Publish 1P1C and 4P4C; drop SPSC/MPMC filter names

.grok/skills/improve-docs/references/STYLE.md

@@ -6,7 +6,8 @@ Keep it simple. Prefer delete and clarify over decorate.
 
 | Say | Never say |
 |-----|-----------|
-| **payload size** (256 B, 4 KiB) for the published filter | fixture; five leftover type names as if they were different shapes |
+| **payload size** (256 B, 4 KiB) for the published size filter | fixture; five leftover type names as if they were different shapes |
+| **pattern** as **1P1C**, **4P4C**, **1P4C**, **4P1C** | SPSC, MPMC on the published filter |
 | **data type** only as the catalog id (`size_256`, `size_4096`) | fixture, fixtures, fixtureKey, `dataset.fixtures` |
 
 The published axis is how many bytes each queued item is. Experiment 13 showed that 256 B and 4 KiB tell different stories; 512 B / 1 KiB / 2 KiB do not. Use **payload size** in README, Dashboard copy, and findings. Catalog ids stay `type_id` values (`size_256`, `size_4096`). Older folders may still say `message` / `document` (same lengths). Do not keep `fixture*` as an internal alias. Old `configs.json` may still contain `fixtures`; read it as a fallback, write `data_types`.
@@ -29,7 +30,7 @@ Do not rank a library on payload size. Every queue in a cell moves the same byte
 ## Content style
 
 - **One idea per paragraph.** Prefer tables for role/path matrices **on the site / Dashboard**, not by fattening the root README.
-- **User terms:** payload size (256 B, 4 KiB), mode (bytes/stream), ops/s, latency, Pareto, baseline.
+- **User terms:** payload size (256 B, 4 KiB), pattern (1P1C, 4P4C), ops/s, latency, Pareto, baseline. Do not say SPSC or MPMC on the dashboard filter.
 - **Never say “fixture”.** The published sample is a **payload size**. The catalog id is a **data type** (`size_256`, `size_4096`). Say that in docs, Dashboard copy, skills, comments, identifiers, and JSON keys. Do not keep `fixture*` as an internal alias “until later.”
 - **Avoid in user copy:** fixture, median size (we do not measure payload size as a result), harness (prefer benchmark runner), unexplained IQR/P95.
 - **Honesty line** when ranks appear: within one language; cross-lang directional — prefer **one** place (e.g. Statistics / Method), not a second essay block on README.

README.md

@@ -131,7 +131,7 @@ Every language writes the same columns (nanoseconds). Domain mapping:
 | `TimeEnq` | Enqueue ns |
 | `TimeDeq` | Dequeue ns |
 | `TimeHandoff` | Handoff ns |
-| `Pattern` | `bytes` = **SPSC**, `stream` = **MPMC** (not I/O) |
+| `Pattern` | `bytes` = **1P1C**, `4p4c` = **4P4C** (`stream` is leftover 2P2C, not I/O) |
 | `CpuTimeNs` | Process CPU time (spin vs block) |
 
 See [Metrics](https://leo-gan.github.io/queue-benchmark/analysis/METRICS/) and [architecture](https://leo-gan.github.io/queue-benchmark/analysis/architecture/).

config/benchmark_config.yaml

@@ -67,15 +67,15 @@ test_data:
       description: "4 KiB payload (copy cost starts to matter)"
       supports_circular: false
 
-# Pattern axis (CSV StringOrStream). Logged names are serializer ABI leftovers:
-# bytes = SPSC, stream = MPMC. Docs and dashboard say SPSC / MPMC, never "stream API".
+# Pattern axis (CSV StringOrStream). Logged names are serializer ABI leftovers.
+# Docs and dashboard say 1P1C / 2P2C / 1P4C / 4P1C / 4P4C, never "stream API".
 io_modes:
   - name: bytes
-    aliases: [string, buffer, spsc]
-    description: "SPSC: single producer, single consumer"
+    aliases: [string, buffer, spsc, 1p1c]
+    description: "1P1C: one producer, one consumer"
   - name: stream
-    aliases: [stream, mpmc]
-    description: "MPMC: 2 producers, 2 consumers (when the library supports it)"
+    aliases: [stream, mpmc, 2p2c]
+    description: "2P2C: two producers, two consumers (when the library supports it)"
   - name: 1p4c
     aliases: [1p4c]
     description: "1 producer, 4 consumers"

config/library/README.md

@@ -4,7 +4,7 @@ Named **run configs** select the measurement matrix:
 
