Calibration Deep Junior 7 \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 .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 .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 .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 .elo-badge.revised { background: #4caf50; }\n .stat-line { font-size: 0.9em; margin: 5px 0; color: #bbb; }\n .highlight { color: #fff; font-weight: bold; }\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 .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; }\n .ssdf-table td { padding: 6px 10px; border-bottom: 1px solid #2e2e2e; color: #ccc; }\n .ssdf-table .peak-row td { color: #D4AF37; font-weight: bold; }\n .ssdf-table .peak-row td:first-child::before { content: \"\u2605 \"; }\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; }\n .recap-table th:first-child { text-align: left; }\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.total-row td { background: #1d1a10; color: #D4AF37; font-weight: bold; border-top: 2px solid #D4AF37; }\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 .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 .match-box { background: #1a1a14; border: 1px solid #D4AF37; border-radius: 8px; padding: 16px 20px; margin-top: 10px; }\n .match-box .match-header { display: flex; justify-content: space-between; align-items: center; margin-bottom: 10px; }\n .match-box .match-title { color: #D4AF37; font-weight: bold; font-size: 1em; }\n .match-box .match-score { font-size: 1.6em; font-weight: bold; color: #fff; font-family: 'Consolas', monospace; }\n .match-box .match-score .draw { color: #D4AF37; }\n .match-games { width: 100%; border-collapse: collapse; font-size: 0.8em; }\n .match-games th { color: #888; padding: 4px 8px; text-align: left; border-bottom: 1px solid #333; }\n .match-games td { padding: 5px 8px; border-bottom: 1px solid #222; color: #bbb; }\n .match-games .win-jr { color: #4caf50; font-weight: bold; }\n .match-games .win-kas { color: #e57373; font-weight: bold; }\n .match-games .draw-g { color: #aaa; }\n .match-context { font-size: 0.8em; color: #888; margin-top: 10px; font-style: italic; }\n .wccc-timeline { display: flex; flex-wrap: wrap; gap: 10px; margin-top: 10px; }\n .wccc-item { background: #1c1c1c; border: 1px solid #444; border-radius: 6px; padding: 10px 14px; flex: 1; min-width: 130px; text-align: center; }\n .wccc-item .wccc-year { color: #D4AF37; font-weight: bold; font-size: 0.9em; }\n .wccc-item .wccc-result { color: #fff; font-size: 1.1em; font-weight: bold; margin: 3px 0; }\n .wccc-item .wccc-place { color: #888; font-size: 0.75em; }\n .wccc-item.gold { border-color: #D4AF37; background: #1d1a10; }\n .wccc-item.gold .wccc-result { color: #D4AF37; }\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 .section-inner { padding: 0 20px 20px 20px; }\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 .alert-bar { background: #1a0a0a; border: 1px solid #c0392b; border-left: 4px solid #c0392b; border-radius: 