Gauntlet Glaurung 2.2 x64 8 CPU Ryzen 9 \u2014 Chroniques Software<\/title>\n<\/head>\n<body>\n<div class=\"gauntlet-container\" style=\"all: initial; display: block; font-family: 'Segoe UI', Helvetica, Arial, sans-serif; background-color: #0a0a0a; padding: 40px; color: #e0e0e0; line-height: 1.6;\">\n <style>\n .gauntlet-wrapper { max-width: 950px; margin: 0 auto; background: #161616; border: 1px solid #333; border-radius: 12px; overflow: hidden; box-shadow: 0 20px 50px rgba(0,0,0,0.6); }\n .gauntlet-header { background: linear-gradient(135deg, #2c240a 0%, #000000 100%); padding: 30px; text-align: center; border-bottom: 3px solid #D4AF37; }\n .gauntlet-header h1 { color: #D4AF37; margin: 0; font-size: 2em; text-transform: uppercase; letter-spacing: 3px; }\n\n .section-title { color: #D4AF37; border-left: 4px solid #D4AF37; padding-left: 15px; margin: 30px 20px 15px 20px; font-size: 1.4em; text-transform: uppercase; }\n .main-grid { display: grid; grid-template-columns: 1fr 1fr; gap: 20px; padding: 0 20px 20px 20px; }\n\n .engine-card { background: #222; border: 1px solid #444; padding: 15px; border-radius: 8px; position: relative; }\n .engine-card.hero { border: 2px solid #D4AF37; background: #1d1a10; grid-column: span 2; }\n\n .elo-badge { position: absolute; top: 15px; right: 15px; background: #D4AF37; color: #000; padding: 2px 10px; border-radius: 4px; font-weight: bold; font-size: 0.9em; }\n .stat-line { font-size: 0.9em; margin: 5px 0; color: #bbb; }\n .hw-info { font-family: 'Consolas', monospace; font-size: 0.8em; color: #D4AF37; display: block; margin-top: 8px; border-top: 1px solid #333; padding-top: 5px; }\n .highlight { color: #fff; font-weight: bold; }\n\n .bio-section { padding: 0 20px 20px 20px; }\n .bio-card { margin-bottom: 20px; padding-bottom: 15px; border-bottom: 1px solid #333; }\n .bio-name { color: #D4AF37; font-weight: bold; font-size: 1.1em; margin-bottom: 5px; display: block; }\n .footer-note { text-align: center; padding: 20px; font-size: 0.8em; color: #666; border-top: 1px solid #333; }\n\n .ssdf-table { width: 100%; border-collapse: collapse; font-size: 0.82em; margin-top: 10px; }\n .ssdf-table th { background: #2a2a2a; color: #D4AF37; padding: 7px 10px; text-align: left; border-bottom: 2px solid #D4AF37; font-size: 0.85em; letter-spacing: 0.05em; }\n .ssdf-table td { padding: 6px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; }\n .ssdf-table tr:hover td { background: #1e1e1e; }\n .ssdf-table .peak-row td { color: #D4AF37; font-weight: bold; }\n .ssdf-table .ryzen-row td { color: #fff; font-weight: bold; background: #1d1a10; }\n .ssdf-table .ryzen-row td:first-child::before { content: \"\u26a1 \"; }\n\n .timeline { padding: 0; list-style: none; margin: 0; }\n .timeline li { display: flex; gap: 16px; padding: 10px 0; border-bottom: 1px solid #2a2a2a; font-size: 0.87em; }\n .timeline li:last-child { border-bottom: none; }\n .tl-year { min-width: 52px; color: #D4AF37; font-weight: bold; font-family: 'Consolas', monospace; }\n .tl-event { color: #ccc; }\n .tl-event strong { color: #fff; }\n\n .innov-grid { display: grid; grid-template-columns: 1fr 1fr; gap: 12px; margin-top: 10px; }\n .innov-card { background: #1c1c1c; border: 1px solid #3a3a2a; border-radius: 6px; padding: 12px 14px; }\n .innov-card .innov-title { color: #D4AF37; font-weight: bold; font-size: 0.88em; margin-bottom: 5px; }\n .innov-card p { color: #aaa; font-size: 0.82em; margin: 0; line-height: 1.5; }\n\n .quote-box { background: #111; border-left: 3px solid #D4AF37; padding: 14px 18px; margin: 12px 0; border-radius: 0 6px 6px 0; }\n .quote-box p { color: #bbb; font-style: italic; font-size: 0.87em; margin: 0 0 6px 0; line-height: 1.6; }\n .quote-box .quote-src { color: #666; font-size: 0.78em; }\n\n .full-width { grid-column: span 2; }\n .section-inner { padding: 0 20px 20px 20px; }\n\n .recap-table { width: 100%; border-collapse: collapse; font-size: 0.84em; margin-top: 10px; }\n .recap-table th { background: #2a2a2a; color: #D4AF37; padding: 8px 10px; text-align: center; border-bottom: 2px solid #D4AF37; font-size: 0.85em; letter-spacing: 0.05em; }\n .recap-table td { padding: 7px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; text-align: center; }\n .recap-table td:first-child { text-align: left; }\n .recap-table tr:hover td { background: #1e1e1e; }\n .recap-table .total-row td { color: #D4AF37; font-weight: bold; border-top: 2px solid #D4AF37; }\n .pct-bar-wrap { display: flex; align-items: center; gap: 8px; }\n .pct-bar { height: 8px; border-radius: 4px; display: inline-block; }\n .pct-bar.low { background: #e57373; }\n .pct-bar.mid { background: #ffb74d; }\n .pct-bar.high { background: #4caf50; }\n\n .perf-summary { display: grid; grid-template-columns: repeat(4, 1fr); gap: 12px; margin-top: 14px; }\n .perf-box { background: #1c1c1c; border: 1px solid #3a3a2a; border-radius: 8px; padding: 14px; text-align: center; }\n .perf-box.gold-box { border-color: #D4AF37; background: #1d1a10; }\n .perf-box .perf-label { color: #888; font-size: 0.75em; text-transform: uppercase; letter-spacing: 0.07em; margin-bottom: 6px; }\n .perf-box .perf-value { color: #fff; font-size: 1.4em; font-weight: bold; font-family: 'Consolas', monospace; }\n .perf-box.gold-box .perf-value { color: #D4AF37; }\n .perf-box .perf-sub { color: #666; font-size: 0.72em; margin-top: 4px; }\n\n .hw-compare-table { width: 100%; border-collapse: collapse; font-size: 0.84em; margin-top: 10px; }\n .hw-compare-table th { background: #2a2a2a; color: #D4AF37; padding: 8px 10px; text-align: center; border-bottom: 2px solid #D4AF37; }\n .hw-compare-table th:first-child { text-align: left; }\n .hw-compare-table td { padding: 7px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; text-align: center; }\n .hw-compare-table td:first-child { text-align: left; }\n .hw-compare-table tr:hover td { background: #1e1e1e; }\n .hw-compare-table .ryzen-row td { color: #D4AF37; font-weight: bold; background: #1d1a10; }\n .hw-compare-table .ref-row td { color: #666; font-style: italic; }\n .hw-compare-table .gain-pos { color: #4caf50; font-weight: bold; }\n .hw-compare-table .nps-col { font-family: 'Consolas', monospace; font-size: 0.85em; }\n <\/style>\n\n <div class=\"gauntlet-wrapper\">\n\n \n <div class=\"gauntlet-header\">\n <h1>Gauntlet : Glaurung 2.2 x64 (8 Threads) vs The Era<\/h1>\n <p style=\"color: #888; margin-top: 10px;\">Hardware Target : Glaurung 2.2 x64 \u2014 8 threads | AMD Ryzen 9 7950X3D | Pool mixte (75 parties)<\/p>\n <\/div>\n\n \n <div class=\"section-title\">L'Anc\u00eatre de Stockfish<\/div>\n <div class=\"main-grid\">\n <div class=\"engine-card hero\">\n <div class=\"elo-badge\">Perf. 3163 Elo \u00b1 23<\/div>\n <h2 style=\"margin: 0; color: #D4AF37;\">Glaurung 2.2 x64 \u2014 8 Threads (2008)<\/h2>\n <div class=\"stat-line\">Hardware : <span class=\"highlight\">AMD Ryzen 9 7950X3D \u2014 8 threads actifs<\/span><\/div>\n <div class=\"stat-line\">Performance Globale : <span class=\"highlight\">50,0%<\/span> \u2014 