bench(vm-jit): add SortedSet/string-text/concurrency/deep-copy kernels

Round out the slow-path baseline with the four remaining gaps:

- sorted_set_ops   — ordered insert+contains (the one missing collection)
- string_text_processing — split/pad_left/starts_with intrinsic group
- task_group_spawn — structured concurrency spawn + async-let join
- deep_copy_list   — per-iteration List copy (Rc/clone traffic)

Two separable runtime bugs surfaced and are flagged (README findings #3, #4;
plan §0.2):

- hash-Set is ~750x slower than an ordered set at the same insert+contains
  workload (set 1680x vs sorted_set_ops 2.2x reg/rust) — isolates the cost to
  the hash-Set impl, not dispatch.
- task_group_spawn scales ~quadratically in round count under every interpreted
  mode (eval == vm == jit: 0.34/9.9/725 ms at 100/1k/10k), a runtime bug
  (unreclaimed task frames), not a JIT one. Pinned to size 2000 and the runner
  gained a per-mode --timeout guard so a pathological case degrades to n/a
  instead of hanging the whole suite.

Refresh baseline-20260620.json (now 37 cases) and the README findings.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: olwang

benchmarks/vm-jit/README.md

@@ -17,9 +17,13 @@ From the repo root, inside the dev container:
 
 ```sh
 docker compose run --rm dev ./benchmarks/vm-jit/run-baseline.sh
-# options: --iterations N  --warmup N  --out PATH
+# options: --iterations N  --warmup N  --timeout SECS  --out PATH
 ```
 
+`--timeout` (default 180 s) caps each mode per case: a pathological kernel — e.g.
+a super-linear runtime path like `task_group_spawn` at large sizes — degrades to
+an `n/a` cell instead of hanging the whole suite. `--timeout 0` disables it.
+
 Each case runs through four modes and reports mean ms + tier ratios:
 
 | column | mode | meaning |
@@ -69,6 +73,7 @@ parsing doesn't taint per-function eligibility.
 | Linear recursion (depth) | ★ `linear_recursion` | sustained frame push/pop depth |
 | Float arithmetic | ★ `float_loop_sum` | Float family + Int↔Float / Float→String |
 | String build/inspect | ★ `string_build_scan` | concat alloc, slice, format |
+| String text-processing | ★ `string_text_processing` | `split`/`pad_left`/`starts_with` intrinsics |
 | String-keyed map | `map_string_keys` | string hashing |
 | User sum-type make/match | ★ `variant_match_loop` | `MakeVariant`/`MatchVariant` |
 | Option match | `match_option_loop` | option destructure |
@@ -82,11 +87,14 @@ parsing doesn't taint per-function eligibility.
 | Sorted map insert | `sorted_map_insert` | ordered insert |
 | Sorted map scan | `sorted_map_scan` | keys + get scan |
 | Set membership | ★ `set_insert_contains` | Set insert/contains |
+| Sorted-set membership | ★ `sorted_set_ops` | ordered insert + `contains` |
 | Deque FIFO | `deque_queue` | push_back/pop_front |
 | Bytes | ★ `bytes_scan` | Bytes construct/slice/len |
+| Deep value copy | ★ `deep_copy_list` | per-iter List copy (Rc/clone traffic) |
 | Stored `owned Fn` dynamic call | ★ `dynamic_closure_call` | indirect closure dispatch |
 | Json | `json_parse_access` | parse + field access |
 | Async call/await | `async_call_loop` | park/resume frame state |
+| Structured concurrency | ★ `task_group_spawn` | `task_group` spawn + async-let join |
 | Realistic mixed | `selfhost_manifest_inspector`, `selfhost_mailbox_bench` | end-to-end blends |
 
