;
-        }
-      if (this.#stepTrace) STEP_T.read += performance.now() - tRead;
-      const tEmit = this.#stepTrace ? performance.now() : 0;
-      await this.#emitRows(toks); // also filters #pendingToks on eviction
-      if (this.#stepTrace) { STEP_T.emit += performance.now() - tEmit; STEP_T.n++; }
-    }
-    if (this.#running.some(row => row.req.continuation)) this.#captureContinuations();
-    if (this.#stepTrace) STEP_T.lastEnd = performance.now();
-  }
-
-  readonly #checkpointAt = new WeakMap<Row, number>();
-  /** Cache extraction belongs to this existing ordinary numerical driver;
-   * the optional request port owns checkpoint eligibility and persistence. */
-  #captureContinuations(): void {
-    if (!this.#pendingToks || !this.#running.some(row => row.req.continuation)) return;
-    let pending: number[] | undefined;
-    const failed = new Set<number>();
-    for (let index = 0; index < this.#running.length; index++) {
-      const row = this.#running[index]!, policy = row.req.continuation;
-      if (!policy || row.fedTainted || row.fed.length !== row.generated ||
-          row.generated < (this.#checkpointAt.get(row) ?? policy.interval) ||
-          this.#pendingReal?.[index] === false) continue;
-      pending ??= this.#pendingToks.toIntTokens();
-      const tokens = [...row.req.promptIds, ...row.fed];
-      const caches = this.#running.length === 1 && !row.merged
-        ? cloneSingleRowState(this.#inners! as Cache[], this.#stateCodecs)
-        : this.#extractRowCaches(index, tokens.length);
-      if (!caches) continue;
-      try {
-        policy.captureOwned({ caches, cacheTokens: tokens, generatedTokens: row.generated,
-          pendingToken: pending[index]! });
-        this.#checkpointAt.set(row, (Math.floor(row.generated / policy.interval) + 1) * policy.interval);
-      } catch (error) { row.reject(error); failed.add(index); }
-    }
-    if (failed.size) this.#applyFilter(this.#running.flatMap((_, index) => failed.has(index) ? [] : [index]), true);
-  }
-
-  /** Read out the pipeline register (if any): emit its tokens and evict
-   *  finished rows, leaving the pipeline cold. Called before a join merges. */
-  async #flushPipeline(): Promise<void> {
-    const prev = this.#pendingToks;
-    if (!prev) return;
-    this.#pendingToks = null;
-    this.#pendingReal = null; // flushed values never fed — nothing to account
-    const toks = prev.toIntTokens();
-    prev.dispose();
-    // Grammar rows: the flushed tokens are EMITTED below, so their matchers
-    // must advance here — #stepGrammar's accept only covers tokens it reads
-    // from a live pending array, and after this flush it cold-starts (no
-    // accept). Skipping this left the matcher one token behind its stream on
-    // every mid-decode join → one-step-stale masks → invalid output (found
-    // by the feature-matrix conformance gate, 2026-07-03; the fill fired
-    // here is awaited by the next #stepGrammar's ready()).
-    for (let b = 0; b < this.#running.length && b < toks.length; b++) {
-      const g = this.#running[b]!.req.grammar;
-      if (g && !g.isTerminated) g.accept(toks[b]!);
-    }
-    await this.#emitRows(toks);
-  }
-
-  /** The read-before-build decode step for batches with ≥1 live grammar row
-   *  (B1). Mirrors the serial loop's grammar resolution: read the previous
-   *  step's [B] tokens NOW (acceptToken needs JS numbers), accept() per row
-   *  (fires async fills), build the forward graph (fills overlap), await
-   *  ready(), sample per live row (the closure applies applyMask), then emit
-   *  the values read in step 1 — no second readback. Terminated grammar rows
-   *  keep their [B] slot through the forward (placeholder, not sampled) and
-   *  finish("stop") in #emitRows. On a cold start (first step after prefill)
-   *  there is no pending array to read; the prefill already accepted token 0
-   *  and fired the fill, so we just await ready() + sample. */
-  async #stepGrammar(): Promise<void> {
-    const rows = this.#running;
-    const B = rows.length;
-    const inners = this.#inners!;
-
-    // (1) Read the pending [B] token array (host copy for accept). The device
-    //     array is kept for the forward input below; disposed after sampling.
-    //     Cold start: pendingToks null → prevVals empty (prefill handled tok0).