Score : <span class=\"highlight\">37,5 \u2013 37,5<\/span><\/div>\n <div class=\"stat-line\">Elo moyen du pool adverse : <span class=\"highlight\">3163<\/span> | Performance calcul\u00e9e : <span class=\"highlight\">3163 Elo<\/span> (= pool, 50% exact) | Erreur : <span class=\"highlight\">\u00b1 23 Elo<\/span><\/div>\n <p style=\"font-size: 0.85em; color: #D4AF37; font-style: italic; margin-top: 10px;\">\n Glaurung 2.2 sur 8 threads du Ryzen 9 7950X3D r\u00e9alise une performance de 3163 Elo \u2014 r\u00e9sultat remarquable pour un moteur open source de 2008, anc\u00eatre direct de Stockfish. Sur un pool d'opposition particuli\u00e8rement relev\u00e9 (3163 Elo de moyenne), incluant Wasp 2.01 et Rybka 2.4MP tous deux en 8 threads sur Ryzen 9, Glaurung affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%. Ce score masque cependant une r\u00e9alit\u00e9 contrast\u00e9e : domin\u00e9 par les moteurs commerciaux \u00e0 hardware \u00e9gal (Wasp, Rybka), Glaurung compense en \u00e9crasant les adversaires brid\u00e9s par leur hardware (Shredder 10 mono-thread, Zappa sur Q6600). La surprise Naum \u2014 qui impose 70% des points \u00e0 Glaurung depuis un modeste Q6600 4 CPU \u2014 reste l'enseignement algorithmique le plus marquant de ce gauntlet.\n <\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">R\u00e9sultats du Gauntlet \u2014 Match par Match<\/div>\n <div class=\"section-inner\">\n <table class=\"recap-table\">\n <thead>\n <tr>\n <th style=\"text-align:left\">Adversaire<\/th>\n <th>Elo Adv.<\/th>\n <th>Hardware Adv.<\/th>\n <th>Glaurung Score<\/th>\n <th>Adv. Score<\/th>\n <th>Parties<\/th>\n <th>% Glaurung<\/th>\n <th>R\u00e9sultat<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr>\n <td><strong style=\"color:#fff\">Wasp 2.01 (8 CPU)<\/strong><\/td>\n <td>3297<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 8 threads<\/td>\n <td>5<\/td>\n <td>10<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">33,3%<\/span>\n <div class=\"pct-bar low\" style=\"width:20px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Rybka 2.4MP (8 CPU)<\/strong><\/td>\n <td>3230<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 8 threads<\/td>\n <td>4,5<\/td>\n <td>10,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">30,0%<\/span>\n <div class=\"pct-bar low\" style=\"width:18px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Naum 4.2 x64<\/strong><\/td>\n <td>3143<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td>4,5<\/td>\n <td>10,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">30,0%<\/span>\n <div class=\"pct-bar low\" style=\"width:18px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Shredder 10 UCI<\/strong><\/td>\n <td>3087<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 1 thread<\/td>\n <td>12<\/td>\n <td>3<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#4caf50;font-weight:bold\">80,0%<\/span>\n <div class=\"pct-bar high\" style=\"width:48px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#4caf50;font-weight:bold\">\u2714 Victoire<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Zappa Mexico II x64<\/strong><\/td>\n <td>3059<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td>11,5<\/td>\n <td>3,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#4caf50;font-weight:bold\">76,7%<\/span>\n <div class=\"pct-bar high\" style=\"width:46px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#4caf50;font-weight:bold\">\u2714 Victoire<\/td>\n <\/tr>\n <tr class=\"total-row\">\n <td><strong>TOTAL<\/strong><\/td>\n <td>3163 moy.<\/td>\n <td style=\"font-size:0.8em\">\u2014 Pool mixte<\/td>\n <td><strong>37,5<\/strong><\/td>\n <td><strong>37,5<\/strong><\/td>\n <td><strong>75<\/strong><\/td>\n <td><strong>50,0%<\/strong><\/td>\n <td style=\"color:#D4AF37;font-weight:bold\">2V \/ 0= \/ 3D<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <\/div>\n\n \n <div class=\"section-title\">Synth\u00e8se de Performance<\/div>\n <div class=\"section-inner\">\n <div class=\"perf-summary\">\n <div class=\"perf-box\">\n <div class=\"perf-label\">Score Total<\/div>\n <div class=\"perf-value\">37,5 \u2013 37,5<\/div>\n <div class=\"perf-sub\">75 parties jou\u00e9es<\/div>\n <\/div>\n <div class=\"perf-box\">\n <div class=\"perf-label\">% de gain<\/div>\n <div class=\"perf-value\">50,0%<\/div>\n <div class=\"perf-sub\">2 victoires \/ 0 nul \/ 3 d\u00e9faites<\/div>\n <\/div>\n <div class=\"perf-box\">\n <div class=\"perf-label\">Elo moyen pool<\/div>\n <div class=\"perf-value\">3163<\/div>\n <div class=\"perf-sub\">Opposition moyenne<\/div>\n <\/div>\n <div class=\"perf-box gold-box\">\n <div class=\"perf-label\">Performance Elo<\/div>\n <div class=\"perf-value\">3163 \u00b1 23<\/div>\n <div class=\"perf-sub\">= pool \u2014 50,0% exact \u2014 8 threads<\/div>\n <\/div>\n <\/div>\n <p style=\"font-size:0.82em; color:#888; margin-top:14px; background:#111; border:1px solid #2a2a2a; border-radius:6px; padding:12px 16px;\">\n <strong style=\"color:#D4AF37\">\ud83d\udcca Calcul de performance :<\/strong> Score de 50,0% (37,5\/75) sur un pool d'opposition moyen \u00e0 3163 Elo.\n \u00c0 50% exactement, le dp est nul par d\u00e9finition : la performance Elo est \u00e9gale \u00e0 la moyenne du pool.\n Performance r\u00e9sultante : <strong style=\"color:#fff\">3163 Elo<\/strong>.\n Erreur standard : \u03c3 = 400 \u00d7 \u221a(0,50 \u00d7 0,50 \/ 75) = <strong style=\"color:#fff\">\u00b1 23 Elo<\/strong>.\n Intervalle de confiance 95% : <strong style=\"color:#fff\">3117 \u2014 3209 Elo<\/strong>.\n Le bilan 2V\/3D masque une r\u00e9alit\u00e9 binaire : <strong style=\"color:#fff\">30-33% contre les moteurs \u00e0 hardware \u00e9gal (Ryzen 9 8T)<\/strong> \u2014 <strong style=\"color:#fff\">~78% contre les moteurs sur hardware inf\u00e9rieur (Q6600 \/ mono-thread)<\/strong>.\n <\/p>\n <\/div>\n\n \n <div class=\"section-title\">Opposition & Hardware<\/div>\n <div class=\"main-grid\">\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3297 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Wasp 2.01 (8 CPU)<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">5 \u2013 10<\/span> <span style=\"color:#e57373\">(33,3%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 8 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3230 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Rybka 2.4MP (8 CPU)<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">4,5 \u2013 10,5<\/span> <span style=\"color:#e57373\">(30,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 8 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3143 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Naum 4.2 x64<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">4,5 \u2013 10,5<\/span> <span style=\"color:#e57373\">(30,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Q6600 2.4 GHz \u2014 4 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3087 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Shredder 10 UCI<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">12 \u2013 3<\/span> <span style=\"color:#4caf50\">(80,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 1 thread<\/span>\n <\/div>\n <div class=\"engine-card\" style=\"grid-column: span 2;\">\n <div class=\"elo-badge\">3059 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Zappa Mexico II x64<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">11,5 \u2013 3,5<\/span> <span style=\"color:#4caf50\">(76,7%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Q6600 2.4 GHz \u2014 4 threads<\/span>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Classements SSDF Officiels \u2014 Glaurung 2.2 sur diff\u00e9rents Hardwares<\/div>\n <div class=\"section-inner\">\n <p style=\"font-size:0.84em; color:#888; margin: 0 0 14px 0;\">\n Sources : SSDF. Glaurung 2.2, moteur open source fondateur, dispose d'une seule mesure officielle SSDF sur Q6600 4 CPU, compl\u00e9t\u00e9e par le gauntlet maison 2026 sur Ryzen 9 7950X3D.