6px; padding: 14px 18px; margin: 12px 0; }\n .alert-bar .alert-title { color: #e74c3c; font-weight: bold; font-size: 0.85em; text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 6px; }\n .alert-bar p { color: #aaa; font-size: 0.84em; margin: 0; }\n .note-bar { background: #0a1020; border: 1px solid #2980b9; border-left: 4px solid #2980b9; border-radius: 6px; padding: 14px 18px; margin: 12px 0; }\n .note-bar p { color: #aaa; font-size: 0.84em; margin: 0; }\n .win { color: #4caf50; }\n .loss { color: #e05555; }\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.high { background: #4caf50; }\n .pct-bar.mid { background: #ffb74d; }\n .pct-bar.low { background: #e57373; }\n <\/style>\n\n <div class=\"gauntlet-wrapper\">\n\n <div class=\"gauntlet-header\">\n <h1>Calibration : Deep Junior 7 (8 Threads)<\/h1>\n <p style=\"color: #888; margin-top: 10px;\">Recalibration 2026 \u00b7 AMD Ryzen 9 7950X3D \u00b7 56 parties directes \u00b7 Ancres r\u00e9vis\u00e9es<\/p>\n <\/div>\n\n \n <div class=\"section-title\">Le Pr\u00e9tendant Isra\u00e9lien \u2014 8 Threads<\/div>\n <div class=\"main-grid\">\n <div class=\"engine-card hero\">\n <div class=\"elo-badge revised\">3031 Elo (r\u00e9vis\u00e9)<\/div>\n <h2 style=\"margin: 0; color: #D4AF37;\">Deep Junior 7 \u2014 Amir Ban & Shay Bushinsky<\/h2>\n <div class=\"stat-line\">Hardware : <span class=\"highlight\">AMD Ryzen 9 7950X3D \u00b7 8 threads<\/span><\/div>\n <div class=\"stat-line\">Type \u00e9valuation : <span class=\"highlight\">HCE (\u00e9valuation manuscrite)<\/span><\/div>\n <div class=\"stat-line\">Gauntlet direct : <span class=\"highlight\">56 parties \u00b7 \u03c3 = \u00b111<\/span><\/div>\n <div class=\"stat-line\">Ancien rating (pool mixte BES\/Ryzen) : <span style=\"color:#666;text-decoration:line-through;\">2996<\/span><\/div>\n <p style=\"font-size: 0.85em; color: #D4AF37; font-style: italic; margin-top: 10px;\">\n La recalibration contre quatre ancres recalibr\u00e9es du laboratoire r\u00e9v\u00e8le un rating de 3031 \u2014 35 points au-dessus de la mesure initiale \u00e9tablie via le pool mixte BES\/Ryzen de 2025. Deep Junior 7 (3031) se positionne d\u00e9sormais \u00e0 6 points seulement de Deep Sjeng 1.6 (3037), les deux moteurs 8 threads \u00e9tant statistiquement indiscernables dans cette zone. L'\u00e9cart avec Junior 7 mono-thread (2980) \u2014 seulement 51 points \u2014 illustre l'inefficacit\u00e9 structurelle du parall\u00e9lisme dans la recherche s\u00e9lective de la s\u00e9rie Junior.\n <\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-inner\" style=\"padding-top:0;\">\n <div class=\"alert-bar\">\n <div class=\"alert-title\">Rating r\u00e9vis\u00e9 \u2014 +35 points<\/div>\n <p>Le rating initial de <b>2996<\/b> avait \u00e9t\u00e9 \u00e9tabli contre un pool mixte combinant ancres BES (Athlon 1200 MHz simul\u00e9) et ancres Ryzen. Le gauntlet de recalibration directe contre quatre ancres natives Ryzen \u2014 toutes recalibr\u00e9es dans la vague de corrections 2026 \u2014 produit <b>3031 \u00b111<\/b> sur 56 parties. Les ancres Junior 7 et Junior 8 (50% exacts sur 14 parties chacun) sont les points de r\u00e9f\u00e9rence les plus solides.