[codex] Add CollectiveX benchmark dashboard

.github/collectivex-source-run.env

@@ -0,0 +1,5 @@
+source_run_id=28534841204
+source_sha=8cbd7c87d2b52f223b2e891b53dbbb4c65d76a83
+source_workflow=CollectiveX Sweep
+source_run_url=https://github.com/SemiAnalysisAI/InferenceX/actions/runs/28534841204
+triggered_at=2026-07-01T19:15:20Z

.github/workflows/update-collectivex-data.yml

@@ -0,0 +1,83 @@
+name: Update CollectiveX Data
+
+on:
+  workflow_dispatch:
+    inputs:
+      source_run_id:
+        description: Optional CollectiveX workflow run ID to include before it reaches completed status
+        type: string
+        required: false
+        default: ''
+  repository_dispatch:
+    types: [update-collectivex-data]
+
+concurrency:
+  group: update-collectivex-data
+  cancel-in-progress: false
+
+permissions:
+  contents: write
+
+jobs:
+  update:
+    timeout-minutes: 30
+    runs-on: ubuntu-latest
+    env:
+      DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
+      SOURCE_RUN_ID: ${{ github.event.client_payload.source_run_id || inputs.source_run_id || '' }}
+    steps:
+      - name: Checkout app repository
+        uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
+        with:
+          ref: ${{ github.event.repository.default_branch }}
+          token: ${{ secrets.PAT }}
+          fetch-depth: 0
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@0e279bb959325dab635dd2c09392533439d90093 # v6.0.8
+
+      - name: Setup Node.js
+        uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
+        with:
+          node-version: '24'
+          cache: pnpm
+
+      - name: Install app dependencies
+        run: pnpm install --frozen-lockfile --filter @semianalysisai/inferencex-app...
+        env:
+          CYPRESS_INSTALL_BINARY: '0'
+
+      - name: Generate static CollectiveX snapshot
+        env:
+          GITHUB_TOKEN: ${{ secrets.PAT }}
+        run: |
+          set -euo pipefail
+          if [ -n "$SOURCE_RUN_ID" ]; then
+            pnpm --filter @semianalysisai/inferencex-app generate:collectivex -- \
+              --source-run-id "$SOURCE_RUN_ID"
+          else
+            pnpm --filter @semianalysisai/inferencex-app generate:collectivex
+          fi
+
+      - name: Commit snapshot when data changed
+        run: |
+          set -euo pipefail
+          SNAPSHOT=packages/app/public/data/collectivex.json
+          git add "$SNAPSHOT"
+          if git diff --cached --quiet; then
+            echo "CollectiveX snapshot is already current." >> "$GITHUB_STEP_SUMMARY"
+            exit 0
+          fi
+
+          git config user.name "InferenceX Data Bot"
+          git config user.email "actions@users.noreply.github.com"
+          git commit -m "chore: update CollectiveX data"
+          git pull --rebase origin "$DEFAULT_BRANCH"
+          git push origin "HEAD:$DEFAULT_BRANCH"
+
+          {
+            echo "Updated \`$SNAPSHOT\`."
+            if [ -n "$SOURCE_RUN_ID" ]; then
+              echo "Included source run: \`$SOURCE_RUN_ID\`."
+            fi
+          } >> "$GITHUB_STEP_SUMMARY"

.gitignore

@@ -52,6 +52,7 @@
 **/public/data/*
 !**/public/data/github-stars.json
 !**/public/data/evaluation/dummy_eval_data.json
+!**/public/data/collectivex.json
 
 # cypress
 **/cypress/videos/

packages/app/cypress/component/tab-nav.cy.tsx

@@ -70,6 +70,11 @@ describe('TabNav — unofficialrun URL preservation (issue #319)', () => {
       'href',
       '/submissions?unofficialruns=12345',
     );
+    cy.get('[data-testid="tab-trigger-collectivex"]').should(
+      'have.attr',
+      'href',
+      '/collectivex?unofficialruns=12345',
+    );
     cy.get('[data-testid="tab-trigger-historical"]').should(
       'have.attr',
       'href',
@@ -109,6 +114,7 @@ describe('TabNav — Hidden popover for gated tabs', () => {
     mountTabNav({});
     cy.get('[data-testid="tab-trigger-inference"]').should('exist');
     cy.get('[data-testid="tab-trigger-gpu-specs"]').should('exist');
+    cy.get('[data-testid="tab-trigger-collectivex"]').should('exist');
     cy.get('[data-testid="tab-trigger-submissions"]').should('exist');
     cy.get('[data-testid="tab-trigger-hidden"]').should('not.exist');
     cy.get('[data-testid="tab-trigger-feedback"]').should('not.exist');