-    const prev = this.#pendingToks;
-    const prevVals: number[] =
-      prev ? prev.toIntTokens() : [];
-    if (this.#runtime.value("MLX_BUN_GRAMMAR_DEBUG") === "1")
-      console.log(`[sg] B=${B} prevVals=${JSON.stringify(prevVals)} current=${JSON.stringify(rows.map((r) => r.current))} sampled=${JSON.stringify(rows.map((r) => r.sampled))}`);
-
-    // (2) accept() per live grammar row — fires that row's async bitmask fill.
-    for (let b = 0; b < B && prevVals.length; b++) {
-      const g = rows[b]!.req.grammar;
-      if (g && !g.isTerminated) g.accept(prevVals[b]!);
-    }
-
-    let nextToks: MlxArray | null = null;
-    let nextReal: boolean[] | null = null;
-    const anyLive = rows.some((r) => r.sampled < r.req.maxTokens);
-    if (anyLive) {
-      // Fed-token accounting (prompt-cache put) — mirror of #step's read:
-      // real slots push, a placeholder that fed taints its row.
-      const prevReal = this.#pendingReal;
-      if (prev)
-        for (let b = 0; b < B; b++) {
-          if (!prevReal || prevReal[b]) rows[b]!.fed.push(prevVals[b]!);
-          else rows[b]!.fedTainted = true;
-        }
-      else for (const r of rows) r.fed.push(r.current);
-      // Unpadded fast path — same rule as #step (bare caches == serial graph).
-      const unpadded = this.#fullLeftPad.every((p) => p === 0);
-      const fwd: Cache[] = inners.map((c) =>
-        isRowBatchCache(c) || isBatchableCache(c) || unpadded
-          ? c
-          : isQuantizedKvCache(c)
-            ? new BatchedQuantDecodeMaskCache(c, B, this.#fullLeftPad)
-            : new BatchedDecodeMaskCache(c, B, this.#fullLeftPad, null),
-      );
-      try {
-        // (3) Build the forward graph (host-side; the fills overlap it).
-        const ids = prev
-          ? ops.reshape(prev, [B, 1]) // feed the unread tokens (device array)
-          : ops.fromInt32(rows.map((r) => r.current), [B, 1]); // pipeline cold
-        const h = await this.#forwardHidden(ids, fwd);
-        ids.dispose();
-        const lg = this.model.logitsFromHidden(h); // [B,1,V]
-        h.dispose();
-        const V = lg.shape[2]!;
-
-        // (4) await ready() on every live grammar row — the fills fired in (2)
-        //     overlapped the graph build above. Usually already resolved.
-        const liveGrammar = rows.filter(
-          (r) => r.req.grammar && !r.req.grammar.isTerminated,
-        );
-        if (liveGrammar.length)
-          await Promise.all(liveGrammar.map((r) => r.req.grammar!.ready()));
-
-        // (5) Sample per live row (the closure applies applyMask after the
-        //     logits processors). Terminated grammar rows + length-doomed rows
-        //     take a placeholder (one harmless KV write, then evicted).
-        const sampled: MlxArray[] = [];
-        nextReal = rows.map(
-          (r) => r.sampled < r.req.maxTokens && !r.req.grammar?.isTerminated,
-        );
-        for (let b = 0; b < B; b++) {
-          const row = rows[b]!;
-          if (
-            row.sampled >= row.req.maxTokens ||
-            row.req.grammar?.isTerminated
-          ) {
-            sampled.push(ops.fromInt32([0], [1]));
-            continue;
-          }
-          const rl = lg.slice([b, 0, 0], [b + 1, 1, V]);
-          const rl2 = ops.reshape(rl, [1, V]);
-          rl.dispose();
-          sampled.push(row.req.sample!(rl2, row.sampled));
-          row.sampled++;
-          rl2.dispose();
-        }
-        lg.dispose();
-        nextToks = ops.concatAxis(sampled, 0); // [B]
-        for (const t of sampled) t.dispose();
-        ops.asyncEvalAll([nextToks]);
-      } finally {
-        for (const c of fwd)
-          (c as { releaseRopeArr?: () => void }).releaseRopeArr?.();
-      }
-    }
-    this.#steps++;
-    if (this.#steps % 256 === 0) clearCache();
-
-    // The device array `prev` has now fed the forward + been read for accept;
-    // dispose it and install the new pending array. On cold start prev is null.