 - `types` (axis W): `type_id` + `type_config` (payload shape)
 - `data_type_instance_count` (axis C): items moved per repetition
-- `execution.io_modes`: `bytes` = **SPSC**, `stream` = **MPMC** (legacy name; not I/O)
+- `execution.io_modes`: `bytes` = **1P1C**, `4p4c` = **4P4C** (CSV leftover `stream` is 2P2C; not I/O)
 
 Published type ids: `size_256` (256 B) · `size_4096` (4 KiB).  
 Older experiment folders still use `message` / `document` (same lengths).  
@@ -14,8 +14,8 @@ Catalog: `schemas/data_catalog_v2.yaml`.
 
 | File | Purpose |
 |------|---------|
-| `smoke.yaml` | CI / quick sanity (`size_256`, n=1, SPSC) |
-| `default.yaml` | Publication matrix (256 B and 4 KiB × [100, 1000], SPSC+MPMC) |
+| `smoke.yaml` | CI / quick sanity (`size_256`, n=1, 1P1C) |
+| `default.yaml` | Publication matrix (256 B and 4 KiB × [100, 1000], 1P1C and 4P4C) |
 
 ## Usage
 

config/library/default.yaml

@@ -1,6 +1,6 @@
 # Run config: default (publication matrix)
 id: default
-description: "Two payload sizes (256 B, 4 KiB) × [100, 1000] items; SPSC, 2P2C, 1P4C, 4P1C, 4P4C"
+description: "Two payload sizes (256 B, 4 KiB) × [100, 1000] items; 1P1C and 4P4C"
 data_model_version: 2
 
 types:
@@ -16,8 +16,9 @@ compression:
 
 execution:
   mode: full
-  # bytes = SPSC; stream = MPMC (CSV ABI name; not a stream API)
-  io_modes: [bytes, stream, 1p4c, 4p1c, 4p4c]
+  # bytes = 1P1C (CSV leftover name). 4p4c = four producers and four consumers.
+  # 2P2C / 1P4C / 4P1C stay on experiment 3, not the published matrix.
+  io_modes: [bytes, 4p4c]
 
 budget:
   soft_seconds_per_10_serializers: 60

config/library/smoke.yaml

@@ -1,6 +1,6 @@
 # Run config: smoke (CI / fast path)
 id: smoke
-description: "Minimal matrix — 256-byte payload, single instance, SPSC only"
+description: "Minimal matrix — 256-byte payload, single instance, 1P1C only"
 data_model_version: 2
 
 types:

dashboard/index.html

@@ -105,9 +105,9 @@
       </div>
 
       <div class="filter-group filter-group-secondary" aria-label="Secondary filters">
-        <span class="filter-label" title="Concurrency pattern: SPSC (1 producer / 1 consumer) vs MPMC (2 / 2). CSV still logs bytes/stream.">Pattern:</span>
+        <span class="filter-label" title="How many producers and consumers share the queue. Published cells are 1P1C and 4P4C.">Pattern:</span>
         <div class="select-wrapper">
-          <select id="mode-select" aria-label="Select pattern (SPSC or MPMC)">
+          <select id="mode-select" aria-label="Select pattern (1P1C or 4P4C)">
             <!-- Dynamically populated -->
           </select>
         </div>
@@ -156,7 +156,7 @@
         </p>
         <p class="orientation-gloss section-help">
           <strong>Pareto</strong> = not both slower and more CPU-hungry ·
-          <strong>Pattern</strong> = SPSC vs MPMC ·
+          <strong>Pattern</strong> = 1P1C or 4P4C ·
           <strong>Samples</strong> = which trials after warmup (default drops outliers) ·
           <a href="../analysis/COMPARISON_RULES/" id="orientation-rules-link">Comparison rules</a> ·
           <a href="../analysis/" id="orientation-method-link">Benchmarks</a> ·
@@ -302,7 +302,7 @@ <h2 class="chart-title compare-h2">Compare</h2>
             <div class="select-wrapper">
               <select id="same-data-select" aria-label="Same-language payload size"></select>
             </div>
-            <span class="filter-label" title="SPSC vs MPMC pattern">Pattern</span>
+            <span class="filter-label" title="How many producers and consumers share the queue">Pattern</span>
             <div class="select-wrapper">
               <select id="same-mode-select" aria-label="Same-language mode"></select>
             </div>
@@ -317,7 +317,7 @@ <h2 class="chart-title compare-h2">Compare</h2>
             <div class="select-wrapper">
               <select id="xl-data-select" aria-label="Cross-language payload size"></select>
             </div>
-            <span class="filter-label" title="SPSC vs MPMC pattern">Pattern</span>
+            <span class="filter-label" title="How many producers and consumers share the queue">Pattern</span>
             <div class="select-wrapper">
               <select id="xl-mode-select" aria-label="Cross-language mode"></select>
             </div>

dashboard/main.js

@@ -1687,9 +1687,9 @@ function applyFilterPolicyToAllGroups({ refreshSelectors = false } = {}) {
     if (modeOptions.includes(wantMode)) {
       state.currentMode = wantMode;
     } else {
-      // Prefer SPSC (bytes) when nothing is selected yet. Insertion order in
-      // the stats file often starts at 1p4c, which hid the usual first view.
-      const prefer = ['bytes', 'stream', '1p4c', '4p1c', '4p4c'];
+      // Prefer 1P1C when nothing is selected yet. Insertion order in the
+      // stats file often starts at 1p4c, which hid the usual first view.
+      const prefer = ['bytes', '4p4c', 'stream', '1p4c', '4p1c'];
       state.currentMode =
         prefer.find((m) => modeOptions.includes(m)) || modeOptions[0] || '';
     }
@@ -2545,8 +2545,8 @@ function normalizeMode(mode) {
 }
 
 function modeDisplayLabel(norm) {
-  if (norm === 'bytes' || norm === 'spsc') return 'SPSC';
-  if (norm === 'stream' || norm === 'mpmc') return 'MPMC (2P2C)';
+  if (norm === 'bytes' || norm === 'spsc' || norm === '1p1c') return '1P1C';
+  if (norm === 'stream' || norm === 'mpmc' || norm === '2p2c') return '2P2C';
   if (/^\d+p\d+c$/.test(String(norm || ''))) return String(norm).toUpperCase();
   return norm || '—';
 }
@@ -2589,7 +2589,7 @@ function summarizeStreamHonesty(groups) {
 function updateStreamHonestyChip() {
   const chip = document.getElementById('stream-honesty-chip');
   if (!chip) return;
-  // Pattern=stream means MPMC, not stream I/O. Do not show serializer honesty.
+  // Pattern=stream means 2P2C, not stream I/O. Do not show serializer honesty.
   chip.hidden = true;
   chip.textContent = '';
   chip.removeAttribute('title');