 ### Known not covered (intentionally)
@@ -107,11 +115,11 @@ speedup, **<1 = faster than the plain VM**).
 
    | kernel | reg_vm_ms | native_ms | rust_ms | nat/reg |
    |---|--:|--:|--:|--:|
-   | `native_scalar_loop` | 153.5 | 3.18 | 2.31 | **0.02** |
-   | `native_read_heap`   | 115.6 | 6.58 | 0.58 | **0.06** |
-   | `native_call_chain`  | 202.0 | 5.41 | 5.58 | **0.03** |
-   | `int_divmod_loop`    | 168.9 | 4.23 | 1.90 | **0.03** |
-   | `bool_logic_loop`    | 326.7 | 6.60 | 2.99 | **0.02** |
+   | `native_scalar_loop` | 153.5 | 3.20 | 2.41 | **0.02** |
+   | `native_read_heap`   | 115.5 | 6.77 | 0.61 | **0.06** |
+   | `native_call_chain`  | 196.3 | 5.33 | 5.56 | **0.03** |
+   | `int_divmod_loop`    | 179.3 | 4.20 | 1.96 | **0.02** |
+   | `bool_logic_loop`    | 347.2 | 6.70 | 3.21 | **0.02** |
 
    So the problem is **not** codegen quality — it's **eligibility/coverage**.
    The moment a function does anything outside the pure-scalar/read-heap subset
@@ -120,33 +128,53 @@ speedup, **<1 = faster than the plain VM**).
    eligibility (Phase 3) is likely the highest-ROI lever**, not just dispatch.
 
 2. **On the real (ineligible) kernels both JIT tiers do ~nothing** — `jit/reg`
-   and `nat/reg` hover at ~1.00, and frequently **>1.0 (native makes it
-   slower)**: `list_sort` 1.31, `map_int` 1.19, `closure_alloc` 1.13,
-   `recursion_fib` 1.09 — the tier translates part of the function, bails, and
-   eats the overhead. Tier-0 similarly regresses `json` (1.48), `dynamic_closure`
-   (1.66), `float` (1.27). A cheap early win: **don't attempt native on
-   functions that will predictably bail.**
-
-3. **`set_insert_contains` is pathological: reg/rust ≈ 1680×** (1774 ms vs
-   1.06 ms), vs the comparable `map_int` at 4.5×. Almost certainly a Set
-   implementation bug, not dispatch cost — flagged for its own investigation,
-   separate from the tier work.
-
-4. **Heap-variant & combinator paths are 290–655×** (`option_result_chain` 655×,
-   `match_option_loop` 331×, `variant_match_loop` 288×, `nested_struct_field`
-   380×) — make/match/allocate of Option/Result/sum/struct values dominates.
-
-5. **`linear_recursion` is the slowest non-Set kernel at 314×** (4.6 s) — deep
-   frame push/pop depth is expensive; the one kernel where tier-0 actually helps
-   a little (0.88).
-
-6. **String/Map/Json/Bytes kernels are closest to Rust** (1.5–9×) because their
-   work already lives in native runtime helpers — least to gain from VM-tier work.
+   and `nat/reg` hover at ~1.00, and frequently **>1.0 (the tier makes it
+   slower)**: tier-0 regresses `string_text_processing` **1.80×** (the worst in
+   the suite), `selfhost_mailbox` 1.21; native regresses `float` 1.23,
+   `closure_alloc` 1.16, `list_closure` 1.09 — the tier translates part of the
+   function, bails, and eats the overhead. (The exact offenders shift run to run;
+   the pattern — tiers *regress* ineligible code — is stable.) A cheap early win:
+   **don't attempt the JIT on functions that will predictably bail.**
+
+3. **`set_insert_contains` is pathological: reg/rust ≈ 1680×** (1796 ms vs
+   1.07 ms) — and the new `sorted_set_ops` kernel makes it a smoking gun: the
+   *same* insert+contains workload on an **ordered** set is **2.2×** reg/rust
+   (1.30 ms). So an ordered membership structure is ~750× faster than the hash
+   `Set` at the same job. This isolates the cost to the **hash-`Set`
+   implementation specifically** (almost certainly an O(n) or rehash-thrash bug),
+   not to dispatch or to set semantics generally. Its own, likely cheap, fix.
+
+4. **Structured concurrency is both slow and super-linear.** `task_group_spawn`
+   is **337× reg/rust** even at the small size it now runs (2 000 rounds), and
+   worse, it scales **≈ quadratically**: jit-internal measured 0.34 ms / 9.9 ms /
+   725 ms at 100 / 1 000 / 10 000 rounds (and identically under plain `eval`, so
+   it is a **runtime** bug, not a JIT one — completed `task_group` frames look
+   unreclaimed). The original 100 000-round size was uncapped and hung the suite;
+   the kernel is pinned to 2 000 and the runner now has a per-mode `--timeout`
+   guard so a pathological case degrades to `n/a` instead of hanging. Flag for
+   its own investigation alongside the Set bug.
+
+5. **Heap-variant & combinator paths are 340–660×** (`option_result_chain` 662×,
+   `match_option_loop` 362×, `variant_match_loop` 342×, `nested_struct_field`
+   374×) — make/match/allocate of Option/Result/sum/struct values dominates.
+
+6. **`linear_recursion` is the slowest non-pathological kernel at ~294×** (4.1 s)
+   — deep frame push/pop depth is expensive.
+
+7. **Deep value copy (Rc/clone) is a ~22× tax** (`deep_copy_list` 21.7×) — real
+   but well below the variant/struct make/match tier, so Phase 4 (value-rep)
+   stays low priority: clone traffic is not where the big wins are.
+
+8. **String/Map/Json/Bytes/SortedSet kernels are closest to Rust** (1.5–9×)
+   because their work already lives in native runtime helpers — least to gain
+   from VM-tier work.
 
