AutheOSI / PQC-gossip AQVM capacity lab
OSI Runtime Gossip Lab

Reviewer-facing capacity view for the real request -> PQC-gossip -> QSTP/routing -> AQVM -> receipt/evidence path. It separates the questions reviewers ask most often: how many AQVM sessions ran in parallel, how slow each session felt, how many packaged runs completed per second, and whether proof/receipt validation still held under load. Browser runs are dispatch samples; imported backend load-runner JSON is AQVM capacity evidence.

AQVM capacity profile AWRE Trinity PQC-gossip path validation under load failure threshold exploration
A. What is being tested

Current AQVM capacity profile over PQC-gossip

DEGRADATION: FAILED

This lab calls the packaged command-center API with --aqvm ./wrapped/autheo_osi_runtime.aqvm. Single-run correctness remains supported; browser capacity scenarios dispatch multiple packaged runs and summarize end-to-end request timing plus AQVM evidence. Import fixed or ramp backend load-runner JSON when you need capacity evidence beyond the browser safety cap.

client/request -> PQC-gossip -> QSTP / PQCNet routing -> optional PrivacyNet posture -> AQVM -> receipt/evidence -> ZK catalog -> auditor export
P95 end-to-end latency
n/a
how slow one request feelsUse latency, especially p95, to judge whether AQVM feels slow under the selected load.
Completed AQVM runs/sec
n/a
batch completion ratePackaged AQVM runtime sessions completed per wall-clock second; this is not chain transaction TPS.
AQVM parallel sessions
0 requested / 0 peak
requested / observed peakThis answers how many AQVM sessions ran at once. Browser dispatch cap: 16; this cap is not AQVM capacity.
OSI1 shard throughput
n/a
derived shards/secObserved OSI1 shard work per second across the batch; shards are workload units, not sessions.
Success rate
0%
verified runsPercent of completed sessions with AQVM, receipt, evasion, ZK catalog, and non-simulated evidence intact.
Failure rate
0%
failed/timeout runsFailed or timed-out AQVM sessions. Capacity is only useful while this stays low.
Current test mode / load
not run
selected scenarioShows test mode, requested parallelism, and total batch size for the current sample.
Parallel AQVM capacitySet Concurrent AQVM sessions, then compare requested parallelism with max concurrent observed.
Slowdown signalUse p95 and average end-to-end latency to judge how slow a session feels.
Throughput unitsRuns/sec counts completed packaged AQVM runs; shards/sec counts observed OSI1 shard work. Neither is EVM/ETH transaction TPS.
B. Current load scenario

Reviewer load knobs and presets

EVIDENCE EXPORT: READY

Choose a capacity preset or adjust the reviewer knobs. Concurrency and run count are live in this UI: the page dispatches browser samples of packaged AQVM runtime requests in parallel and measures the full request path. Backend load-runner imports show AQVM capacity evidence separately.

Browser dispatch sample knob: UI dispatches up to 16 packaged AQVM runs at once; this is not AQVM capacity.
Total AQVM sessions in the sample; this is not the parallel-session count.
Sent as testMode and mapped to --test.
Sent as evasionProfile and mapped to --evasion.
Backend generates the runtime message; observed shard count comes from evidence.
Used by this page to interpret the returned latency.
Abort threshold for each UI-dispatched runtime request.
Native runner reports observed shards; exact target is metadata for now.
Local reviewer gate; native run still reports actual receipt status.
Controls reviewer posture interpretation without mutating evidence.
Load single-run, AQVM load/ramp, or osi-runtime-gossip-cluster-ramp JSON evidence.
Advanced operator controls
Leave blank for same-origin packaged command center.
Raw .wasm paths are rejected by the API and package wrapper.
Sent to the runner as --label.
Maps to the existing CLI --fault option.
Use paranoia evasion for live decoy behavior; exact target is informational.
Visible for upcoming runtime wiring; current API runs one packaged CLI invocation.
Current API always returns JSON plus Prometheus preview.
Mode, payload, evasion, fault, concurrency, batch, timeout
live request
Latency, receipt, proof strictness, saturation posture
local reviewer gate
Shard target, decoy target, dispatch/export mode
upcoming wiring

Package reviewers can run ./osi-runtime-gossip-test --test full-suite --export-json ./evidence/osi-runtime-gossip-evidence.json --export-prometheus ./evidence/osi-runtime-gossip.prom from the extracted super-node-package, then import the JSON here. For backend AQVM capacity, import ./evidence/aqvm-load.json or ./evidence/aqvm-ramp.json from ./osi-runtime-aqvm-load. For distributed quorum evidence, import ./evidence/osi-gossip-cluster-ramp.json from ./osi-runtime-gossip-cluster.