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Gauntlet Initial \u2014 Pool Mixte BES\/Ryzen (Elo : 2996)<\/div>\n <div class=\"section-inner\">\n <table class=\"recap-table\">\n <thead>\n <tr><th>Opposition (Deep Junior 7 8CPU \u2013 Ancre)<\/th><th>Ancre Elo<\/th><th>Hardware ancre<\/th><th>Score<\/th><th>Parties<\/th><th>%<\/th><\/tr>\n <\/thead>\n <tbody>\n <tr><td>Deep Junior 7 \u2013 Zappa Mexico II x64<\/td><td>3059<\/td><td>Q6600 4T \u00b7 Ryzen<\/td><td>9.5 \u2013 2.5<\/td><td>12<\/td><td class=\"win\">79.2%<\/td><\/tr>\n <tr><td>Deep Junior 7 \u2013 Rybka 2.3.1<\/td><td>2920<\/td><td>Athlon 1.2 GHz (BES)<\/td><td>8.0 \u2013 4.0<\/td><td>12<\/td><td class=\"win\">66.7%<\/td><\/tr>\n <tr><td>Deep Junior 7 \u2013 Fruit 2.3.1<\/td><td>2861<\/td><td>Athlon 1.2 GHz (BES)<\/td><td>7.5 \u2013 4.5<\/td><td>12<\/td><td class=\"win\">62.5%<\/td><\/tr>\n <tr><td>Deep Junior 7 \u2013 Fritz 6<\/td><td>2947<\/td><td>Ryzen 9 7950X3D<\/td><td>7.5 \u2013 4.5<\/td><td>12<\/td><td class=\"win\">62.5%<\/td><\/tr>\n <tr><td>Deep Junior 7 \u2013 Shredder 10<\/td><td>2825<\/td><td>Athlon 1.2 GHz (BES)<\/td><td>7.5 \u2013 4.5<\/td><td>12<\/td><td class=\"win\">62.5%<\/td><\/tr>\n <tr><td>Deep Junior 7 \u2013 Wasp 2<\/td><td>2956<\/td><td>Athlon 1.2 GHz (BES)<\/td><td>3.0 \u2013 9.0<\/td><td>12<\/td><td class=\"loss\">25.0%<\/td><\/tr>\n <tr class=\"total-row\"><td>Total cumul\u00e9<\/td><td>moy. pond. 2928<\/td><td>Pool mixte<\/td><td>43 \u2013 29<\/td><td>72<\/td><td>59.7%<\/td><\/tr>\n <\/tbody>\n <\/table>\n <div class=\"note-bar\" style=\"margin-top:14px;\">\n <p>Performance calcul\u00e9e : 2928 + 400 \u00d7 log\u2081\u2080(0.597 \/ 0.403) \u2248 <b>2996<\/b>. Ce gauntlet m\u00e9lange des ancres natives Ryzen (Zappa, Fritz 6) et des ancres simul\u00e9es via BES (Rybka, Fruit, Shredder 10, Wasp 2) \u2014 m\u00e9thode qui peut introduire un biais structurel selon la sensibilit\u00e9 de chaque moteur au profil cache\/IPC du Ryzen vs Athlon r\u00e9el.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Gauntlet de Recalibration 2026 \u2014 Ancres Directes (Elo : 3031)<\/div>\n <div class=\"section-inner\">\n <table class=\"recap-table\">\n <thead>\n <tr><th>Opposition (Deep Junior 7 8CPU \u2013 Ancre 1CPU)<\/th><th>Ancre Elo<\/th><th>Score<\/th><th>Parties<\/th><th>%<\/th><th>Perf ind.