 Implications for the plan: (a) Phase 3 (widen native eligibility) is re-weighted
 **up** — native is already near-Rust where it runs; (b) add a "predict-and-skip
-bail" guard so the tiers stop *regressing* ineligible code; (c) the Set anomaly
-(#3) is a separate, likely cheap, high-impact bug fix.
+bail" guard so the tiers stop *regressing* ineligible code (#2); (c) the hash-Set
+anomaly (#3) and the quadratic `task_group` runtime (#4) are separate, likely
+cheap, high-impact bug fixes that do not depend on the tier work; (d) Rc/clone
+traffic (#7) confirms Phase 4 can stay deferred.
 
 ## Adding a kernel
 

benchmarks/vm-jit/baseline/baseline-20260620.json

@@ -2,38 +2,42 @@
   "iterations": 5,
   "warmup": 1,
   "cases": [
-    {"category":"native-scalar","case":"native_scalar_loop.rss","size":"2000000","reg_vm_ms":153.457683,"jit_ms":153.111989,"native_ms":3.184386,"rust_ms":2.314222},
-    {"category":"native-read-heap","case":"native_read_heap.rss","size":"2000000","reg_vm_ms":115.55809,"jit_ms":114.279249,"native_ms":6.577557,"rust_ms":0.583562},
-    {"category":"native-call","case":"native_call_chain.rss","size":"1500000","reg_vm_ms":201.961433,"jit_ms":206.206385,"native_ms":5.410725,"rust_ms":5.574784},
-    {"category":"int-divmod","case":"int_divmod_loop.rss","size":"2000000","reg_vm_ms":168.89069,"jit_ms":179.23963,"native_ms":4.232961,"rust_ms":1.899341},
-    {"category":"bool-logic","case":"bool_logic_loop.rss","size":"2000000","reg_vm_ms":326.721717,"jit_ms":336.44246,"native_ms":6.596299,"rust_ms":2.987159},
-    {"category":"int-arith","case":"pure_loop_sum.rss","size":"2000000","reg_vm_ms":247.106757,"jit_ms":257.33228,"native_ms":247.356443,"rust_ms":3.835658},
-    {"category":"struct-field","case":"struct_field_rw.rss","size":"200000","reg_vm_ms":44.128567,"jit_ms":43.695347,"native_ms":42.81113,"rust_ms":0.166535},
-    {"category":"struct-nested","case":"nested_struct_field.rss","size":"300000","reg_vm_ms":93.073526,"jit_ms":94.55333,"native_ms":94.570614,"rust_ms":0.244677},
-    {"category":"call-static","case":"function_call_hot_loop.rss","size":"200000","reg_vm_ms":37.099722,"jit_ms":37.16694,"native_ms":36.640618,"rust_ms":0.561213},
-    {"category":"recursion-tree","case":"recursion_fib.rss","size":"30","reg_vm_ms":210.577398,"jit_ms":216.230689,"native_ms":228.890624,"rust_ms":2.969392},
-    {"category":"recursion-linear","case":"linear_recursion.rss","size":"50000","reg_vm_ms":4647.048691,"jit_ms":4097.047955,"native_ms":4151.038991,"rust_ms":14.818337},
-    {"category":"float","case":"float_loop_sum.rss","size":"1000000","reg_vm_ms":75.306713,"jit_ms":95.797992,"native_ms":76.451107,"rust_ms":1.526888},
-    {"category":"string","case":"string_build_scan.rss","size":"200000","reg_vm_ms":36.914004,"jit_ms":37.777445,"native_ms":37.132889,"rust_ms":13.220889},
-    {"category":"string-map","case":"map_string_keys.rss","size":"16000","reg_vm_ms":17.810704,"jit_ms":12.4346,"native_ms":6.881151,"rust_ms":2.92115},
