Stadium relaunch: full 127-page deep-dive report -- per-cell, per-ISA,
every factor interaction, and a raw-data appendix Expanded the campaign-mechanism validation report from a condensed 6-page summary into the full depth Captain Bob asked for: analyze all 9 cells as a conglomerate Latin square, then dive into each cell's own data, then cover every within-ISA and cross-factor interaction explicitly rather than averaging it away. Report structure (127 pages, compiled clean, no undefined references): - Front matter: context, methodology, the SWAP-MTX bug narrative (console-interleaving fix + the Fisher-Yates correctness bug and its fix, both already committed separately) - Layer 1: aggregate 3x3 Latin square (heatmap, invariant-metrics table) - Per-Cell Deep Dive (9 sections): cfg-level distribution, summary table, and a rep-order execution-trajectory chart per cell -- the trajectory charts are what actually visualize the order-dependence finding rather than just stating it - Per-ISA Deep Dive (3 sections): within-architecture seed comparison (violin plots, Kruskal-Wallis, per-factor main effects) - Factor Interactions (6 sections, every pairwise combination of the 4 L8 binary factors): both infer_dec_q and early_exit interaction plots faceted by architecture, plus the three-way factor x factor x architecture significance test - Per-Factor Response (4 sections): linear response by architecture, with an explicit note that a true quadratic term isn't identifiable from this 2-level factorial design - Appendix: full run_id-ordered raw data, all 4,320 rows across all 9 cells, as the primary-source backing for every statistic above Generated programmatically (analyse_stadium_relaunch_fixed.R for the aggregate layer, generate_stadium_deepdive.R for the per-cell/per-ISA/ interaction/appendix layers) rather than hand-authored, since content at this scale needs to be data-driven to stay honest. Also includes analyse_stadium_relaunch.R, the earlier script built against the pre-fix (buggy-shuffle) dataset -- superseded but kept for the record, matching how the underlying data commits were handled. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Sonnet 5
parent
dbe4b671a1
commit
cf5b08fb65
@@ -0,0 +1,180 @@
|
||||
<?xml version='1.0' encoding='UTF-8' ?>
|
||||
<svg xmlns='http://www.w3.org/2000/svg' xmlns:xlink='http://www.w3.org/1999/xlink' width='864.00pt' height='432.00pt' viewBox='0 0 864.00 432.00'>
|
||||
<g class='svglite'>
|
||||
<defs>
|
||||
<style type='text/css'><![CDATA[
|
||||
.svglite line, .svglite polyline, .svglite polygon, .svglite path, .svglite rect, .svglite circle {
|
||||
fill: none;
|
||||
stroke: #000000;
|
||||
stroke-linecap: round;
|
||||
stroke-linejoin: round;
|
||||
stroke-miterlimit: 10.00;
|
||||
}
|
||||
.svglite text {
|
||||
white-space: pre;
|
||||
}
|
||||
.svglite g.glyphgroup path {
|
||||
fill: inherit;
|
||||
stroke: none;
|
||||
}
|
||||
]]></style>
|
||||
</defs>
|
||||
<rect width='100%' height='100%' style='stroke: none; fill: #FFFFFF;'/>
|
||||
<defs>
|
||||
<clipPath id='cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA='>
|
||||
<rect x='0.00' y='0.00' width='864.00' height='432.00' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
<rect x='0.000000000000064' y='0.000000000000057' width='864.00' height='432.00' style='stroke-width: 1.07; stroke: none; fill: #FFFFFF;' />
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpMzIuNzh8MzA0LjM3fDM0LjE0fDM2NS4wMg=='>
|
||||
<rect x='32.78' y='34.14' width='271.59' height='330.89' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpMzIuNzh8MzA0LjM3fDM0LjE0fDM2NS4wMg==)'>
|
||||
<polyline points='32.78,262.63 304.37,262.63 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='32.78,156.03 304.37,156.03 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='32.78,49.44 304.37,49.44 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='106.85,365.02 106.85,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='230.30,365.02 230.30,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='106.85,151.59 230.30,182.09 ' style='stroke-width: 2.13; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='106.85,225.32 230.30,247.09 ' style='stroke-width: 2.13; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<circle cx='106.85' cy='151.59' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='230.30' cy='182.09' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='106.85' cy='225.32' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<circle cx='230.30' cy='247.09' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<polyline points='100.68,49.18 113.03,49.18 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='106.85,49.18 106.85,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='100.68,254.01 113.03,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='224.13,80.66 236.48,80.66 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='230.30,80.66 230.30,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='224.13,283.52 236.48,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='100.68,121.46 113.03,121.46 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='106.85,121.46 106.85,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='100.68,329.18 113.03,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='224.13,144.19 236.48,144.19 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='230.30,144.19 230.30,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='224.13,349.98 236.48,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpMzA5Ljg1fDU4MS40NXwzNC4xNHwzNjUuMDI='>