<\/th><\/tr>\n <\/thead>\n <tbody>\n <tr><td>Deep Junior 7 8CPU \u2013 Fritz 6 1CPU<\/td><td>2905<\/td><td>10.5 \u2013 3.5<\/td><td>14<\/td><td class=\"win\">75.00%<\/td><td>3096<\/td><\/tr>\n <tr><td>Deep Junior 7 8CPU \u2013 List 5.12 1CPU<\/td><td>2965<\/td><td>9.0 \u2013 5.0<\/td><td>14<\/td><td class=\"win\">64.29%<\/td><td>3067<\/td><\/tr>\n <tr><td>Deep Junior 7 8CPU \u2013 Junior 7 1CPU<\/td><td>2980<\/td><td>7.0 \u2013 7.0<\/td><td>14<\/td><td>50.00%<\/td><td>2980<\/td><\/tr>\n <tr><td>Deep Junior 7 8CPU \u2013 Junior 8 1CPU<\/td><td>2999<\/td><td>7.0 \u2013 7.0<\/td><td>14<\/td><td>50.00%<\/td><td>2999<\/td><\/tr>\n <tr class=\"total-row\"><td>Total cumul\u00e9<\/td><td>moy. pond. 2962<\/td><td>33.5 \u2013 22.5<\/td><td>56<\/td><td>59.82%<\/td><td><b>3031<\/b><\/td><\/tr>\n <\/tbody>\n <\/table>\n\n <div class=\"perf-summary\">\n <div class=\"perf-box\"><div class=\"perf-label\">Score Total<\/div><div class=\"perf-value\">33.5 \u2013 22.5<\/div><div class=\"perf-sub\">56 parties<\/div><\/div>\n <div class=\"perf-box\"><div class=\"perf-label\">% de gain<\/div><div class=\"perf-value\">59.82%<\/div><div class=\"perf-sub\">pool 100% Ryzen<\/div><\/div>\n <div class=\"perf-box\"><div class=\"perf-label\">Moy. pond. ancres<\/div><div class=\"perf-value\">2962<\/div><div class=\"perf-sub\">4 ancres recalibr\u00e9es<\/div><\/div>\n <div class=\"perf-box gold-box\"><div class=\"perf-label\">Performance Elo<\/div><div class=\"perf-value\">3031<\/div><div class=\"perf-sub\">\u03c3 = \u00b111 \u00b7 rang 23<\/div><\/div>\n <\/div>\n\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\">Note m\u00e9thodologique :<\/strong> Les quatre ancres sont toutes recalibr\u00e9es dans la vague de corrections 2026 (Junior 7 +118, Fritz 6 \u221242, List 5.12 +29, Junior 8 +7). Les deux 50% exacts contre Junior 7 (2980) et Junior 8 (2999) sur 14 parties chacun sont les points de r\u00e9f\u00e9rence les plus robustes du gauntlet.\n <\/p>\n <\/div>\n\n \n <div class=\"section-title\">M\u00e9thodologie BES \u2014 Limites et Biais<\/div>\n <div class=\"section-inner\">\n <div class=\"innov-grid\">\n <div class=\"innov-card\">\n <div class=\"innov-title\">Qu'est-ce que BES ?<\/div>\n <p>Battle Encoder Shirase est un outil Windows qui ralentit un processus en le suspendant et le reprenant \u00e0 intervalles r\u00e9guliers. Utilis\u00e9 pour faire tourner le Ryzen 9 7950X3D \u00e0 la vitesse d'un Athlon 1200 MHz, permettant une comparaison directe avec les listes SSDF historiques.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">La limite structurelle<\/div>\n <p>BES r\u00e9duit la fr\u00e9quence effective par suspension, non par r\u00e9duction r\u00e9elle. L'IPC, le cache L3 (32 MB sur Zen 4 vs ~256 KB sur Athlon) et la bande passante m\u00e9moire restent ceux du Ryzen. Ces param\u00e8tres avantagent ou d\u00e9savantagent diff\u00e9remment chaque moteur par rapport \u00e0 un Athlon r\u00e9el.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Impact sur Deep Junior 7<\/div>\n <p>L'\u00e9cart entre la mesure BES (2996) et la mesure directe (3031) est de +35 points \u2014 soit 3\u03c3. La mesure BES sous-estimait l\u00e9g\u00e8rement le niveau de Deep Junior 7. Le gauntlet direct sur ancres Ryzen-natif est retenu comme r\u00e9f\u00e9rence d\u00e9finitive.