-    prev?.dispose();
-    this.#pendingToks = nextToks;
-    this.#pendingReal = nextReal;
-
-    // (6) Emit the values read in (1) — no second readback. Terminated grammar
-    //     rows finish("stop") via #emit's isTerminated check; the filter evicts.
-    //     Cold start: prevVals empty (prefill emitted tok0) → nothing to emit.
-    if (prevVals.length) await this.#emitRows(prevVals);
-    if (this.#running.some(row => row.req.continuation)) this.#captureContinuations();
-  }
-
-  /** Emit one read-back token per running row; evict finished rows. A row's
-   *  onToken throwing rejects THAT row and evicts it — siblings continue. */
-  async #emitRows(toks: number[]): Promise<void> {
-    const rows = this.#running;
-    const B = rows.length;
-    const keep: number[] = [];
-    const done: { b: number; row: Row }[] = []; // clean finishes (stop/length)
-    for (let b = 0; b < B; b++) {
-      const row = rows[b]!;
-      row.generated++;
-      let disp: "continue" | "stop" | "length";
-      try {
-        disp = await this.#emit(row, toks[b]!);
-      } catch (e) {
-        row.reject(e); // containment: this row only (mlx-lm `remove`)
-        row.merged = true; // poison: a rejected row is never put() back
-        row.fedTainted = true;
-        continue; // not kept → evicted by the filter below
-      }
-      if (disp === "continue") keep.push(b);
-      else {
-        done.push({ b, row });
-        this.#finish(row, disp);
-      }
-    }
-    if (keep.length < B) {
-      // Per-row prompt-cache extraction (mlx-lm server.py:864-880 →
-      // BatchGenerator.extract_cache): a MERGED row's KV lives interleaved
-      // in the batched inners, so it is pulled out into fresh serial caches
-      // BEFORE filter() mutates them. Never-merged lone rows skip this —
-      // #applyFilter's keep=[] path put()s their adopted caches zero-copy.
-      // Rejected rows aren't in `done`; the whole-batch error path calls
-      // #applyFilter(dropOnly) directly and never reaches here.
-      for (const { b, row } of done) this.#extractAndPut(b, row);
-      this.#applyFilter(keep);
-    }
-  }
-
-  /** Account one sampled token for a row. Mirrors generate(): EOS terminates
-   *  WITHOUT an onToken call; otherwise onToken(token) runs and `false` halts;
-   *  reaching maxTokens ends with "length". Advances row.current on continue. */
-  async #emit(row: Row, token: number, logprobs?: import("../../generate").TokenLogprobs): Promise<"continue" | "stop" | "length"> {
-    row.req.signal?.throwIfAborted();
-    this.#firstOutput(row);
-    if (row.generated === 1) clearCache();
-    if (row.req.eosTokenIds.includes(token)) return "stop";
-    const cont = await row.req.onToken(token, logprobs);
-    if (cont === false) return "stop";
-    // Grammar termination: the matcher is satisfied (e.g. closing `}` accepted).
-    // The final token has been delivered via onToken above; halt with "stop" so
-    // the row finishes + evicts rather than sampling into an all--inf mask.
-    if (row.req.grammar?.isTerminated) return "stop";
-    if (row.generated >= row.req.maxTokens) return "length";
-    row.current = token;
-    return "continue";
-  }
-
-  #firstOutput(row: Row): void {
-    if (row.generated === 1) {
-      if (this.#prefill?.rows.includes(row)) this.#preparationPublishedOutput = true;
-      row.firstTokenAt = performance.now();
-      row.closeTokenZero?.();
-      row.closeTokenZero = undefined;
-    }
-  }
-
-  #finish(row: Row, reason: "stop" | "length"): void {
-    row.req.continuation?.complete();
-    const now = performance.now();
-    const first = row.firstTokenAt || now;
-    row.resolve({
-      ...(row.spec ? { spec: row.spec } : {}),
-      promptTokens: row.promptTokens,
-      generatedTokens: row.generated,
-      cachedTokens: row.cachedTokens,
-      finishReason: reason,
-      prefillMs: row.admittedAt ? first - row.admittedAt : 0,
-      decodeMs: row.firstTokenAt ? now - row.firstTokenAt : 0,
-    });
-  }
-
-  /** Finish-time disposition of serial-class caches covering exactly
-   *  `tokens`: put() into the prompt cache when the hook is present and the
-   *  offset lines up (defensive — a mismatch means the accounting is wrong
-   *  and the entry would corrupt future hits), else dispose. `retain` rides
-   *  along per the PromptCacheEntry contract (runs after dispose). */
-  #putOrDispose(caches: Cache[], tokens: number[], retain?: () => void, namespace = ""): void {
-    if (this.#promptCache) {
-      // A never-merged row can still own a row layout. Persistence receives
-      // its existing serial representation, just like merged-row retirement.