-    {"category":"variant-user","case":"variant_match_loop.rss","size":"300000","reg_vm_ms":108.672445,"jit_ms":125.023662,"native_ms":111.837014,"rust_ms":0.376861},
-    {"category":"variant-option","case":"match_option_loop.rss","size":"200000","reg_vm_ms":31.245447,"jit_ms":30.868393,"native_ms":30.649216,"rust_ms":0.094525},
-    {"category":"variant-combinator","case":"option_result_chain.rss","size":"60000","reg_vm_ms":47.72894,"jit_ms":46.930591,"native_ms":52.329538,"rust_ms":0.072825},
-    {"category":"list-scan","case":"list_index_scan.rss","size":"19000","reg_vm_ms":1.973841,"jit_ms":1.656981,"native_ms":1.591647,"rust_ms":0.019266},
-    {"category":"list-closure","case":"list_closure_pipeline.rss","size":"12000","reg_vm_ms":4.225353,"jit_ms":3.598439,"native_ms":3.772532,"rust_ms":0.041908},
-    {"category":"closure-alloc","case":"closure_alloc_loop.rss","size":"300000","reg_vm_ms":33.552675,"jit_ms":33.422957,"native_ms":38.049689,"rust_ms":0.128301},
-    {"category":"list-pipeline","case":"pipeline_chain.rss","size":"16000","reg_vm_ms":3.383271,"jit_ms":3.309203,"native_ms":3.716182,"rust_ms":0.055567},
-    {"category":"list-sort","case":"list_sort.rss","size":"4000","reg_vm_ms":18.553223,"jit_ms":18.217876,"native_ms":24.354669,"rust_ms":0.434442},
-    {"category":"map-int","case":"map_insert_lookup.rss","size":"17000","reg_vm_ms":2.695618,"jit_ms":2.662094,"native_ms":3.200924,"rust_ms":0.593669},
-    {"category":"sorted-map-insert","case":"sorted_map_insert.rss","size":"8000","reg_vm_ms":2.434344,"jit_ms":2.337164,"native_ms":2.433319,"rust_ms":0.565904},
-    {"category":"sorted-map-scan","case":"sorted_map_scan.rss","size":"12000","reg_vm_ms":3.637075,"jit_ms":3.695273,"native_ms":3.674432,"rust_ms":0.918024},
-    {"category":"set","case":"set_insert_contains.rss","size":"30000","reg_vm_ms":1773.996818,"jit_ms":1771.040505,"native_ms":1763.028911,"rust_ms":1.055736},
-    {"category":"deque","case":"deque_queue.rss","size":"50000","reg_vm_ms":4.766012,"jit_ms":4.816395,"native_ms":4.840669,"rust_ms":0.118906},
-    {"category":"bytes","case":"bytes_scan.rss","size":"200000","reg_vm_ms":20.87134,"jit_ms":21.51582,"native_ms":20.855232,"rust_ms":2.205859},
-    {"category":"closure-dynamic","case":"dynamic_closure_call.rss","size":"200000","reg_vm_ms":41.054785,"jit_ms":67.972256,"native_ms":41.517968,"rust_ms":0.36911},
-    {"category":"json","case":"json_parse_access.rss","size":"50000","reg_vm_ms":51.280614,"jit_ms":75.74428,"native_ms":50.29214,"rust_ms":34.17042},
-    {"category":"async","case":"async_call_loop.rss","size":"200000","reg_vm_ms":36.875962,"jit_ms":36.746431,"native_ms":37.798412,"rust_ms":null},
-    {"category":"selfhost-manifest","case":"selfhost_manifest_inspector.rss","size":"benchmarks/micro/fixtures/package-medium/rsspkg.toml","reg_vm_ms":0.097848,"jit_ms":0.046441,"native_ms":0.04854,"rust_ms":0.051507},
-    {"category":"selfhost-mailbox","case":"selfhost_mailbox_bench.rss","size":"60000","reg_vm_ms":112.603209,"jit_ms":80.854286,"native_ms":81.589531,"rust_ms":0.437867}