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Parall\u00e9lisme inefficace sur Junior<\/div>\n <p>L'\u00e9cart DJ7 8T (3031) vs J7 1T (2980) = seulement <b>51 points<\/b>. La recherche s\u00e9lective asym\u00e9trique de Junior cr\u00e9e un arbre cibl\u00e9 qui se parall\u00e9lise mal : gain \u2248 \u00d72 en kn\/s (vs \u00d76\u20138 attendus) et +0.5 demi-coup. C'est pourquoi DJ7 et J7 sont quasi-indiscernables dans ce laboratoire.<\/p>\n <\/div>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Classements SSDF Historiques<\/div>\n <div class=\"section-inner\">\n <p style=\"font-size:0.84em; color:#888; margin: 0 0 14px 0;\">Source : Svenska schackdatorf\u00f6reningen (SSDF). Hardware de r\u00e9f\u00e9rence indiqu\u00e9. Junior 7.0 = version mono-thread ; Deep Junior 7 = version multi-processeur.<\/p>\n <table class=\"ssdf-table\">\n <thead><tr><th>Liste SSDF<\/th><th>Rang<\/th><th>Hardware<\/th><th>Elo<\/th><th>Parties<\/th><th>%<\/th><th>Note<\/th><\/tr><\/thead>\n <tbody>\n <tr><td>2001-10-02<\/td><td>#5<\/td><td>Athlon 1200 \/ 256 MB<\/td><td>2679<\/td><td>140<\/td><td>59%<\/td><td>Entr\u00e9e sur nouveau HW de pointe (Junior 7.0)<\/td><\/tr>\n <tr><td>2001-10-02<\/td><td>#10<\/td><td>K6-2 450 \/ 128 MB<\/td><td>2624<\/td><td>271<\/td><td>65%<\/td><td>+28 pts vs Junior 6<\/td><\/tr>\n <tr class=\"peak-row\"><td>2001-12-27<\/td><td>#4<\/td><td>Athlon 1200 \/ 256 MB<\/td><td>2681<\/td><td>377<\/td><td>59%<\/td><td>Pic historique \u2014 top 5 mondial<\/td><\/tr>\n <tr><td>2001-12-27<\/td><td>#9<\/td><td>K6-2 450 \/ 128 MB<\/td><td>2643<\/td><td>635<\/td><td>70%<\/td><td>Score de victoires \u00e9lev\u00e9 sur K6<\/td><\/tr>\n <tr><td>2002-12-05<\/td><td>#9<\/td><td>Athlon 1200 \/ 256 MB<\/td><td>2687<\/td><td>623<\/td><td>57%<\/td><td>Tass\u00e9 par Fritz 7 \/ Shredder 6<\/td><\/tr>\n <tr><td>2003-05-26<\/td><td>#12<\/td><td>Athlon 1200 \/ 256 MB<\/td><td>2693<\/td><td>719<\/td><td>56%<\/td><td>Dernier relev\u00e9 SSDF avant Junior 8<\/td><\/tr>\n <\/tbody>\n <\/table>\n <p style=\"font-size:0.78em; color:#666; margin-top: 10px;\">\u2605 = pic de classement. Rang calcul\u00e9 sur l'ensemble du pool SSDF du mois concern\u00e9.<\/p>\n <\/div>\n\n \n <div class=\"section-title\">Palmar\u00e8s WCCC \/ WMCCC<\/div>\n <div class=\"section-inner\">\n <div class=\"wccc-timeline\">\n <div class=\"wccc-item gold\"><div class=\"wccc-year\">1997<\/div><div class=\"wccc-result\">\ud83c\udfc6 1er<\/div><div class=\"wccc-place\">WMCCC Paris<\/div><\/div>\n <div class=\"wccc-item gold\"><div class=\"wccc-year\">2001<\/div><div class=\"wccc-result\">\ud83c\udfc6 1er<\/div><div class=\"wccc-place\">WMCCC Maastricht<br><span style=\"font-size:0.8em\">section multi-processeur<\/span><\/div><\/div>\n <div class=\"wccc-item gold\"><div class=\"wccc-year\">2002<\/div><div class=\"wccc-result\">\ud83c\udfc6 1er<\/div><div class=\"wccc-place\">WCCC Maastricht<\/div><\/div>\n <div class=\"wccc-item gold\"><div class=\"wccc-year\">2004<\/div><div class=\"wccc-result\">\ud83c\udfc6 1er<\/div><div class=\"wccc-place\">WCCC