\n <\/p>\n <table class=\"ssdf-table\">\n <thead>\n <tr>\n <th>Version \/ Hardware<\/th>\n <th>Threads<\/th>\n <th>NPS total estim\u00e9<\/th>\n <th>Elo SSDF<\/th>\n <th>Parties<\/th>\n <th>Gain vs base<\/th>\n <th>Source<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr>\n <td>Glaurung 2.2 x64 \u2014 Q6600 2.4 GHz (4 CPU)<\/td>\n <td>4<\/td>\n <td>1 693 \u00d7 4 = 6 772 kNps<\/td>\n <td>2995<\/td>\n <td>\u2014<\/td>\n <td>\u2014 r\u00e9f\u00e9rence<\/td>\n <td>SSDF<\/td>\n <\/tr>\n <tr class=\"ryzen-row\">\n <td>Glaurung 2.2 x64 \u2014 Ryzen 9 7950X3D (8 threads)<\/td>\n <td>8<\/td>\n <td>5 179 \u00d7 8 = 41 432 kNps<\/td>\n <td>3163 \u00b1 23<\/td>\n <td>75<\/td>\n <td style=\"color:#4caf50;font-weight:bold\">+168<\/td>\n <td>Gauntlet maison 2026<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <\/div>\n\n \n <div class=\"section-title\">Analyse \u2014 Loi d'\u00c9chelle Hardware & Performance Elo<\/div>\n <div class=\"section-inner\">\n <table class=\"hw-compare-table\">\n <thead>\n <tr>\n <th>Hardware<\/th>\n <th>NPS\/thread<\/th>\n <th>Threads<\/th>\n <th>NPS total<\/th>\n <th>Ratio vs base<\/th>\n <th>Elo mesur\u00e9<\/th>\n <th>Gain Elo r\u00e9el<\/th>\n <th>Elo\/doublement NPS<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr class=\"ref-row\">\n <td>Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td class=\"nps-col\">1 693<\/td>\n <td>4<\/td>\n <td class=\"nps-col\">6 772<\/td>\n <td>\u00d7 1,0<\/td>\n <td>2995<\/td>\n <td>\u2014<\/td>\n <td>\u2014<\/td>\n <\/tr>\n <tr class=\"ryzen-row\">\n <td>Ryzen 9 7950X3D \u2014 8T<\/td>\n <td class=\"nps-col\">5 179<\/td>\n <td>8<\/td>\n <td class=\"nps-col\">41 432<\/td>\n <td>\u00d7 6,1<\/td>\n <td>3163 \u00b1 23<\/td>\n <td style=\"color:#4caf50;font-weight:bold\">+168<\/td>\n <td>~46 Elo<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <p style=\"font-size:0.82em; color:#888; margin-top:14px; background:#111; border:1px solid #2a2a2a; border-radius:6px; padding:12px 16px;\">\n <strong style=\"color:#D4AF37\">\ud83d\udd0d Analyse de coh\u00e9rence :<\/strong>\n <br><br>\n Du Q6600 4 CPU (6 772 kNps) au Ryzen 9 7950X3D 8T (41 432 kNps), le ratio NPS est de \u00d76,1 soit 2,6 doublements. Le gain Elo attendu \u00e0 ~46 Elo\/doublement serait de <strong style=\"color:#fff\">+120 points<\/strong>. Le gain r\u00e9el observ\u00e9 est de <strong style=\"color:#fff\">+168 points<\/strong> \u2014 l\u00e9g\u00e8rement sup\u00e9rieur aux projections, ce qui reste dans les marges d'incertitude statistique (\u00b1 23 Elo). Contrairement \u00e0 Wasp qui b\u00e9n\u00e9ficiait d'un bonus V-Cache 3D de +105 Elo inexpliqu\u00e9s par le seul NPS, Glaurung suit une loi d'\u00e9chelle NPS coh\u00e9rente : le gain est proportionnel au logarithme du NPS total, sans sur-performance spectaculaire li\u00e9e au cache.\n <br><br>\n <strong style=\"color:#fff\">L'anomalie Naum :<\/strong> Naum 4.2 sur Q6600 4 CPU (3143 Elo) inflige 70% des points \u00e0 Glaurung sur Ryzen 9 8T (3163 Elo). Cet \u00e9cart th\u00e9orique de seulement 20 points se traduit par une domination pratique nette, sugg\u00e9rant que la cote SSDF de Naum 4.2 est sous-estim\u00e9e, ou que son algorithme de recherche est particuli\u00e8rement efficace dans les types de positions auxquelles Glaurung est vuln\u00e9rable.\n <\/p>\n <\/div>\n\n \n <div class=\"section-title\">Commentaire Technique<\/div>\n <div class=\"bio-section\">\n <div class=\"bio-card\" style=\"border-bottom: none;\">\n <span class=\"bio-name\">Analyse du Match-up \u2014 Glaurung 2.2 (8 threads, Ryzen 9 7950X3D)<\/span>\n <p class=\"stat-line\">Le gauntlet de Glaurung 2.2 r\u00e9v\u00e8le un moteur \u00e0 double visage. Face aux adversaires \u00e0 hardware \u00e9gal sur Ryzen 9 en 8 threads \u2014 Wasp 2.01 (3297) et Rybka 2.4MP (3230) \u2014 Glaurung encaisse des d\u00e9faites nettes \u00e0 33% et 30%, confirmant son rang inf\u00e9rieur dans la hi\u00e9rarchie algorithmique de l'\u00e8re commerciale 2005-2011. Face aux moteurs handicap\u00e9s par leur hardware \u2014 Zappa Mexico II sur Q6600 4 CPU (76,7%) et Shredder 10 mono-thread sur Ryzen 9 (80,0%) \u2014 Glaurung exploite pleinement sa sup\u00e9riorit\u00e9 en ressources de calcul. La v\u00e9ritable surprise est Naum 4.2 sur Q6600 4 CPU : malgr\u00e9 un hardware tr\u00e8s inf\u00e9rieur, Naum impose 70% des points \u00e0 Glaurung, d\u00e9monstration \u00e9loquente de la qualit\u00e9 algorithmique de ce moteur serbe. Au final, 50% exact sur un pool \u00e0 3163 : Glaurung 2.2 est parfaitement calibr\u00e9 \u00e0 sa performance estim\u00e9e sur Ryzen 9, confirmant la coh\u00e9rence rigoureuse de la m\u00e9thode de gauntlet maison.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Profil Technique \u2014 Glaurung (Tord Romstad)<\/div>\n <div class=\"section-inner\">\n <div class=\"innov-grid\">\n <div class=\"innov-card\">\n <div class=\"innov-title\">Anc\u00eatre direct de Stockfish<\/div>\n <p>Glaurung, d\u00e9velopp\u00e9 par Tord Romstad, est le pr\u00e9curseur g\u00e9n\u00e9tique direct de Stockfish. Son code a servi de base au projet Stockfish lanc\u00e9 en 2008, faisant de Glaurung 2.2 l'un des moteurs open source les plus influents de l'histoire des \u00e9checs informatiques. Sa force r\u00e9side dans une impl\u00e9mentation alpha-b\u00eata rigoureuse et une \u00e9valuation bien \u00e9quilibr\u00e9e.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Scaling MP remarquable et coh\u00e9rent<\/div>\n <p>Sur Q6600 4 CPU, Glaurung 2.2 atteint 2995 Elo SSDF. Le passage au Ryzen 9 7950X3D en 8 threads apporte +168 Elo, coh\u00e9rent avec le ratio NPS de \u00d76,1 entre les deux configurations. Cette loi d'\u00e9chelle r\u00e9guli\u00e8re t\u00e9moigne d'une impl\u00e9mentation parall\u00e8le propre pour un moteur de 2008, sans irr\u00e9gularit\u00e9 notable.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Pas d'effet V-Cache d\u00e9tect\u00e9<\/div>\n <p>Contrairement \u00e0 Wasp 2.01 qui tire un b\u00e9n\u00e9fice exceptionnel du V-Cache 3D du 7950X3D (+105 Elo non expliqu\u00e9s par le NPS), Glaurung 2.2 affiche un gain strictement conforme \u00e0 la loi NPS classique. Son architecture de table de transposition ne b\u00e9n\u00e9ficie pas du m\u00eame multiplicateur mat\u00e9riel, ce qui distingue nettement les deux moteurs sur ce hardware.