packages/app/cypress/e2e/collectivex.cy.ts

@@ -0,0 +1,187 @@
+function expectToggleOptions(testId: string, labels: string[]) {
+  cy.get(`[data-testid="${testId}"]`)
+    .find('button')
+    .then(($buttons) => {
+      expect($buttons.toArray().map((button) => button.textContent?.trim())).to.deep.equal(labels);
+    });
+}
+
+function xTickLabels() {
+  return cy.get('[data-testid="collectivex-explorer-chart"] .x-axis .tick text').then(($ticks) =>
+    $ticks
+      .toArray()
+      .map((tick) => tick.textContent?.trim())
+      .filter(Boolean),
+  );
+}
+
+describe('CollectiveX', () => {
+  beforeEach(() => {
+    cy.intercept('GET', '/data/collectivex.json', { fixture: 'api/collectivex.json' }).as(
+      'collectivexData',
+    );
+    cy.visit('/collectivex');
+    cy.wait('@collectivexData');
+  });
+
+  it('renders the v3 report hierarchy with headline defaults', () => {
+    cy.get('[data-testid="collectivex-display"]')
+      .should('contain.text', 'CollectiveX')
+      .and('contain.text', 'Retained sweeps');
+
+    cy.contains('[role="tab"]', 'EP dispatch / combine').should(
+      'have.attr',
+      'aria-selected',
+      'true',
+    );
+    cy.contains('[role="tab"]', 'Decision').should('be.visible');
+    cy.contains('[role="tab"]', 'All-reduce').should('be.visible');
+    cy.contains('[role="tab"]', 'KV-cache transfer').should('be.visible');
+
+    cy.get('[data-testid="collectivex-main-chart"]')
+      .should('contain.text', 'Round trip (measured) · decode · p99')
+      .and('contain.text', 'DeepSeek-V3/V4')
+      .and('contain.text', 'BF16')
+      .and('contain.text', 'EP8');
+    cy.get('[data-testid="collectivex-explorer-chart"] svg').should('be.visible');
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should('have.length', 3);
+    cy.get('[data-testid="collectivex-explorer-chart"] .point').should('have.length', 20);
+    xTickLabels().should('include.members', ['1', '8', '128']);
+  });
+
+  it('keeps log axes sparse and byte/tokens axes readable', () => {
+    cy.get('[data-testid="collectivex-explorer-chart"] .y-axis .tick').should(($ticks) => {
+      expect($ticks.length).to.be.greaterThan(0);
+      expect($ticks.length).to.be.at.most(5);
+    });
+
+    cy.contains('[role="tab"]', 'All-reduce').click();
+    cy.get('[data-testid="collectivex-all-reduce-chart"] svg').should('be.visible');
+    cy.get('[data-testid="collectivex-all-reduce-chart"] .x-axis .tick').should(($ticks) => {
+      expect($ticks.length).to.be.at.most(10);
+    });
+  });
+
+  it('exposes model, backend, precision, activation, EP, publication, and axis controls', () => {
+    expectToggleOptions('collectivex-operation-toggle', [
+      'Dispatch',
+      'Combine',
+      'Round trip',
+      'Isolated sum',
+    ]);
+    expectToggleOptions('collectivex-phase-toggle', ['Decode', 'Prefill']);
+    expectToggleOptions('collectivex-percentile-toggle', ['p50', 'p90', 'p99']);
+    expectToggleOptions('collectivex-suite-toggle', [
+      'All',
+      'Backend default',
+      'Resource constrained',
+    ]);
+    expectToggleOptions('collectivex-publication-toggle', [
+      'Official headline',
+      'Publishable',
+      'Official only',
+      'All',
+    ]);
+    expectToggleOptions('collectivex-precision-toggle', ['All', 'BF16', 'FP8']);
+    expectToggleOptions('collectivex-ep-toggle', ['All', 'EP8']);
+    expectToggleOptions('collectivex-x-scale-toggle', ['Log', 'Linear']);
+    expectToggleOptions('collectivex-y-scale-toggle', ['Log', 'Linear']);
+
+    cy.get('[data-testid="collectivex-model-shape-select"]').should(
+      'contain.text',
+      'DeepSeek-V3/V4',
+    );
+    cy.get('[data-testid="collectivex-backend-select"]').click();
+    cy.get('[role="option"]').then(($options) => {
+      expect($options.toArray().map((option) => option.textContent?.trim())).to.include.members([
+        'All',
+        'deepep',
+        'deepep v2',
+      ]);
+    });
+    cy.contains('[role="option"]', 'All').click();
+    cy.get('[data-testid="collectivex-activation-select"]').click();
+    cy.get('[role="option"]').then(($options) => {
+      expect($options.toArray().map((option) => option.textContent?.trim())).to.include.members([