Ramat-Gan<\/div><\/div>\n <div class=\"wccc-item gold\"><div class=\"wccc-year\">2006<\/div><div class=\"wccc-result\">\ud83c\udfc6 1er<\/div><div class=\"wccc-place\">WCCC Turin<br><span style=\"font-size:0.8em\">Junior 10 \u2014 Woodcrest 4 c\u0153urs<\/span><\/div><\/div>\n <div class=\"wccc-item\"><div class=\"wccc-year\">2000<\/div><div class=\"wccc-result\">\ud83e\udd47 1er<\/div><div class=\"wccc-place\">Cadaqu\u00e9s<br><span style=\"font-size:0.8em\">tournoi majeur hors ICGA<\/span><\/div><\/div>\n <\/div>\n <p style=\"font-size:0.78em; color:#666; margin-top:10px;\">Junior remportera encore les WCCC 2009, 2011 et 2013 \u2014 soit 8 titres mondiaux ICGA au total. Deep Junior 7 participait en version multiprocesseur aux WCCC\/WMCCC \u2014 c'est la version 8 threads test\u00e9e dans ce laboratoire.<\/p>\n <\/div>\n\n \n <div class=\"section-title\">Match Kasparov vs Deep Junior \u2014 New York 2003<\/div>\n <div class=\"section-inner\">\n <div class=\"match-box\">\n <div class=\"match-header\">\n <div class=\"match-title\">Man vs Machine WC \u00b7 X3D Technologies \u00b7 Janvier\u2013F\u00e9vrier 2003<\/div>\n <div class=\"match-score\"><span style=\"color:#e57373\">Kasparov<\/span> <span class=\"draw\">3 \u2013 3<\/span> <span style=\"color:#D4AF37\">Deep Junior<\/span><\/div>\n <\/div>\n <table class=\"match-games\">\n <thead><tr><th>Partie<\/th><th>Date<\/th><th>Blancs<\/th><th>R\u00e9sultat<\/th><th>Ouverture<\/th><th>Fait marquant<\/th><\/tr><\/thead>\n <tbody>\n <tr><td>1<\/td><td>26 jan<\/td><td>Kasparov<\/td><td class=\"win-kas\">1 \u2013 0<\/td><td>Semi-Slave<\/td><td>Kasparov domine apr\u00e8s 18 coups \u2014 avantage qualit\u00e9<\/td><\/tr>\n <tr><td>2<\/td><td>28 jan<\/td><td>Deep Junior<\/td><td class=\"draw-g\">\u00bd \u2013 \u00bd<\/td><td>Sicilienne Kan<\/td><td>Kasparov soulage la \"mal\u00e9diction de la partie 2\" vs Deep Blue<\/td><\/tr>\n <tr><td>3<\/td><td>30 jan<\/td><td>Kasparov<\/td><td class=\"win-jr\">0 \u2013 1<\/td><td>Semi-Slave<\/td><td>Kasparov blunde \u2014 Junior profite<\/td><\/tr>\n <tr><td>4<\/td><td>01 f\u00e9v<\/td><td>Deep Junior<\/td><td class=\"draw-g\">\u00bd \u2013 \u00bd<\/td><td>\u2014<\/td><td>Nulle \u00e9quilibr\u00e9e<\/td><\/tr>\n <tr><td>5<\/td><td>05 f\u00e9v<\/td><td>Kasparov<\/td><td class=\"draw-g\">\u00bd \u2013 \u00bd<\/td><td>Nimzo-Indienne<\/td><td style=\"color:#D4AF37;font-weight:bold\">10...Bxh2+! \u2014 sacrifice de fou l\u00e9gendaire. Kasparov : \"move of the match\"<\/td><\/tr>\n <tr><td>6<\/td><td>07 f\u00e9v<\/td><td>Deep Junior<\/td><td class=\"draw-g\">\u00bd \u2013 \u00bd<\/td><td>Sicilienne Najdorf<\/td><td>Kasparov refuse la victoire (23...Rxc3!) \u2014 stupeur du public<\/td><\/tr>\n <\/tbody>\n <\/table>\n <p class=\"match-context\">Match sanctionn\u00e9 par la FIDE et l'ICGA comme \"Man vs Machine World Championship\". Dotation : 1 000 000 $ \u2014 Kasparov 750 000 $, \u00e9quipe Junior 250 000 $. Diffus\u00e9 en direct sur ESPN2. La version jou\u00e9e est Deep Junior 7 en configuration multiprocesseur sur Pentium 4 Xeon.