dashboard/public/data/available_runs.json

@@ -1,5 +1,6 @@
 {
   "csharp": [
+    "2026-09-01-162810",
     "2026-09-01-153914",
     "2026-09-01-143923",
     "2026-09-01-131724",
@@ -17,6 +18,7 @@
     "2026-08-31-141025"
   ],
   "rust": [
+    "2026-09-01-162819",
     "2026-09-01-153934",
     "2026-09-01-144004",
     "2026-09-01-131801",
@@ -35,6 +37,7 @@
     "2026-08-31-141008"
   ],
   "python": [
+    "2026-09-01-162815",
     "2026-09-01-153922",
     "2026-09-01-143936",
     "2026-09-01-131735",
@@ -54,6 +57,7 @@
     "2026-08-31-140959"
   ],
   "javascript": [
+    "2026-09-01-162823",
     "2026-09-01-153947",
     "2026-09-01-144029",
     "2026-09-01-131823",
@@ -72,6 +76,7 @@
     "2026-08-31-141008"
   ],
   "c": [
+    "2026-09-01-162822",
     "2026-09-01-153943",
     "2026-09-01-144022",
     "2026-09-01-131817",

docs/analysis/BENCHMARK_DESIGN.md

@@ -34,13 +34,13 @@ priority, blocking vs spin vs yield, SPSC vs MPMC.
 
 | CSV value | Say this | What actually ran |
 |-----------|----------|-------------------|
-| `bytes` | **SPSC** | One producer, one consumer |
-| `stream` | **MPMC** | Two producers, two consumers |
-| `1p4c` / `4p1c` / `4p4c` | Named P×C | Default full matrix + experiment 3 |
+| `bytes` | **1P1C** | One producer, one consumer |
+| `stream` | **2P2C** | Two producers, two consumers (experiment / old logs) |
+| `1p4c` / `4p1c` / `4p4c` | **1P4C** / **4P1C** / **4P4C** | Experiment 3; 4P4C is also on the default matrix |
 
 There is no stream I/O in this suite. In-process queues move already-built
-payloads. If a library cannot do MPMC, **skip the cell** — do not fake it
-with a mutex around an SPSC structure.
+payloads. If a library cannot run with more than one producer or consumer,
+**skip the cell** — do not fake it with a mutex around a 1P1C structure.
 
 ## What we time
 
@@ -75,7 +75,7 @@ Same run modes as [Modes](modes.md): smoke / all-single / full / research.
 |------|-----|
 | Languages | C, C#, JavaScript, Python, Rust |
 | Categories | T and A on the default matrix; P/S/D opt-in (see [categories](queue_categories.md)) |
-| Pattern | SPSC, 2P2C, 1P4C, 4P1C, 4P4C (skip if the library cannot) |
+| Pattern | 1P1C and 4P4C (skip 4P4C if the library cannot) |
 | Payloads | 256 B (`size_256`) and 4 KiB (`size_4096`) |
 | Experiments | [01](../experiments/01-spsc-handoff/)–[12](../experiments/12-durable-local/) |
 
@@ -94,7 +94,7 @@ Same run modes as [Modes](modes.md): smoke / all-single / full / research.
 
 ### Category A
 
-Same SPSC/MPMC cells as T, but the workers are async tasks. Extra tests:
+Same 1P1C / 4P4C cells as T, but the workers are async tasks. Extra tests:
 experiment 7 (many waiters), 8 (bounded async), 9 (cancel). Do not rank an
 A library against a T library.
 
@@ -125,7 +125,7 @@ is multi-socket and the run says so.
 
 1. Pick a **language**.
 2. Pick a **category** (T or A) in the dashboard Category control.
-3. Pick a **pattern** (SPSC or MPMC) and a **payload**.
+3. Pick a **pattern** (1P1C or 4P4C) and a **payload size**.
 4. Compare families inside that slice.
 
 There is no overall score. A queue that wins T1 can lose T2 or T4.

docs/analysis/COMPARISON_RULES.md

@@ -5,8 +5,8 @@ a chart that looks decisive and is wrong.
 
 ## Never
 
-1. **Never publish one “fastest queue” number.** A library that wins SPSC
-   handoff of a 256-byte ticket can lose at 4 KiB, under 4 producers, or
+1. **Never publish one “fastest queue” number.** A library that wins 1P1C
+   handoff of a 256-byte ticket can lose at 4 KiB, under 4P4C, or
    when the queue is bounded. Report by category × pattern × payload.
 