+        'All',
+        'Normal',
+      ]);
+    });
+    cy.contains('[role="option"]', 'Normal').click();
+    cy.get('[data-testid="collectivex-activation-select"]').should('contain.text', 'Normal');
+  });
+
+  it('filters the EP explorer by precision, activation, routing, and phase', () => {
+    cy.get('[data-testid="collectivex-suite-toggle"]')
+      .contains('button', 'Backend default')
+      .click();
+    cy.get('[data-testid="collectivex-precision-toggle"]').contains('button', 'FP8').click();
+    cy.get('[data-testid="collectivex-main-chart"]')
+      .should('contain.text', 'FP8')
+      .and('contain.text', 'B300 EP8 · deepep v2 · fp8');
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should('have.length', 1);
+
+    cy.get('[data-testid="collectivex-suite-toggle"]').contains('button', 'All').click();
+    cy.get('[data-testid="collectivex-precision-toggle"]').contains('button', 'BF16').click();
+    cy.get('[data-testid="collectivex-routing-select"]').click();
+    cy.contains('[role="option"]', 'zipf').click();
+    cy.get('[data-testid="collectivex-main-chart"]').should('contain.text', 'zipf');
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should(
+      'have.length.at.least',
+      1,
+    );
+
+    cy.get('[data-testid="collectivex-routing-select"]').click();
+    cy.contains('[role="option"]', 'uniform').click();
+    cy.get('[data-testid="collectivex-phase-toggle"]').contains('button', 'Prefill').click();
+    cy.get('[data-testid="collectivex-main-chart"]').should('contain.text', 'prefill');
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should('have.length', 2);
+    xTickLabels().should('include.members', ['128', '256']);
+  });
+
+  it('renders Decision and Evidence tabs from the static snapshot', () => {
+    cy.contains('[role="tab"]', 'Decision').click();
+    cy.get('[data-testid="collectivex-decision"]')
+      .should('contain.text', 'Best backend')
+      .and('contain.text', 'Max tokens under round-trip p99 budget');
+    cy.get('[data-testid="collectivex-summary-card"]').should('have.length', 7);
+    cy.get('[data-testid="collectivex-budget-table"]').should('contain.text', '<= 100 us');
+
+    cy.contains('[role="tab"]', 'Evidence').click();
+    cy.get('[data-testid="collectivex-sensitivity-table"]').should('exist');
+    cy.get('[data-testid="collectivex-failures-table"]').should('contain.text', 'diagnostic');
+    cy.get('[data-testid="collectivex-coverage-table"]').should('contain.text', 'official');
+    cy.get('section[data-testid="collectivex-display"]').should('contain.text', 'Provenance');
+  });
+
+  it('renders all new collective and transfer family tabs', () => {
+    const familyTabs = [
+      ['All-reduce', 'collectivex-all-reduce-chart', 'collectivex-all-reduce-metric-toggle'],
+      ['All-gather', 'collectivex-all-gather-chart', 'collectivex-all-gather-metric-toggle'],
+      ['CPU-GPU offload', 'collectivex-offload-chart', 'collectivex-offload-metric-toggle'],
+      ['KV-cache transfer', 'collectivex-kv-cache-chart', 'collectivex-kv-cache-metric-toggle'],
+      [
+        'Copy-engine / SDMA',
+        'collectivex-copy-engine-chart',
+        'collectivex-copy-engine-metric-toggle',
+      ],
+      ['RL mesh', 'collectivex-rl-mesh-chart', 'collectivex-rl-mesh-metric-toggle'],
+    ];
+
+    for (const [label, chartTestId, metricTestId] of familyTabs) {
+      cy.contains('[role="tab"]', label).click();
+      cy.get(`[data-testid="${metricTestId}"]`).should('be.visible');
+      cy.get(`[data-testid="${chartTestId}"] svg`).should('be.visible');
+      cy.get(`[data-testid="${chartTestId}"] .line-path`).should('have.length.at.least', 1);
+    }
+  });
+
+  it('legend toggles remove and restore a rendered EP series', () => {
+    cy.get('[data-testid="collectivex-main-chart"]').within(() => {
+      cy.contains('label', 'B300 EP8 · deepep · bf16').click();
+    });
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should('have.length', 2);
+    cy.get('[data-testid="collectivex-main-chart"]').contains('button', 'Reset filter').click();
+    cy.get('[data-testid="collectivex-explorer-chart"] .line-path').should('have.length', 3);
+  });
+});