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">R\u00e9f\u00e9rence historique du logiciel libre<\/div>\n <p>En 2008, Glaurung 2.2 repr\u00e9sentait l'\u00e9tat de l'art open source avec ~2995 Elo SSDF. Sa publication en code ouvert a catalys\u00e9 toute une g\u00e9n\u00e9ration de moteurs libres, culminant avec Stockfish. Tester Glaurung en 2026 sur Ryzen 9, c'est mesurer la distance parcourue en 18 ans d'\u00e9volution algorithmique depuis ses propres fondations.<\/p>\n <\/div>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Chronologie Glaurung \u2014 Mesures & H\u00e9ritage<\/div>\n <div class=\"section-inner\">\n <ul class=\"timeline\">\n <li><span class=\"tl-year\">2004<\/span><span class=\"tl-event\">Tord Romstad publie les premi\u00e8res versions de <strong>Glaurung<\/strong>, nomm\u00e9 d'apr\u00e8s le dragon de Tolkien. Entr\u00e9e progressive dans les classements SSDF.<\/span><\/li>\n <li><span class=\"tl-year\">2007<\/span><span class=\"tl-event\"><strong>Glaurung 2.2 x64<\/strong> : version finale et la plus aboutie. Mesur\u00e9e \u00e0 <strong>2995 Elo SSDF<\/strong> sur Q6600 4 CPU \u2014 r\u00e9f\u00e9rence de base du gauntlet.<\/span><\/li>\n <li><span class=\"tl-year\">2008<\/span><span class=\"tl-event\">Tord Romstad, Marco Costalba et Joona Kiiski fondent <strong>Stockfish<\/strong> en partant du code de Glaurung. Glaurung 2.2 entre dans l'histoire comme point de d\u00e9part du moteur open source le plus fort du monde.<\/span><\/li>\n <li><span class=\"tl-year\">2026<\/span><span class=\"tl-event\">Mesure exp\u00e9rimentale : <strong>Glaurung 2.2 x64, 8 threads, Ryzen 9 7950X3D<\/strong> \u2014 performance de <strong>3163 \u00b1 23 Elo<\/strong>. 50% exact sur pool \u00e0 3163 \u2014 coh\u00e9rence parfaite avec les projections NPS. Laboratoire Chroniques Software.<\/span><\/li>\n <\/ul>\n <\/div>\n\n \n <div class=\"section-inner\" style=\"padding-top: 0;\">\n <div class=\"quote-box\">\n <p>\"Glaurung 2.2 sur 8 threads du Ryzen 9 7950X3D atteint 3163 Elo \u2014 exactement \u00e0 hauteur de son pool adverse \u00e0 50,0%. Ce r\u00e9sultat parfaitement \u00e9quilibr\u00e9 cache une r\u00e9alit\u00e9 tranch\u00e9e : domin\u00e9 \u00e0 30-33% par Wasp et Rybka \u00e0 hardware \u00e9gal, dominant sans partage Shredder mono-thread et Zappa sur Q6600. La contre-performance face \u00e0 Naum 4.2 sur Q6600 reste l'enseignement algorithmique le plus marquant du gauntlet.\"<\/p>\n <span class=\"quote-src\">\u2014 Laboratoire Chroniques Software, Gauntlet maison 2026<\/span>\n <\/div>\n <div class=\"quote-box\">\n <p>\"Glaurung 2.2 est l'Adam de Stockfish. Le tester en 2026 sur un Ryzen 9 7950X3D, c'est mesurer combien le hardware moderne peut ressusciter un code vieux de 18 ans \u2014 la r\u00e9ponse est 3163 Elo, le niveau d'un moteur commercial de premier plan de son \u00e9poque, port\u00e9 par la puissance brute d'un processeur que ses cr\u00e9ateurs n'auraient jamais imagin\u00e9. Et contrairement \u00e0 Wasp, Glaurung y arrive par la seule voie du NPS brut, sans le moindre bonus V-Cache.\"<\/p>\n <span class=\"quote-src\">\u2014 Chroniques Software, Analyse comparative Glaurung & h\u00e9ritage Stockfish, 2026<\/span>\n <\/div>\n <\/div>\n\n <div class=\"footer-note\">\n Exp\u00e9rience \"Chroniques Software\" \u2014 Laboratoire 2026 \u2014 Glaurung 2.2 x64 (8 threads) sur AMD Ryzen 9 7950X3D \u2014 Performance confirm\u00e9e : <strong>3163 \u00b1 23 Elo<\/strong> (75 parties, \u03c3 = 400 \u00d7 \u221a(p\u00b7(1\u2212p)\/N)). Donn\u00e9es SSDF : liste officielle SSDF.\n <\/div>\n\n <\/div>\n<\/div>\n<\/body>\n<\/html>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Glaurung 2.2 x64, anc\u00eatre direct de Stockfish, r\u00e9alise une performance de 3163 \u00b1 23 Elo sur 8 threads du Ryzen 9 7950X3D \u2014 soit +168 points par rapport \u00e0 sa cote SSDF officielle sur Q6600 4 CPU. Sur 75 parties, il affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%, dominant les moteurs sur hardware inf\u00e9rieur (76-80%) mais nettement distanc\u00e9 par Wasp et Rybka \u00e0 hardware \u00e9gal (30-33%). Contrairement \u00e0 Wasp, dont le gain exceptionnel sur ce processeur sugg\u00e8re un fort effet V-Cache 3D, Glaurung suit une loi d’\u00e9chelle NPS strictement classique.<\/p>\n","protected":false},"author":1,"featured_media":3496,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"","neve_meta_content_width":0,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":""},"categories":[37],"tags":[],"rttpg_featured_image_url":{"full":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"landscape":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"portraits":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"thumbnail":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-150x150.png",150,150,true],"medium":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-300x249.png",300,249,true],"large":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-1024x848.png",1024,848,true],"1536x1536":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"2048x2048":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"neve-blog":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-930x620.png",930,620,true]},"rttpg_author":{"display_name":"valterd","author_link":"https:\/\/www.chess.lumn.eu\/wordpress\/author\/valterd\/"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/www.chess.lumn.eu\/wordpress\/category\/soft\/\" rel=\"category tag\">Soft<\/a>","rttpg_excerpt":"Glaurung 2.2 x64, anc\u00eatre direct de Stockfish, r\u00e9alise une performance de 3163 \u00b1 23 Elo sur 8 threads du Ryzen 9 7950X3D \u2014 soit +168 points par rapport \u00e0 sa cote SSDF officielle sur Q6600 4 CPU. Sur 75 parties, il affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%, dominant les moteurs sur hardware inf\u00e9rieur (76-80%)…","_links":{"self":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495"}],"collection":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/comments?post=3495"}],"version-history":[{"count":4,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495\/revisions"}],"predecessor-version":[{"id":3501,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495\/revisions\/3501"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media\/3496"}],"wp:attachment":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media?parent=3495"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/categories?post=3495"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/tags?post=3495"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

\n\t\t\t\t\t\t

\n\t\t\t\t

\n\t\t\t\t\t\t\t\t\t

\n\t\t\t\t\t\t

\n\t\t\t\t\t\t\t

\n\t\t\t\t\t

\n\t\t\t

\n\t\t\t\t\t\t\t

\n\t\t\t\t\t\t

\n\t\t\t\t