+    {"category":"native-scalar","case":"native_scalar_loop.rss","size":"2000000","reg_vm_ms":153.450249,"jit_ms":153.371398,"native_ms":3.201323,"rust_ms":2.412036},
+    {"category":"native-read-heap","case":"native_read_heap.rss","size":"2000000","reg_vm_ms":115.515553,"jit_ms":118.760618,"native_ms":6.772098,"rust_ms":0.612261},
+    {"category":"native-call","case":"native_call_chain.rss","size":"1500000","reg_vm_ms":196.302776,"jit_ms":197.774866,"native_ms":5.327675,"rust_ms":5.555101},
+    {"category":"int-divmod","case":"int_divmod_loop.rss","size":"2000000","reg_vm_ms":179.280179,"jit_ms":166.543577,"native_ms":4.197145,"rust_ms":1.958134},
+    {"category":"bool-logic","case":"bool_logic_loop.rss","size":"2000000","reg_vm_ms":347.169502,"jit_ms":338.2524,"native_ms":6.694568,"rust_ms":3.212792},
+    {"category":"int-arith","case":"pure_loop_sum.rss","size":"2000000","reg_vm_ms":240.724204,"jit_ms":241.646979,"native_ms":245.257864,"rust_ms":3.860776},
+    {"category":"struct-field","case":"struct_field_rw.rss","size":"200000","reg_vm_ms":43.875971,"jit_ms":43.066937,"native_ms":43.761612,"rust_ms":0.165308},
+    {"category":"struct-nested","case":"nested_struct_field.rss","size":"300000","reg_vm_ms":92.842627,"jit_ms":95.59277,"native_ms":97.32352,"rust_ms":0.24795},
+    {"category":"call-static","case":"function_call_hot_loop.rss","size":"200000","reg_vm_ms":37.870419,"jit_ms":36.868427,"native_ms":38.519386,"rust_ms":0.545433},
+    {"category":"recursion-tree","case":"recursion_fib.rss","size":"30","reg_vm_ms":224.940183,"jit_ms":213.232579,"native_ms":211.518604,"rust_ms":2.820809},
+    {"category":"recursion-linear","case":"linear_recursion.rss","size":"50000","reg_vm_ms":4156.306189,"jit_ms":4099.619487,"native_ms":4147.873087,"rust_ms":14.13297},
+    {"category":"float","case":"float_loop_sum.rss","size":"1000000","reg_vm_ms":72.741296,"jit_ms":73.878812,"native_ms":89.474229,"rust_ms":1.498632},
+    {"category":"string","case":"string_build_scan.rss","size":"200000","reg_vm_ms":37.227373,"jit_ms":36.949423,"native_ms":36.891548,"rust_ms":14.551392},
+    {"category":"string-text","case":"string_text_processing.rss","size":"100000","reg_vm_ms":44.948627,"jit_ms":80.683534,"native_ms":47.285951,"rust_ms":28.514427},
+    {"category":"string-map","case":"map_string_keys.rss","size":"16000","reg_vm_ms":7.651754,"jit_ms":7.253979,"native_ms":6.903029,"rust_ms":3.07649},
+    {"category":"variant-user","case":"variant_match_loop.rss","size":"300000","reg_vm_ms":125.618428,"jit_ms":108.532653,"native_ms":110.234594,"rust_ms":0.367141},
+    {"category":"variant-option","case":"match_option_loop.rss","size":"200000","reg_vm_ms":30.798301,"jit_ms":29.998618,"native_ms":31.626625,"rust_ms":0.085174},
+    {"category":"variant-combinator","case":"option_result_chain.rss","size":"60000","reg_vm_ms":48.3353,"jit_ms":45.721135,"native_ms":51.190782,"rust_ms":0.073008},