|
||||
<rect x='309.85' y='34.14' width='271.59' height='330.89' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpMzA5Ljg1fDU4MS40NXwzNC4xNHwzNjUuMDI=)'>
|
||||
<polyline points='309.85,262.63 581.45,262.63 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='309.85,156.03 581.45,156.03 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='309.85,49.44 581.45,49.44 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='383.93,365.02 383.93,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='507.38,365.02 507.38,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='383.93,151.59 507.38,182.09 ' style='stroke-width: 2.13; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='383.93,225.32 507.38,247.09 ' style='stroke-width: 2.13; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<circle cx='383.93' cy='151.59' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='507.38' cy='182.09' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='383.93' cy='225.32' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<circle cx='507.38' cy='247.09' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<polyline points='377.75,49.18 390.10,49.18 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='383.93,49.18 383.93,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='377.75,254.01 390.10,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='501.20,80.66 513.55,80.66 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='507.38,80.66 507.38,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='501.20,283.52 513.55,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='377.75,121.46 390.10,121.46 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='383.93,121.46 383.93,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='377.75,329.18 390.10,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='501.20,144.19 513.55,144.19 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='507.38,144.19 507.38,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='501.20,349.98 513.55,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpNTg2LjkzfDg1OC41MnwzNC4xNHwzNjUuMDI='>
|
||||
<rect x='586.93' y='34.14' width='271.59' height='330.89' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpNTg2LjkzfDg1OC41MnwzNC4xNHwzNjUuMDI=)'>
|
||||
<polyline points='586.93,262.63 858.52,262.63 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='586.93,156.03 858.52,156.03 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='586.93,49.44 858.52,49.44 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='661.00,365.02 661.00,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='784.45,365.02 784.45,34.14 ' style='stroke-width: 1.07; stroke: #E5E5E5; stroke-linecap: butt;' />
|
||||
<polyline points='661.00,151.59 784.45,182.09 ' style='stroke-width: 2.13; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='661.00,225.32 784.45,247.09 ' style='stroke-width: 2.13; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<circle cx='661.00' cy='151.59' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='784.45' cy='182.09' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<circle cx='661.00' cy='225.32' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<circle cx='784.45' cy='247.09' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<polyline points='654.83,49.18 667.17,49.18 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='661.00,49.18 661.00,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='654.83,254.01 667.17,254.01 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='778.28,80.66 790.62,80.66 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='784.45,80.66 784.45,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='778.28,283.52 790.62,283.52 ' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<polyline points='654.83,121.46 667.17,121.46 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='661.00,121.46 661.00,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='654.83,329.18 667.17,329.18 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='778.28,144.19 790.62,144.19 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='784.45,144.19 784.45,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<polyline points='778.28,349.98 790.62,349.98 ' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpMzIuNzh8MzA0LjM3fDI0LjI5fDM0LjE0'>