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Innovations Techniques de la S\u00e9rie Junior<\/div>\n <div class=\"section-inner\">\n <div class=\"innov-grid\">\n <div class=\"innov-card\">\n <div class=\"innov-title\">Comptage asym\u00e9trique des coups<\/div>\n <p>Les coups ordinaires comptent pour 2, les coups \"int\u00e9ressants\" pour 1 ou moins dans le budget de recherche. Les variations prometteuses sont explor\u00e9es plus profond\u00e9ment \u2014 g\u00e9n\u00e9ralisation des extensions s\u00e9lectives. Ce m\u00e9canisme est la signature architecturale de toute la s\u00e9rie Junior.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Opponent Modeling<\/div>\n <p>Junior peut jouer des coups exploitant les faiblesses connues de l'adversaire. Une forme primitive d'adaptation dynamique au style de jeu adverse, particuli\u00e8rement efficace contre les humains \u2014 comme Kasparov l'a constat\u00e9 en 2003.<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Style tactique ultra-agressif<\/div>\n <p>Junior est r\u00e9put\u00e9 pour ses attaques de roi impitoyables. IM Hannu Wegner (2001) : \"La derni\u00e8re version de Junior a clairement progress\u00e9 dans la tactique et notamment dans l'ex\u00e9cution d'attaques sur l'aile-roi brillantes.\"<\/p>\n <\/div>\n <div class=\"innov-card\">\n <div class=\"innov-title\">Support multiprocesseur (Deep Junior)<\/div>\n <p>\"Deep Junior\" d\u00e9signe la version multi-c\u0153urs. En 2006 (Turin), optimis\u00e9 pour Intel Woodcrest quad-core non encore commercialis\u00e9, atteignant 9,3 M n\u0153uds\/sec \u2014 +33% vs Reykjavik 2005. La version 7 test\u00e9e dans ce laboratoire tourne en 8 threads sur Ryzen 9 7950X3D.<\/p>\n <\/div>\n <\/div>\n <\/div>\n\n \n <div class=\"section-title\">Chronologie Junior<\/div>\n <div class=\"section-inner\">\n <ul class=\"timeline\">\n <li><span class=\"tl-year\">1993<\/span><span class=\"tl-event\">Amir Ban et Shay Bushinsky d\u00e9butent Junior comme projet amateur, en dehors de leurs heures de travail.<\/span><\/li>\n <li><span class=\"tl-year\">1997<\/span><span class=\"tl-event\"><strong>Commercialisation via ChessBase.<\/strong> Junior remporte le <strong>WMCCC \u00e0 Paris<\/strong>. D\u00e9but de la rivalit\u00e9 avec Shredder.<\/span><\/li>\n <li><span class=\"tl-year\">2000<\/span><span class=\"tl-event\">Victoire au grand tournoi de <strong>Cadaqu\u00e9s<\/strong>. Shredder remporte la WCCC de Londres.<\/span><\/li>\n <li><span class=\"tl-year\">2001<\/span><span class=\"tl-event\"><strong>WMCCC Maastricht<\/strong> \u2014 Junior remporte la section multi-processeur. SSDF oct. 2001 : Junior 7.0 entre \u00e0 la <strong>5e place mondiale<\/strong> (2679 Elo, Athlon 1200 MHz).<\/span><\/li>\n <li><span class=\"tl-year\">Avr. 2001<\/span><span class=\"tl-event\">Brain Games qualifier : Junior m\u00e8ne Fritz <strong>5-0<\/strong> sur 24 parties avant que Fritz \u00e9galise in extremis et gagne le playoff.