\n\t\t\t\n\n\n\n\nGauntlet Glaurung 2.2 x64 8 CPU Ryzen 9 \u2014 Chroniques Software<\/title>\n<\/head>\n<body>\n<div class=\"gauntlet-container\" style=\"all: initial; display: block; font-family: 'Segoe UI', Helvetica, Arial, sans-serif; background-color: #0a0a0a; padding: 40px; color: #e0e0e0; line-height: 1.6;\">\n <style>\n .gauntlet-wrapper { max-width: 950px; margin: 0 auto; background: #161616; border: 1px solid #333; border-radius: 12px; overflow: hidden; box-shadow: 0 20px 50px rgba(0,0,0,0.6); }\n .gauntlet-header { background: linear-gradient(135deg, #2c240a 0%, #000000 100%); padding: 30px; text-align: center; border-bottom: 3px solid #D4AF37; }\n .gauntlet-header h1 { color: #D4AF37; margin: 0; font-size: 2em; text-transform: uppercase; letter-spacing: 3px; }\n\n .section-title { color: #D4AF37; border-left: 4px solid #D4AF37; padding-left: 15px; margin: 30px 20px 15px 20px; font-size: 1.4em; text-transform: uppercase; }\n .main-grid { display: grid; grid-template-columns: 1fr 1fr; gap: 20px; padding: 0 20px 20px 20px; }\n\n .engine-card { background: #222; border: 1px solid #444; padding: 15px; border-radius: 8px; position: relative; }\n .engine-card.hero { border: 2px solid #D4AF37; background: #1d1a10; grid-column: span 2; }\n\n .elo-badge { position: absolute; top: 15px; right: 15px; background: #D4AF37; color: #000; padding: 2px 10px; border-radius: 4px; font-weight: bold; font-size: 0.9em; }\n .stat-line { font-size: 0.9em; margin: 5px 0; color: #bbb; }\n .hw-info { font-family: 'Consolas', monospace; font-size: 0.8em; color: #D4AF37; display: block; margin-top: 8px; border-top: 1px solid #333; padding-top: 5px; }\n .highlight { color: #fff; font-weight: bold; }\n\n .bio-section { padding: 0 20px 20px 20px; }\n .bio-card { margin-bottom: 20px; padding-bottom: 15px; border-bottom: 1px solid #333; }\n .bio-name { color: #D4AF37; font-weight: bold; font-size: 1.1em; margin-bottom: 5px; display: block; }\n .footer-note { text-align: center; padding: 20px; font-size: 0.8em; color: #666; border-top: 1px solid #333; }\n\n .ssdf-table { width: 100%; border-collapse: collapse; font-size: 0.82em; margin-top: 10px; }\n .ssdf-table th { background: #2a2a2a; color: #D4AF37; padding: 7px 10px; text-align: left; border-bottom: 2px solid #D4AF37; font-size: 0.85em; letter-spacing: 0.05em; }\n .ssdf-table td { padding: 6px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; }\n .ssdf-table tr:hover td { background: #1e1e1e; }\n .ssdf-table .peak-row td { color: #D4AF37; font-weight: bold; }\n .ssdf-table .ryzen-row td { color: #fff; font-weight: bold; background: #1d1a10; }\n .ssdf-table .ryzen-row td:first-child::before { content: \"\u26a1 \"; }\n\n .timeline { padding: 0; list-style: none; margin: 0; }\n .timeline li { display: flex; gap: 16px; padding: 10px 0; border-bottom: 1px solid #2a2a2a; font-size: 0.87em; }\n .timeline li:last-child { border-bottom: none; }\n .tl-year { min-width: 52px; color: #D4AF37; font-weight: bold; font-family: 'Consolas', monospace; }\n .tl-event { color: #ccc; }\n .tl-event strong { color: #fff; }\n\n .innov-grid { display: grid; grid-template-columns: 1fr 1fr; gap: 12px; margin-top: 10px; }\n .innov-card { background: #1c1c1c; border: 1px solid #3a3a2a; border-radius: 6px; padding: 12px 14px; }\n .innov-card .innov-title { color: #D4AF37; font-weight: bold; font-size: 0.88em; margin-bottom: 5px; }\n .innov-card p { color: #aaa; font-size: 0.82em; margin: 0; line-height: 1.5; }\n\n .quote-box { background: #111; border-left: 3px solid #D4AF37; padding: 14px 18px; margin: 12px 0; border-radius: 0 6px 6px 0; }\n .quote-box p { color: #bbb; font-style: italic; font-size: 0.87em; margin: 0 0 6px 0; line-height: 1.6; }\n .quote-box .quote-src { color: #666; font-size: 0.78em; }\n\n .full-width { grid-column: span 2; }\n .section-inner { padding: 0 20px 20px 20px; }\n\n .recap-table { width: 100%; border-collapse: collapse; font-size: 0.84em; margin-top: 10px; }\n .recap-table th { background: #2a2a2a; color: #D4AF37; padding: 8px 10px; text-align: center; border-bottom: 2px solid #D4AF37; font-size: 0.85em; letter-spacing: 0.05em; }\n .recap-table td { padding: 7px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; text-align: center; }\n .recap-table td:first-child { text-align: left; }\n .recap-table tr:hover td { background: #1e1e1e; }\n .recap-table .total-row td { color: #D4AF37; font-weight: bold; border-top: 2px solid #D4AF37; }\n .pct-bar-wrap { display: flex; align-items: center; gap: 8px; }\n .pct-bar { height: 8px; border-radius: 4px; display: inline-block; }\n .pct-bar.low { background: #e57373; }\n .pct-bar.mid { background: #ffb74d; }\n .pct-bar.high { background: #4caf50; }\n\n .perf-summary { display: grid; grid-template-columns: repeat(4, 1fr); gap: 12px; margin-top: 14px; }\n .perf-box { background: #1c1c1c; border: 1px solid #3a3a2a; border-radius: 8px; padding: 14px; text-align: center; }\n .perf-box.gold-box { border-color: #D4AF37; background: #1d1a10; }\n .perf-box .perf-label { color: #888; font-size: 0.75em; text-transform: uppercase; letter-spacing: 0.07em; margin-bottom: 6px; }\n .perf-box .perf-value { color: #fff; font-size: 1.4em; font-weight: bold; font-family: 'Consolas', monospace; }\n .perf-box.gold-box .perf-value { color: #D4AF37; }\n .perf-box .perf-sub { color: #666; font-size: 0.72em; margin-top: 4px; }\n\n .hw-compare-table { width: 100%; border-collapse: collapse; font-size: 0.84em; margin-top: 10px; }\n .hw-compare-table th { background: #2a2a2a; color: #D4AF37; padding: 8px 10px; text-align: center; border-bottom: 2px solid #D4AF37; }\n .hw-compare-table th:first-child { text-align: left; }\n .hw-compare-table td { padding: 7px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; text-align: center; }\n .hw-compare-table td:first-child { text-align: left; }\n .hw-compare-table tr:hover td { background: #1e1e1e; }\n .hw-compare-table .ryzen-row td { color: #D4AF37; font-weight: bold; background: #1d1a10; }\n .hw-compare-table .ref-row td { color: #666; font-style: italic; }\n .hw-compare-table .gain-pos { color: #4caf50; font-weight: bold; }\n .hw-compare-table .nps-col { font-family: 'Consolas', monospace; font-size: 0.85em; }\n <\/style>\n\n <div class=\"gauntlet-wrapper\">\n\n \n <div class=\"gauntlet-header\">\n <h1>Gauntlet : Glaurung 2.2 x64 (8 Threads) vs The Era<\/h1>\n <p style=\"color: #888; margin-top: 10px;\">Hardware Target : Glaurung 2.2 x64 \u2014 8 threads | AMD Ryzen 9 7950X3D | Pool mixte (75 parties)<\/p>\n <\/div>\n\n \n <div class=\"section-title\">L'Anc\u00eatre de Stockfish<\/div>\n <div class=\"main-grid\">\n <div class=\"engine-card hero\">\n <div class=\"elo-badge\">Perf. 3163 Elo \u00b1 23<\/div>\n <h2 style=\"margin: 0; color: #D4AF37;\">Glaurung 2.2 x64 \u2014 8 Threads (2008)<\/h2>\n <div class=\"stat-line\">Hardware : <span class=\"highlight\">AMD Ryzen 9 7950X3D \u2014 8 threads actifs<\/span><\/div>\n <div class=\"stat-line\">Performance Globale : <span class=\"highlight\">50,0%<\/span> \u2014 Score : <span class=\"highlight\">37,5 \u2013 37,5<\/span><\/div>\n <div class=\"stat-line\">Elo moyen du pool adverse : <span class=\"highlight\">3163<\/span> | Performance calcul\u00e9e : <span class=\"highlight\">3163 Elo<\/span> (= pool, 50% exact) | Erreur : <span class=\"highlight\">\u00b1 23 Elo<\/span><\/div>\n <p style=\"font-size: 0.85em; color: #D4AF37; font-style: italic; margin-top: 10px;\">\n Glaurung 2.2 sur 8 threads du Ryzen 9 7950X3D r\u00e9alise une performance de 3163 Elo \u2014 r\u00e9sultat remarquable pour un moteur open source de 2008, anc\u00eatre direct de Stockfish. Sur un pool d'opposition particuli\u00e8rement relev\u00e9 (3163 Elo de moyenne), incluant Wasp 2.01 et Rybka 2.4MP tous deux en 8 threads sur Ryzen 9, Glaurung affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%. Ce score masque cependant une r\u00e9alit\u00e9 contrast\u00e9e : domin\u00e9 par les moteurs commerciaux \u00e0 hardware \u00e9gal (Wasp, Rybka), Glaurung compense en \u00e9crasant les adversaires brid\u00e9s par leur hardware (Shredder 10 mono-thread, Zappa sur Q6600). La surprise Naum \u2014 qui impose 70% des points \u00e0 Glaurung depuis un modeste Q6600 4 CPU \u2014 reste l'enseignement algorithmique le plus marquant de ce gauntlet.\n <\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">R\u00e9sultats du Gauntlet \u2014 Match par Match<\/div>\n <div class=\"section-inner\">\n <table class=\"recap-table\">\n <thead>\n <tr>\n <th style=\"text-align:left\">Adversaire<\/th>\n <th>Elo Adv.<\/th>\n <th>Hardware Adv.<\/th>\n <th>Glaurung Score<\/th>\n <th>Adv. Score<\/th>\n <th>Parties<\/th>\n <th>% Glaurung<\/th>\n <th>R\u00e9sultat<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr>\n <td><strong style=\"color:#fff\">Wasp 2.01 (8 CPU)<\/strong><\/td>\n <td>3297<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 8 threads<\/td>\n <td>5<\/td>\n <td>10<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">33,3%<\/span>\n <div class=\"pct-bar low\" style=\"width:20px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Rybka 2.4MP (8 CPU)<\/strong><\/td>\n <td>3230<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 8 threads<\/td>\n <td>4,5<\/td>\n <td>10,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">30,0%<\/span>\n <div class=\"pct-bar low\" style=\"width:18px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Naum 4.2 x64<\/strong><\/td>\n <td>3143<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td>4,5<\/td>\n <td>10,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#e57373;font-weight:bold\">30,0%<\/span>\n <div class=\"pct-bar low\" style=\"width:18px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#e57373;font-weight:bold\">\u2718 D\u00e9faite<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Shredder 10 UCI<\/strong><\/td>\n <td>3087<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Ryzen 9 7950X3D \u2014 1 thread<\/td>\n <td>12<\/td>\n <td>3<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#4caf50;font-weight:bold\">80,0%<\/span>\n <div class=\"pct-bar high\" style=\"width:48px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#4caf50;font-weight:bold\">\u2714 Victoire<\/td>\n <\/tr>\n <tr>\n <td><strong style=\"color:#fff\">Zappa Mexico II x64<\/strong><\/td>\n <td>3059<\/td>\n <td style=\"font-size:0.8em;color:#aaa\">Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td>11,5<\/td>\n <td>3,5<\/td>\n <td>15<\/td>\n <td>\n <div class=\"pct-bar-wrap\">\n <span style=\"color:#4caf50;font-weight:bold\">76,7%<\/span>\n <div class=\"pct-bar high\" style=\"width:46px\"><\/div>\n <\/div>\n <\/td>\n <td style=\"color:#4caf50;font-weight:bold\">\u2714 Victoire<\/td>\n <\/tr>\n <tr class=\"total-row\">\n <td><strong>TOTAL<\/strong><\/td>\n <td>3163 moy.<\/td>\n <td style=\"font-size:0.8em\">\u2014 Pool mixte<\/td>\n <td><strong>37,5<\/strong><\/td>\n <td><strong>37,5<\/strong><\/td>\n <td><strong>75<\/strong><\/td>\n <td><strong>50,0%<\/strong><\/td>\n <td style=\"color:#D4AF37;font-weight:bold\">2V \/ 0= \/ 3D<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <\/div>\n\n \n <div class=\"section-title\">Synth\u00e8se de Performance<\/div>\n <div class=\"section-inner\">\n <div class=\"perf-summary\">\n <div class=\"perf-box\">\n <div class=\"perf-label\">Score Total<\/div>\n <div class=\"perf-value\">37,5 \u2013 37,5<\/div>\n <div class=\"perf-sub\">75 parties jou\u00e9es<\/div>\n <\/div>\n <div class=\"perf-box\">\n <div class=\"perf-label\">% de gain<\/div>\n <div class=\"perf-value\">50,0%<\/div>\n <div class=\"perf-sub\">2 victoires \/ 0 nul \/ 3 d\u00e9faites<\/div>\n <\/div>\n <div class=\"perf-box\">\n <div class=\"perf-label\">Elo moyen pool<\/div>\n <div class=\"perf-value\">3163<\/div>\n <div class=\"perf-sub\">Opposition moyenne<\/div>\n <\/div>\n <div class=\"perf-box gold-box\">\n <div class=\"perf-label\">Performance Elo<\/div>\n <div class=\"perf-value\">3163 \u00b1 23<\/div>\n <div class=\"perf-sub\">= pool \u2014 50,0% exact \u2014 8 threads<\/div>\n <\/div>\n <\/div>\n <p style=\"font-size:0.82em; color:#888; margin-top:14px; background:#111; border:1px solid #2a2a2a; border-radius:6px; padding:12px 16px;\">\n <strong style=\"color:#D4AF37\">\ud83d\udcca Calcul de performance :<\/strong> Score de 50,0% (37,5\/75) sur un pool d'opposition moyen \u00e0 3163 Elo.\n \u00c0 50% exactement, le dp est nul par d\u00e9finition : la performance Elo est \u00e9gale \u00e0 la moyenne du pool.\n Performance r\u00e9sultante : <strong style=\"color:#fff\">3163 Elo<\/strong>.\n Erreur standard : \u03c3 = 400 \u00d7 \u221a(0,50 \u00d7 0,50 \/ 75) = <strong style=\"color:#fff\">\u00b1 23 Elo<\/strong>.\n Intervalle de confiance 95% : <strong style=\"color:#fff\">3117 \u2014 3209 Elo<\/strong>.\n Le bilan 2V\/3D masque une r\u00e9alit\u00e9 binaire : <strong style=\"color:#fff\">30-33% contre les moteurs \u00e0 hardware \u00e9gal (Ryzen 9 8T)<\/strong> \u2014 <strong style=\"color:#fff\">~78% contre les moteurs sur hardware inf\u00e9rieur (Q6600 \/ mono-thread)<\/strong>.