Third-party cluster smoke ramp, add-node helper, and topology JSON

Run this from the extracted super-node-package after each topology node is serving ./run-command-center.sh. To test more super-nodes without hand-editing JSON, use ./cluster/add_osi_gossip_node.mjs; the manual cluster/osi-gossip-topology.json edit remains available for reviewers who want to inspect or patch the topology directly. 

./cluster/add_osi_gossip_node.mjs \
  --name reviewer-node-1 \
  --base-url http://<reviewer-node-host>:8080 \
  --ssh root@<reviewer-node-host> \
  --package-path /root/super-node-package \
  --quorum 2
mkdir -p ./evidence
./osi-runtime-gossip-cluster \
  --topology ./cluster/osi-gossip-topology.json \
  --test full-suite \
  --ramp 1,2,4 \
  --runs-per-level 12 \
  --payload-bytes 4096 \
  --quorum 2 \
  --timeout-ms 120000 \
  --connect-timeout-ms 5000 \
  --fanout-timeout-ms 10000 \
  --evasion strict \
  --export-json ./evidence/osi-gossip-cluster-smoke-ramp.json \
  --export-prometheus ./evidence/osi-gossip-cluster-smoke-ramp.prom

Manual JSON alternative: append an object with a unique name, reachable base_url, and optional aqvm_runner SSH package path. 

{
  "role": "super-node",
  "name": "supernode-c",
  "base_url": "http://<super-node-host>:8080",
  "aqvm_runner": "ssh://root@<super-node-host>/root/super-node-package"
}
D. Degradation / failure posture

Where AQVM breaks down

The posture combines latency thresholds, failure rate, queue backlog, and whether runtime/receipt/ZK evidence continues to verify while the AQVM path is loaded.

FAILED

No verified AQVM runtime evidence has been collected yet.

End-to-end path summary

Request -> PQC-gossip -> AQVM -> evidence

Requestclient payload + reviewer scenario
PQC-gossippackage command-center dispatch
QSTP / routingPQCNet path; PrivacyNet optional posture
AQVMwrapped AutheOSI runtime compute
ReceiptCSI/OSI receipt integrity
ZK catalogreassembly catalog validation
n/aingress / transportderived overhead
n/aAQVM computeruntime evidence
n/areceipt generationderived overhead
n/aevidence export / ZK catalogderived overhead
E. Correctness under current load

Validation Evidence

VALIDATION: MISSING

These proof indicators confirm whether the AQVM run stayed valid while loaded. They support the capacity story instead of replacing it.

Runtime OK
MISSING
Receipt verified
MISSING
Evasion verified
MISSING
ZK catalog verified
MISSING
AQVM artifact
MISSING
Session posture
MISSING

AQVM compute module

FAIL
Theo/AQVM quantum VM compute artifact under validation.

Runtime

FAIL
AutheOSI runtime execution result.

OSI1 shards

FAIL
Shard counts and OSI1 sent counts must agree.

Evasion

FAIL
Object shard evasion verification.

Receipt

FAIL
CSI/OSI receipt integrity check.

ZK catalog

FAIL
Reassembly circuit catalog evidence.

Simulated success

FAIL
Must remain false for a PASS.
Detailed evidence

Runtime, capacity, and proof rows

Auditor package

Third-party JSON export

{"suite":"osi-runtime-gossip","result":"fail","aqvm_required":true,"aqvm_available":false,"simulated_success":false,"error":"not_run"}
Observability bridge

Prometheus preview/export

osi_runtime_gossip_simulated_success_total 0
Reviewer interpretation

Load, latency, throughput, verification outcome

Awaiting AQVM capacity evidence. Target profile: throughput measured, latency below threshold, runtime ok, receipt verified, evasion verified, ZK catalog verified, and simulated success false.

    Scenario history

    Recent scaling / trend comparison

    No runs yetRun or import evidence to compare scenarios.
    Metric provenance

    Real vs derived capacity signals

    Real signals: packaged runtime result, AQVM availability/path, runtime OK, receipt/evasion/ZK verification, payload size, observed shards, AQVM latency, and AQVM shards/ms when returned by evidence. Derived signals: UI-measured end-to-end latency, completed AQVM runs/sec, OSI1 shards/sec, p95, queue depth, saturation/degradation posture, and timing overhead outside AQVM. Browser parallelism is a dispatch sample capped at 16 and is not AQVM capacity; imported backend load-runner fixed/ramp JSON is the AQVM capacity evidence for stable max, first degraded concurrency, and production recommendation. Live controls: mode, payload, evasion, fault, concurrency, batch size, timeout, and AQVM artifact path. Future/metadata controls: exact shard target, exact decoy target, dispatch mode, and export mode.