|
||||
<rect x='32.78' y='24.29' width='271.59' height='9.85' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpMzIuNzh8MzA0LjM3fDI0LjI5fDM0LjE0)'>
|
||||
<text x='168.58' y='32.99' text-anchor='middle' style='font-size: 11.00px; font-weight: bold; font-family: "Liberation Sans";' textLength='34.88px' lengthAdjust='spacingAndGlyphs'>amd64</text>
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpMzA5Ljg1fDU4MS40NXwyNC4yOXwzNC4xNA=='>
|
||||
<rect x='309.85' y='24.29' width='271.59' height='9.85' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpMzA5Ljg1fDU4MS40NXwyNC4yOXwzNC4xNA==)'>
|
||||
<text x='445.65' y='32.99' text-anchor='middle' style='font-size: 11.00px; font-weight: bold; font-family: "Liberation Sans";' textLength='41.62px' lengthAdjust='spacingAndGlyphs'>aarch64</text>
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id='cpNTg2LjkzfDg1OC41MnwyNC4yOXwzNC4xNA=='>
|
||||
<rect x='586.93' y='24.29' width='271.59' height='9.85' />
|
||||
</clipPath>
|
||||
</defs>
|
||||
<g clip-path='url(#cpNTg2LjkzfDg1OC41MnwyNC4yOXwzNC4xNA==)'>
|
||||
<text x='722.72' y='32.99' text-anchor='middle' style='font-size: 11.00px; font-weight: bold; font-family: "Liberation Sans";' textLength='37.97px' lengthAdjust='spacingAndGlyphs'>riscv64</text>
|
||||
</g>
|
||||
<g clip-path='url(#cpMC4wMHw4NjQuMDB8MC4wMHw0MzIuMDA=)'>
|
||||
<text x='106.85' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>hi</text>
|
||||
<text x='230.30' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>lo</text>
|
||||
<text x='383.93' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>hi</text>
|
||||
<text x='507.38' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>lo</text>
|
||||
<text x='661.00' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>hi</text>
|
||||
<text x='784.45' y='376.01' text-anchor='middle' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>lo</text>
|
||||
<text x='27.85' y='265.66' text-anchor='end' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='9.78px' lengthAdjust='spacingAndGlyphs'>42</text>
|
||||
<text x='27.85' y='159.06' text-anchor='end' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='9.78px' lengthAdjust='spacingAndGlyphs'>44</text>
|
||||
<text x='27.85' y='52.46' text-anchor='end' style='font-size: 8.80px;fill: #4D4D4D; font-family: "Liberation Sans";' textLength='9.78px' lengthAdjust='spacingAndGlyphs'>46</text>
|
||||
<text x='445.65' y='388.15' text-anchor='middle' style='font-size: 11.00px; font-family: "Liberation Sans";' textLength='106.44px' lengthAdjust='spacingAndGlyphs'>temporal-decay factor</text>
|
||||
<text transform='translate(13.05,199.58) rotate(-90)' text-anchor='middle' style='font-size: 11.00px; font-family: "Liberation Sans";' textLength='88.12px' lengthAdjust='spacingAndGlyphs'>mean infer_dec_q</text>
|
||||
<text x='380.02' y='417.74' style='font-size: 11.00px; font-family: "Liberation Sans";' textLength='67.31px' lengthAdjust='spacingAndGlyphs'>entropy factor</text>
|
||||
<line x1='454.23' y1='413.95' x2='465.57' y2='413.95' style='stroke-width: 2.13; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<circle cx='459.90' cy='413.95' r='3.02' style='stroke-width: 0.71; stroke: #D62728; fill: #D62728;' />
|
||||
<line x1='454.23' y1='413.95' x2='465.57' y2='413.95' style='stroke-width: 1.07; stroke: #D62728; stroke-linecap: butt;' />
|
||||
<line x1='486.20' y1='413.95' x2='497.54' y2='413.95' style='stroke-width: 2.13; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<circle cx='491.87' cy='413.95' r='3.02' style='stroke-width: 0.71; stroke: #4A90D9; fill: #4A90D9;' />
|
||||
<line x1='486.20' y1='413.95' x2='497.54' y2='413.95' style='stroke-width: 1.07; stroke: #4A90D9; stroke-linecap: butt;' />
|
||||
<text x='472.46' y='416.98' style='font-size: 8.80px; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>hi</text>
|
||||
<text x='504.44' y='416.98' style='font-size: 8.80px; font-family: "Liberation Sans";' textLength='6.84px' lengthAdjust='spacingAndGlyphs'>lo</text>
|
||||
<text x='445.65' y='23.35' text-anchor='middle' style='font-size: 9.00px;fill: #666666; font-family: "Liberation Sans";' textLength='366.84px' lengthAdjust='spacingAndGlyphs'>Non-parallel lines = interaction effect; divergence across panels = ISA-dependent interaction</text>
|
||||
<text x='445.65' y='14.98' text-anchor='middle' style='font-size: 12.00px; font-weight: bold; font-family: "Liberation Sans";' textLength='285.28px' lengthAdjust='spacingAndGlyphs'>ent_in x tmp_in Interaction on infer_dec_q, by ISA</text>
|
||||
</g>
|
||||
</g>
|
||||
</svg>
|
||||
|
After Width: | Height: | Size: 15 KiB |
Reference in New Issue
Block a user