\n <\/p>\n <\/div>\n\n \n <div class=\"section-title\">Opposition & Hardware<\/div>\n <div class=\"main-grid\">\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3297 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Wasp 2.01 (8 CPU)<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">5 \u2013 10<\/span> <span style=\"color:#e57373\">(33,3%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 8 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3230 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Rybka 2.4MP (8 CPU)<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">4,5 \u2013 10,5<\/span> <span style=\"color:#e57373\">(30,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 8 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3143 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Naum 4.2 x64<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">4,5 \u2013 10,5<\/span> <span style=\"color:#e57373\">(30,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Q6600 2.4 GHz \u2014 4 threads<\/span>\n <\/div>\n <div class=\"engine-card\">\n <div class=\"elo-badge\">3087 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Shredder 10 UCI<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">12 \u2013 3<\/span> <span style=\"color:#4caf50\">(80,0%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Ryzen 9 7950X3D \u2014 1 thread<\/span>\n <\/div>\n <div class=\"engine-card\" style=\"grid-column: span 2;\">\n <div class=\"elo-badge\">3059 Elo<\/div>\n <h3 style=\"margin: 0; color: #fff;\">Zappa Mexico II x64<\/h3>\n <div class=\"stat-line\">Score Glaurung : <span class=\"highlight\">11,5 \u2013 3,5<\/span> <span style=\"color:#4caf50\">(76,7%)<\/span><\/div>\n <span class=\"hw-info\">Hardware: Q6600 2.4 GHz \u2014 4 threads<\/span>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Classements SSDF Officiels \u2014 Glaurung 2.2 sur diff\u00e9rents Hardwares<\/div>\n <div class=\"section-inner\">\n <p style=\"font-size:0.84em; color:#888; margin: 0 0 14px 0;\">\n Sources : SSDF. Glaurung 2.2, moteur open source fondateur, dispose d'une seule mesure officielle SSDF sur Q6600 4 CPU, compl\u00e9t\u00e9e par le gauntlet maison 2026 sur Ryzen 9 7950X3D.\n <\/p>\n <table class=\"ssdf-table\">\n <thead>\n <tr>\n <th>Version \/ Hardware<\/th>\n <th>Threads<\/th>\n <th>NPS total estim\u00e9<\/th>\n <th>Elo SSDF<\/th>\n <th>Parties<\/th>\n <th>Gain vs base<\/th>\n <th>Source<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr>\n <td>Glaurung 2.2 x64 \u2014 Q6600 2.4 GHz (4 CPU)<\/td>\n <td>4<\/td>\n <td>1 693 \u00d7 4 = 6 772 kNps<\/td>\n <td>2995<\/td>\n <td>\u2014<\/td>\n <td>\u2014 r\u00e9f\u00e9rence<\/td>\n <td>SSDF<\/td>\n <\/tr>\n <tr class=\"ryzen-row\">\n <td>Glaurung 2.2 x64 \u2014 Ryzen 9 7950X3D (8 threads)<\/td>\n <td>8<\/td>\n <td>5 179 \u00d7 8 = 41 432 kNps<\/td>\n <td>3163 \u00b1 23<\/td>\n <td>75<\/td>\n <td style=\"color:#4caf50;font-weight:bold\">+168<\/td>\n <td>Gauntlet maison 2026<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <\/div>\n\n \n <div class=\"section-title\">Analyse \u2014 Loi d'\u00c9chelle Hardware & Performance Elo<\/div>\n <div class=\"section-inner\">\n <table class=\"hw-compare-table\">\n <thead>\n <tr>\n <th>Hardware<\/th>\n <th>NPS\/thread<\/th>\n <th>Threads<\/th>\n <th>NPS total<\/th>\n <th>Ratio vs base<\/th>\n <th>Elo mesur\u00e9<\/th>\n <th>Gain Elo r\u00e9el<\/th>\n <th>Elo\/doublement NPS<\/th>\n <\/tr>\n <\/thead>\n <tbody>\n <tr class=\"ref-row\">\n <td>Q6600 2.4 GHz \u2014 4 CPU<\/td>\n <td class=\"nps-col\">1 693<\/td>\n <td>4<\/td>\n <td class=\"nps-col\">6 772<\/td>\n <td>\u00d7 1,0<\/td>\n <td>2995<\/td>\n <td>\u2014<\/td>\n <td>\u2014<\/td>\n <\/tr>\n <tr class=\"ryzen-row\">\n <td>Ryzen 9 7950X3D \u2014 8T<\/td>\n <td class=\"nps-col\">5 179<\/td>\n <td>8<\/td>\n <td class=\"nps-col\">41 432<\/td>\n <td>\u00d7 6,1<\/td>\n <td>3163 \u00b1 23<\/td>\n <td style=\"color:#4caf50;font-weight:bold\">+168<\/td>\n <td>~46 Elo<\/td>\n <\/tr>\n <\/tbody>\n <\/table>\n <p style=\"font-size:0.82em; color:#888; margin-top:14px; background:#111; border:1px solid #2a2a2a; border-radius:6px; padding:12px 16px;\">\n <strong style=\"color:#D4AF37\">\ud83d\udd0d Analyse de coh\u00e9rence :<\/strong>\n <br><br>\n Du Q6600 4 CPU (6 772 kNps) au Ryzen 9 7950X3D 8T (41 432 kNps), le ratio NPS est de \u00d76,1 soit 2,6 doublements. Le gain Elo attendu \u00e0 ~46 Elo\/doublement serait de <strong style=\"color:#fff\">+120 points<\/strong>. Le gain r\u00e9el observ\u00e9 est de <strong style=\"color:#fff\">+168 points<\/strong> \u2014 l\u00e9g\u00e8rement sup\u00e9rieur aux projections, ce qui reste dans les marges d'incertitude statistique (\u00b1 23 Elo). Contrairement \u00e0 Wasp qui b\u00e9n\u00e9ficiait d'un bonus V-Cache 3D de +105 Elo inexpliqu\u00e9s par le seul NPS, Glaurung suit une loi d'\u00e9chelle NPS coh\u00e9rente : le gain est proportionnel au logarithme du NPS total, sans sur-performance spectaculaire li\u00e9e au cache.\n <br><br>\n <strong style=\"color:#fff\">L'anomalie Naum :<\/strong> Naum 4.2 sur Q6600 4 CPU (3143 Elo) inflige 70% des points \u00e0 Glaurung sur Ryzen 9 8T (3163 Elo). Cet \u00e9cart th\u00e9orique de seulement 20 points se traduit par une domination pratique nette, sugg\u00e9rant que la cote SSDF de Naum 4.2 est sous-estim\u00e9e, ou que son algorithme de recherche est particuli\u00e8rement efficace dans les types de positions auxquelles Glaurung est vuln\u00e9rable.\n <\/p>\n <\/div>\n\n \n <div class=\"section-title\">Commentaire Technique<\/div>\n <div class=\"bio-section\">\n <div class=\"bio-card\" style=\"border-bottom: none;\">\n <span class=\"bio-name\">Analyse du Match-up \u2014 Glaurung 2.2 (8 threads, Ryzen 9 7950X3D)<\/span>\n <p class=\"stat-line\">Le gauntlet de Glaurung 2.2 r\u00e9v\u00e8le un moteur \u00e0 double visage. Face aux adversaires \u00e0 hardware \u00e9gal sur Ryzen 9 en 8 threads \u2014 Wasp 2.01 (3297) et Rybka 2.4MP (3230) \u2014 Glaurung encaisse des d\u00e9faites nettes \u00e0 33% et 30%, confirmant son rang inf\u00e9rieur dans la hi\u00e9rarchie algorithmique de l'\u00e8re commerciale 2005-2011. Face aux moteurs handicap\u00e9s par leur hardware \u2014 Zappa Mexico II sur Q6600 4 CPU (76,7%) et Shredder 10 mono-thread sur Ryzen 9 (80,0%) \u2014 Glaurung exploite pleinement sa sup\u00e9riorit\u00e9 en ressources de calcul. La v\u00e9ritable surprise est Naum 4.2 sur Q6600 4 CPU : malgr\u00e9 un hardware tr\u00e8s inf\u00e9rieur, Naum impose 70% des points \u00e0 Glaurung, d\u00e9monstration \u00e9loquente de la qualit\u00e9 algorithmique de ce moteur serbe. Au final, 50% exact sur un pool \u00e0 3163 : Glaurung 2.2 est parfaitement calibr\u00e9 \u00e0 sa performance estim\u00e9e sur Ryzen 9, confirmant la coh\u00e9rence rigoureuse de la m\u00e9thode de gauntlet maison.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Profil Technique \u2014 Glaurung (Tord Romstad)<\/div>\n <div class=\"section-inner\">\n <div class=\"innov-grid\">\n <div class=\"innov-card\">\n <div class=\"innov-title\">Anc\u00eatre direct de Stockfish<\/div>\n <p>Glaurung, d\u00e9velopp\u00e9 par Tord Romstad, est le pr\u00e9curseur g\u00e9n\u00e9tique direct de Stockfish. Son code a servi de base au projet Stockfish lanc\u00e9 en 2008, faisant de Glaurung 2.2 l'un des moteurs open source les plus influents de l'histoire des \u00e9checs informatiques. Sa force r\u00e9side dans une impl\u00e9mentation alpha-b\u00eata rigoureuse et une \u00e9valuation bien \u00e9quilibr\u00e9e.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Scaling MP remarquable et coh\u00e9rent<\/div>\n <p>Sur Q6600 4 CPU, Glaurung 2.2 atteint 2995 Elo SSDF. Le passage au Ryzen 9 7950X3D en 8 threads apporte +168 Elo, coh\u00e9rent avec le ratio NPS de \u00d76,1 entre les deux configurations. Cette loi d'\u00e9chelle r\u00e9guli\u00e8re t\u00e9moigne d'une impl\u00e9mentation parall\u00e8le propre pour un moteur de 2008, sans irr\u00e9gularit\u00e9 notable.