 2. **Never rank across languages.** A C ring and a Python deque are not
@@ -25,14 +25,14 @@ a chart that looks decisive and is wrong.
    concurrency. It does not hand a payload from a producer to a consumer.
    It lives under Other, not A.
 
-6. **Never invent MPMC.** If the library cannot run two producers and two
-   consumers, skip the cell. Do not wrap an SPSC structure in a mutex and
-   log it as MPMC.
+6. **Never invent a multi-worker cell.** If the library cannot run four
+   producers and four consumers, skip 4P4C. Do not wrap a 1P1C structure
+   in a mutex and log it as 4P4C.
 
 ## Always
 
 - Compare **inside one language and one communication category**.
-- Say **SPSC / MPMC** in prose. The CSV still says `bytes` / `stream`.
+- Say **1P1C / 4P4C** in prose. The CSV still says `bytes` / `4p4c`.
 - A failed fidelity check is an error, not a speed win.
 - Warmup index 0 stays in the raw CSV; analysis drops it.
 

docs/analysis/METRICS.md

@@ -24,7 +24,7 @@ Do **not** invent `msgs_per_cpu_sec` from wall time.
 | `Size` / `SizeGzip` / `SizeZstd` / `median_size_bytes` | Serializer leftovers. Payload bytes are the **data type**, not a result. | **Removed.** New CSVs and stats do not write these. Old logs may still contain `Size`; analysis ignores it. |
 | `mean_fidelity` | `FidelityScore` = fraction of items that arrived in order. | **Gate**, not a score. If `< 1`, the row is invalid for ranking. Typical value is `1.0`. |
 | `mean_memory_peak_bytes` | Process RSS (`getrusage` / `PeakWorkingSet64` / `process.memoryUsage().rss`). | Provenance. Usually the process, not the queue. Do not run a “most compact” contest on it. |
-| `StreamMode` / honesty | Serializer “native vs adapted stream I/O”. There is no stream I/O in this suite. `Pattern=stream` means **MPMC**. | Do not show an honesty column. |
+| `StreamMode` / honesty | Serializer “native vs adapted stream I/O”. There is no stream I/O in this suite. `Pattern=stream` means **2P2C**. | Do not show an honesty column. |
 | `OpPerSecEnq` / `Deq` / `Handoff` in the CSV | Convenience copies of `1e9 / Time*`. | Analysis recomputes from times. Do not treat CSV ops as a second clock. |
 
 ## Identity (not metrics)
@@ -33,8 +33,8 @@ Do **not** invent `msgs_per_cpu_sec` from wall time.
 `DataTypeInstanceCount`, `TypeConfigHash`, `Repetitions`, `RepetitionIndex`,
 `RunOrder`, `SchedulePosition`, `NativeKind` (implementation family label).
 
-`Pattern`: `bytes` = SPSC, `stream` = MPMC 2P2C, plus named `1p4c` / `4p1c` /
-`4p4c`.
+`Pattern`: `bytes` = 1P1C, `stream` = 2P2C (old logs / experiment 3), plus
+named `1p4c` / `4p1c` / `4p4c`. The published matrix is 1P1C and 4P4C.
 
 ## Derived stats (not CSV columns)
 

docs/analysis/architecture.md

@@ -71,14 +71,15 @@ serializer-benchmark columns so historical logs keep loading.
 | `TimeEnq` | Enqueue ns |
 | `TimeDeq` | Dequeue ns |
 | `TimeHandoff` | Handoff ns |
-| `Pattern` | `bytes` = **SPSC**, `stream` = **MPMC** (not I/O) |
+| `Pattern` | `bytes` = **1P1C**, `4p4c` = **4P4C** (`stream` is leftover 2P2C, not I/O) |
 | `CpuTimeNs` | Process CPU time (spin vs block) |
 
 ## Patterns
 
-- **SPSC** (CSV `bytes`): one producer, one consumer. Every library must implement this.
-- **MPMC** (CSV `stream`): two producers, two consumers. Libraries that cannot do MPMC
-  skip the cell rather than fake it.
+- **1P1C** (CSV `bytes`): one producer, one consumer. Every library must implement this.
+- **4P4C** (CSV `4p4c`): four producers, four consumers. Libraries that cannot
+  run that cell skip it rather than fake it.
 