+    {"category":"list-scan","case":"list_index_scan.rss","size":"19000","reg_vm_ms":1.62429,"jit_ms":1.609049,"native_ms":1.583207,"rust_ms":0.017175},
+    {"category":"list-closure","case":"list_closure_pipeline.rss","size":"12000","reg_vm_ms":3.47659,"jit_ms":3.600923,"native_ms":3.777506,"rust_ms":0.038183},
+    {"category":"closure-alloc","case":"closure_alloc_loop.rss","size":"300000","reg_vm_ms":34.232657,"jit_ms":34.585007,"native_ms":39.661796,"rust_ms":0.124166},
+    {"category":"list-pipeline","case":"pipeline_chain.rss","size":"16000","reg_vm_ms":3.468223,"jit_ms":3.383273,"native_ms":3.550698,"rust_ms":0.054074},
+    {"category":"list-sort","case":"list_sort.rss","size":"4000","reg_vm_ms":18.384282,"jit_ms":17.547816,"native_ms":17.495865,"rust_ms":0.482924},
+    {"category":"map-int","case":"map_insert_lookup.rss","size":"17000","reg_vm_ms":2.705407,"jit_ms":2.761598,"native_ms":2.62059,"rust_ms":0.573258},
+    {"category":"sorted-map-insert","case":"sorted_map_insert.rss","size":"8000","reg_vm_ms":2.427648,"jit_ms":2.422907,"native_ms":2.405448,"rust_ms":0.567841},
+    {"category":"sorted-map-scan","case":"sorted_map_scan.rss","size":"12000","reg_vm_ms":3.707281,"jit_ms":3.768256,"native_ms":3.700639,"rust_ms":0.926999},
+    {"category":"set","case":"set_insert_contains.rss","size":"30000","reg_vm_ms":1796.190578,"jit_ms":1745.084837,"native_ms":1732.635993,"rust_ms":1.067999},
+    {"category":"sorted-set","case":"sorted_set_ops.rss","size":"8000","reg_vm_ms":1.297166,"jit_ms":1.243474,"native_ms":1.288466,"rust_ms":0.582866},
+    {"category":"deque","case":"deque_queue.rss","size":"50000","reg_vm_ms":4.658297,"jit_ms":4.760022,"native_ms":4.720656,"rust_ms":0.109524},
+    {"category":"bytes","case":"bytes_scan.rss","size":"200000","reg_vm_ms":20.877156,"jit_ms":21.167498,"native_ms":20.374739,"rust_ms":2.347498},
+    {"category":"deep-copy","case":"deep_copy_list.rss","size":"100000","reg_vm_ms":68.505373,"jit_ms":69.119698,"native_ms":69.659689,"rust_ms":3.151415},
+    {"category":"closure-dynamic","case":"dynamic_closure_call.rss","size":"200000","reg_vm_ms":37.551014,"jit_ms":37.879705,"native_ms":39.199713,"rust_ms":0.360008},
+    {"category":"json","case":"json_parse_access.rss","size":"50000","reg_vm_ms":50.076174,"jit_ms":50.481908,"native_ms":49.539274,"rust_ms":31.489825},
+    {"category":"async","case":"async_call_loop.rss","size":"200000","reg_vm_ms":36.381123,"jit_ms":36.349031,"native_ms":37.444839,"rust_ms":null},
+    {"category":"async-taskgroup","case":"task_group_spawn.rss","size":"2000","reg_vm_ms":31.691751,"jit_ms":31.264826,"native_ms":30.398209,"rust_ms":0.0941},
+    {"category":"selfhost-manifest","case":"selfhost_manifest_inspector.rss","size":"benchmarks/micro/fixtures/package-medium/rsspkg.toml","reg_vm_ms":0.046441,"jit_ms":0.043958,"native_ms":0.049116,"rust_ms":0.045941},
+    {"category":"selfhost-mailbox","case":"selfhost_mailbox_bench.rss","size":"60000","reg_vm_ms":79.176401,"jit_ms":96.132959,"native_ms":85.825131,"rust_ms":0.433083}
   ]
 }