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Pas d'effet V-Cache d\u00e9tect\u00e9<\/div>\n <p>Contrairement \u00e0 Wasp 2.01 qui tire un b\u00e9n\u00e9fice exceptionnel du V-Cache 3D du 7950X3D (+105 Elo non expliqu\u00e9s par le NPS), Glaurung 2.2 affiche un gain strictement conforme \u00e0 la loi NPS classique. Son architecture de table de transposition ne b\u00e9n\u00e9ficie pas du m\u00eame multiplicateur mat\u00e9riel, ce qui distingue nettement les deux moteurs sur ce hardware.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">R\u00e9f\u00e9rence historique du logiciel libre<\/div>\n <p>En 2008, Glaurung 2.2 repr\u00e9sentait l'\u00e9tat de l'art open source avec ~2995 Elo SSDF. Sa publication en code ouvert a catalys\u00e9 toute une g\u00e9n\u00e9ration de moteurs libres, culminant avec Stockfish. Tester Glaurung en 2026 sur Ryzen 9, c'est mesurer la distance parcourue en 18 ans d'\u00e9volution algorithmique depuis ses propres fondations.<\/p>\n <\/div>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Chronologie Glaurung \u2014 Mesures & H\u00e9ritage<\/div>\n <div class=\"section-inner\">\n <ul class=\"timeline\">\n <li><span class=\"tl-year\">2004<\/span><span class=\"tl-event\">Tord Romstad publie les premi\u00e8res versions de <strong>Glaurung<\/strong>, nomm\u00e9 d'apr\u00e8s le dragon de Tolkien. Entr\u00e9e progressive dans les classements SSDF.<\/span><\/li>\n <li><span class=\"tl-year\">2007<\/span><span class=\"tl-event\"><strong>Glaurung 2.2 x64<\/strong> : version finale et la plus aboutie. Mesur\u00e9e \u00e0 <strong>2995 Elo SSDF<\/strong> sur Q6600 4 CPU \u2014 r\u00e9f\u00e9rence de base du gauntlet.<\/span><\/li>\n <li><span class=\"tl-year\">2008<\/span><span class=\"tl-event\">Tord Romstad, Marco Costalba et Joona Kiiski fondent <strong>Stockfish<\/strong> en partant du code de Glaurung. Glaurung 2.2 entre dans l'histoire comme point de d\u00e9part du moteur open source le plus fort du monde.<\/span><\/li>\n <li><span class=\"tl-year\">2026<\/span><span class=\"tl-event\">Mesure exp\u00e9rimentale : <strong>Glaurung 2.2 x64, 8 threads, Ryzen 9 7950X3D<\/strong> \u2014 performance de <strong>3163 \u00b1 23 Elo<\/strong>. 50% exact sur pool \u00e0 3163 \u2014 coh\u00e9rence parfaite avec les projections NPS. Laboratoire Chroniques Software.<\/span><\/li>\n <\/ul>\n <\/div>\n\n \n <div class=\"section-inner\" style=\"padding-top: 0;\">\n <div class=\"quote-box\">\n <p>\"Glaurung 2.2 sur 8 threads du Ryzen 9 7950X3D atteint 3163 Elo \u2014 exactement \u00e0 hauteur de son pool adverse \u00e0 50,0%. Ce r\u00e9sultat parfaitement \u00e9quilibr\u00e9 cache une r\u00e9alit\u00e9 tranch\u00e9e : domin\u00e9 \u00e0 30-33% par Wasp et Rybka \u00e0 hardware \u00e9gal, dominant sans partage Shredder mono-thread et Zappa sur Q6600. La contre-performance face \u00e0 Naum 4.2 sur Q6600 reste l'enseignement algorithmique le plus marquant du gauntlet.\"<\/p>\n <span class=\"quote-src\">\u2014 Laboratoire Chroniques Software, Gauntlet maison 2026<\/span>\n <\/div>\n <div class=\"quote-box\">\n <p>\"Glaurung 2.2 est l'Adam de Stockfish. Le tester en 2026 sur un Ryzen 9 7950X3D, c'est mesurer combien le hardware moderne peut ressusciter un code vieux de 18 ans \u2014 la r\u00e9ponse est 3163 Elo, le niveau d'un moteur commercial de premier plan de son \u00e9poque, port\u00e9 par la puissance brute d'un processeur que ses cr\u00e9ateurs n'auraient jamais imagin\u00e9. Et contrairement \u00e0 Wasp, Glaurung y arrive par la seule voie du NPS brut, sans le moindre bonus V-Cache.\"<\/p>\n <span class=\"quote-src\">\u2014 Chroniques Software, Analyse comparative Glaurung & h\u00e9ritage Stockfish, 2026<\/span>\n <\/div>\n <\/div>\n\n <div class=\"footer-note\">\n Exp\u00e9rience \"Chroniques Software\" \u2014 Laboratoire 2026 \u2014 Glaurung 2.2 x64 (8 threads) sur AMD Ryzen 9 7950X3D \u2014 Performance confirm\u00e9e : <strong>3163 \u00b1 23 Elo<\/strong> (75 parties, \u03c3 = 400 \u00d7 \u221a(p\u00b7(1\u2212p)\/N)). Donn\u00e9es SSDF : liste officielle SSDF.\n <\/div>\n\n <\/div>\n<\/div>\n<\/body>\n<\/html>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Glaurung 2.2 x64, anc\u00eatre direct de Stockfish, r\u00e9alise une performance de 3163 \u00b1 23 Elo sur 8 threads du Ryzen 9 7950X3D \u2014 soit +168 points par rapport \u00e0 sa cote SSDF officielle sur Q6600 4 CPU. Sur 75 parties, il affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%, dominant les moteurs sur hardware inf\u00e9rieur (76-80%) mais nettement distanc\u00e9 par Wasp et Rybka \u00e0 hardware \u00e9gal (30-33%). Contrairement \u00e0 Wasp, dont le gain exceptionnel sur ce processeur sugg\u00e8re un fort effet V-Cache 3D, Glaurung suit une loi d’\u00e9chelle NPS strictement classique.<\/p>\n","protected":false},"author":1,"featured_media":3496,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"","neve_meta_content_width":0,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":""},"categories":[37],"tags":[],"rttpg_featured_image_url":{"full":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"landscape":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"portraits":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"thumbnail":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-150x150.png",150,150,true],"medium":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-300x249.png",300,249,true],"large":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-1024x848.png",1024,848,true],"1536x1536":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"2048x2048":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e.png",1136,941,false],"neve-blog":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_5h5eb65h5eb65h5e-930x620.png",930,620,true]},"rttpg_author":{"display_name":"valterd","author_link":"https:\/\/www.chess.lumn.eu\/wordpress\/author\/valterd\/"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/www.chess.lumn.eu\/wordpress\/category\/soft\/\" rel=\"category tag\">Soft<\/a>","rttpg_excerpt":"Glaurung 2.2 x64, anc\u00eatre direct de Stockfish, r\u00e9alise une performance de 3163 \u00b1 23 Elo sur 8 threads du Ryzen 9 7950X3D \u2014 soit +168 points par rapport \u00e0 sa cote SSDF officielle sur Q6600 4 CPU. Sur 75 parties, il affiche un bilan parfaitement \u00e9quilibr\u00e9 \u00e0 50%, dominant les moteurs sur hardware inf\u00e9rieur (76-80%)…","_links":{"self":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495"}],"collection":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/comments?post=3495"}],"version-history":[{"count":4,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495\/revisions"}],"predecessor-version":[{"id":3501,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3495\/revisions\/3501"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media\/3496"}],"wp:attachment":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media?parent=3495"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/categories?post=3495"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/tags?post=3495"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}