-      const extracted: Cache[] = [], replaced: Cache[] = [];
-      let persistent: Cache[];
-      try {
-        persistent = caches.map(cache => {
-          if (!isBatchableCache(cache) || cache.batchSize !== 1) return cache;
-          const row = cache.extractRow(0);
-          extracted.push(row); replaced.push(cache);
-          return row;
-        });
-      } catch (error) { disposeResources(extracted); throw error; }
-      try { disposeResources(replaced); }
-      catch (error) { disposeResources(extracted); throw error; }
-      caches = persistent;
-      const withOff = caches.find(
-        (c) => typeof (c as { offset?: unknown }).offset === "number",
-      ) as { offset: number } | undefined;
-      if (withOff && withOff.offset === tokens.length) {
-        this.#promptCache.put(tokens, caches, namespace, retain);
-        return;
-      }
-    }
-    for (const c of caches) c.dispose();
-    retain?.();
-  }
-
-  /** Extract a finishing MERGED row's KV into fresh serial caches and put()
-   *  them keyed by [promptIds + fed] — the batch-lane mirror of mlx-lm
-   *  server.py:872 (extract_cache → prompt_cache.insert_cache). Gates:
-   *  promptTokens >= 256 (the boundary-snapshot substantiality gate,
-   *  server.ts) and an exact coverage key (!fedTainted). Refusal
-   *  (#extractRowCaches null) just disposes-by-omission — the row's KV dies
-   *  with the filter, exactly the pre-extraction behavior. */
-  #extractAndPut(b: number, row: Row): void {
-    if (!this.#promptCache || !row.merged || row.fedTainted) return;
-    if (row.promptTokens < 256) return;
-    const tokens = [...row.req.promptIds, ...row.fed];
-    const caches = this.#extractRowCaches(b, tokens.length);
-    if (!caches) return;
-    // Materialize the owned copies without stalling the in-flight step (the
-    // slices depend on this step's KV writes): async — the batched source
-    // buffers free once the copies land, instead of being pinned by a lazy
-    // graph inside an idle cache entry.
-    withResource(leaseCacheStates(caches), state => ops.asyncEvalAll([...state]));
-    this.#putOrDispose(caches, tokens, undefined, row.cacheNamespace);
-  }
-
-  /** Row `b` of every layer as OWNED serial-class caches, or null when a
-   *  layer kind can't be extracted (then the caller drops the row's KV as
-   *  before). Bit-exactness: merge/extend/filter/decode are byte-preserving
-   *  per row (tests/batched-decode-parity, tests/batched-rotating,
-   *  tests/batched-rotating-quant) and each extract is a pure slice+copy of
-   *  those bytes (tests/batched-extract), so an extracted row's bytes ==
-   *  the solo run's. */
-  #extractRowCaches(b: number, expectTokens: number): Cache[] | null {
-    const out: Cache[] = [];
-    for (const inner of this.#inners!) {
-      let c: Cache | null;
-      if (isBatchableCache(inner)) c = inner.extractRow(b);
-      else if (isRowBatchCache(inner) && cacheSignature(inner) === "ssm")
-        // Coverage gate: recurrent state is UNTRIMMABLE, so an entry is only
-        // valid when the row's own advance count equals the [promptIds+fed]
-        // key EXACTLY (a mismatched entry would silently corrupt every future
-        // exact-hit). Defensive — a miss here means the fed accounting broke.