-The word *stream* is a leftover ABI name. The dashboard and docs say SPSC / MPMC.
+The word *stream* is a leftover ABI name for 2P2C. The dashboard and docs say
+1P1C / 4P4C.
 See [Benchmark design](BENCHMARK_DESIGN.md).

docs/analysis/modes.md

@@ -5,7 +5,7 @@ hard-code repetition counts.
 
 | Mode | Reps | Matrix | Use |
 |------|------|--------|-----|
-| `smoke` | 2 | `config/library/smoke.yaml` (256 B, n=1, SPSC) | CI, “does it run?” |
+| `smoke` | 2 | `config/library/smoke.yaml` (256 B, n=1, 1P1C) | CI, “does it run?” |
 | `all-single` | 10 | default matrix | Quick full pass |
 | `full` | 100 | default matrix | Publication |
 | `research` | 500 | default matrix | High-power stats |

docs/analysis/queue_categories.md

@@ -61,13 +61,15 @@ These are **properties** that can apply inside T or A:
 - bounded vs unbounded (backpressure)
 - FIFO vs priority
 - blocking vs spin vs yield
-- SPSC vs MPMC (workload / pattern)
+- 1P1C vs 4P4C (workload / pattern)
 
 ## Patterns
 
 | Say | CSV `Pattern` | Work |
 |-----|---------------|------|
-| SPSC | `bytes` | 1 producer, 1 consumer |
-| MPMC | `stream` | 2 producers, 2 consumers |
+| 1P1C | `bytes` | 1 producer, 1 consumer |
+| 4P4C | `4p4c` | 4 producers, 4 consumers |
+| 2P2C | `stream` | 2 producers, 2 consumers (old logs / experiment 3) |
+| 1P4C / 4P1C | `1p4c` / `4p1c` | Experiment 3 only |
 
-Logged values `bytes` / `stream` are pattern tags, not I/O.
+Logged values `bytes` / `stream` / `4p4c` are pattern tags, not I/O.

docs/c/index.md

@@ -11,7 +11,7 @@
 |----------|----------|---------------|---------|-------|
 | `mutex-queue` | locked | T (thread) | harness | pthread mutex + ring |
 | `lfqueue` | concurrent | T (thread) | lfqueue (vendored) | Lock-free MPMC FIFO |
-| `spsc-ring` | spsc | T (thread) | harness | Single-producer ring; MPMC skipped |
+| `spsc-ring` | spsc | T (thread) | harness | Single-producer ring; 1P1C only |
 | `steal-deque` | work-stealing | T (thread) | harness | Owner-push / steal-from-top |
 | `pipe-ipc` | concurrent | P (process) | harness | Opt-in pipe + fork |
 | `shared-ring` | spsc | S (shared) | harness | Opt-in mmap + fork ring |

docs/experiments/index.md

@@ -4,8 +4,8 @@ One question per folder under `experiments/`. Read [PLAN.md](../../experiments/P
 
 | # | Question |
 |---|----------|
-| [01-spsc-handoff](../../experiments/01-spsc-handoff/) | SPSC handoff of a small message |
-| [02-payload-size](../../experiments/02-payload-size/) | Same SPSC at 4 KiB |
+| [01-spsc-handoff](../../experiments/01-spsc-handoff/) | 1P1C handoff of a small message |
+| [02-payload-size](../../experiments/02-payload-size/) | Same 1P1C at 4 KiB |
 | [03-contention](../../experiments/03-contention/) | 1P4C / 4P1C / 4P4C |
 | [04-backpressure](../../experiments/04-backpressure/) | Bounded queue, slow consumer |
 | [05-wakeup](../../experiments/05-wakeup/) | Empty-queue wake |