-        c = (inner as SSMCache).conv && (inner as SSMCache).rowOffset(b) === expectTokens
-          ? inner.extractRow(b)
-          : null;
-      else if (isRowBatchCache(inner)) c = inner.extractRow(b);
-      else if (
-        isRotatingQuantizedCache(inner) || // adopted serial state never
-        isRotatingPlainCache(inner) //     coexists with a merged row (defensive)
-      )
-        c = null;
-      else if (isQuantizedKvCache(inner))
-        c = inner.keys ? extractQuantRow(inner, this.#fullLeftPad[b]!, b) : null;
-      else if (isPlainKvCache(inner))
-        c = inner.keys ? extractKVRow(inner, this.#fullLeftPad[b]!, b) : null;
-      else c = null;
-      if (!c) {
-        for (const d of out) d.dispose();
-        return null;
-      }
-      out.push(c);
-    }
-    return out;
-  }
-
-  /** Evict rows not in `keep` (sorted ascending) from the batched KV and the
-   *  pipeline register. */
-  #applyFilter(keep: number[], dropOnly = false): void {
-    if (this.#method) {
-      this.#method.filterRows(keep, dropOnly);
-      this.#running = keep.map(row => this.#running[row]!);
-      return;
-    }
-    const inners = this.#inners!;
-    if (keep.length === 0) {
-      // Prompt-cache put (Phase 3.2): a lone NEVER-MERGED row's inners are
-      // its adopted serial-class caches, covering exactly prompt+fed — hand
-      // them back to the cache instead of disposing. dropOnly (the batch-
-      // drop error path) and poisoned rows dispose as before; MERGED rows'
-      // KV was already extracted per row in #emitRows (owned copies), so
-      // disposing the batched inners here is safe either way.
-      const solo =
-        !dropOnly && this.#running.length === 1 && !this.#running[0]!.merged
-          ? this.#running[0]!
-          : null;
-      if (solo) {
-        this.#putOrDispose(
-          inners as Cache[],
-          [...solo.req.promptIds, ...solo.fed],
-          this.#adoptedRetain ?? undefined,
-          solo.cacheNamespace,
-        );
-      } else {
-        for (const c of inners) c.dispose();
-        this.#adoptedRetain?.();
-      }
-      this.#adoptedRetain = null;
-      this.#inners = null;
-      this.#fullLeftPad = [];
-      this.#running = [];
-      this.#pendingToks?.dispose();
-      this.#pendingToks = null;
-      this.#pendingReal = null;
-      return;
-    }
-    const keptFullPad = keep.map((i) => this.#fullLeftPad[i]!);
-    // mlx-lm BatchKVCache.filter removes padding shared by all survivors.
-    // Full-attention width and padding move together; each row's absolute
-    // RoPE position stays unchanged. Rotating caches keep their own layout.
-    const trimFull = this.#kinds.includes("full") ? Math.min(...keptFullPad) : 0;
-    const out: LayerInner[] = [];
-    for (const inner of inners) {
-      if (isBatchableCache(inner)) {
-        inner.filterRows(keep);
-        out.push(inner);
-      } else if (isRowBatchCache(inner)) {
-        inner.filterRows(keep);
-        out.push(inner);
-      } else if (isRotatingQuantizedCache(inner) || isRotatingPlainCache(inner)) {
-        // Adopted lone-row state exists only at B=1, where the only filter
-        // is the keep=[] dispose-all handled above — unreachable.
-        throw new Error("applyFilter: adopted serial rotating cache cannot be row-filtered");
-      } else if (isQuantizedKvCache(inner)) {
-        const [k0, v0] = inner.temporalView();
-        const f = filterQuantRows(k0, v0, keep, trimFull);
-        for (const t of [k0, v0]) { t.packed.dispose(); t.scales.dispose(); t.biases.dispose(); }
-        const c = new QuantizedKVCache(inner.groupSize, inner.bits);
-        c.restoreState(f.keys, f.values, inner.offset - trimFull);
-        out.push(c);
-        inner.dispose();
-      } else if (isPlainKvCache(inner)) {
-        const [k0, v0] = inner.temporalView();
-        const f = filterKVRows(k0, v0, keep, trimFull);
-        k0.dispose(); v0.dispose();
-        const c = new KVCache();
-        c.restoreState(f.keys, f.values, inner.offset - trimFull);
-        out.push(c);
-        inner.dispose();
-      } else {
-        throw new Error(`applyFilter: unsupported cache signature ${cacheSignature(inner)}`);
-      }
-    }
-    this.#inners = out;
-    this.#fullLeftPad = keptFullPad.map((padding) => padding - trimFull);
-    this.#running = keep.map((i) => this.#running[i]!);
-    if (this.#pendingToks) {
-      const idx = ops.fromInt32(keep, [keep.length]);
-      const next = ops.takeAxis(this.#pendingToks, idx, 0);
-      idx.dispose();
-      this.#pendingToks.dispose();
-      this.#pendingToks = next;
-    }
-    if (this.#pendingReal) this.#pendingReal = keep.map((i) => this.#pendingReal![i]!);
-  }
-
-  #readToken(t: MlxArray): number {
-    const v = t.toIntTokens()[0]!;
-    t.dispose();
-    return v;
-  }
-}