<\/span><\/li>\n <li><span class=\"tl-year\">2002<\/span><span class=\"tl-event\"><strong>WCCC Maastricht<\/strong> \u2014 Junior champion du monde. Premi\u00e8re au <strong>championnat national isra\u00e9lien de clubs<\/strong>.<\/span><\/li>\n <li><span class=\"tl-year\">Jan. 2003<\/span><span class=\"tl-event\"><strong>Match Kasparov vs Deep Junior<\/strong>, New York \u2014 nul <strong>3-3<\/strong>. 1 M$ de dotation. Kasparov cite <em>10...Bxh2+!<\/em> comme meilleur coup du match. C'est <strong>Deep Junior 7<\/strong> qui dispute ce match.<\/span><\/li>\n <li><span class=\"tl-year\">2004<\/span><span class=\"tl-event\"><strong>WCCC Ramat-Gan<\/strong> (Isra\u00ebl) \u2014 Junior titre mondial \u00e0 domicile.<\/span><\/li>\n <li><span class=\"tl-year\">2005<\/span><span class=\"tl-event\">WCCC Reykjavik \u2014 Junior perd sa couronne au profit de Zappa et Fruit.<\/span><\/li>\n <li><span class=\"tl-year\">2006<\/span><span class=\"tl-event\"><strong>WCCC Turin<\/strong> \u2014 Junior 10 sur Woodcrest 4 c\u0153urs. <strong>5e titre mondial<\/strong>.<\/span><\/li>\n <li><span class=\"tl-year\">2009\u20132013<\/span><span class=\"tl-event\">3 titres WCCC suppl\u00e9mentaires. Total : <strong>8 titres mondiaux ICGA<\/strong> \u2014 record absolu de la s\u00e9rie.<\/span><\/li>\n <\/ul>\n <\/div>\n\n \n <div class=\"section-title\">Profil des Auteurs<\/div>\n <div class=\"bio-section\">\n <div class=\"bio-card\">\n <span class=\"bio-name\">Amir Ban<\/span>\n <p class=\"stat-line\">Champion isra\u00e9lien de math\u00e9matiques olympiques. Co-inventeur, avec Dov Moran, de la <strong>cl\u00e9 USB (\"disc on key\")<\/strong>. La vente des droits \u00e0 SanDisk (2006) lui a procur\u00e9 les ressources pour consacrer du temps \u00e0 Junior. La Troph\u00e9e Shannon (t\u00eate de cheval sculpt\u00e9e, surnomm\u00e9e \"Shanny\") a tr\u00f4n\u00e9 plusieurs ann\u00e9es dans son salon apr\u00e8s chaque titre.<\/p>\n <\/div>\n <div class=\"bio-card\">\n <span class=\"bio-name\">Shay Bushinsky<\/span>\n <p class=\"stat-line\">Co-auteur de Junior depuis les origines. Architecte principal du moteur de recherche et des innovations tactiques. L'innovation cl\u00e9 de Junior r\u00e9side dans son comptage asym\u00e9trique des coups \u2014 les coups int\u00e9ressants comptent moins dans le budget de recherche et sont donc analys\u00e9s plus profond\u00e9ment. Ce m\u00e9canisme explique \u00e0 la fois la force tactique de Junior et la mauvaise parall\u00e9lisation de son arbre de recherche.<\/p>\n <\/div>\n <div class=\"bio-card\" style=\"border-bottom: none;\">\n <span class=\"bio-name\">GM Boris Alterman<\/span>\n <p class=\"stat-line\">Grand-Ma\u00eetre isra\u00e9lien, conseiller \u00e9chiqu\u00e9en de l'\u00e9quipe. Responsable du <strong>livre d'ouvertures<\/strong> de Junior depuis les versions 6 et 7. Auteur de la s\u00e9rie vid\u00e9o \"Alterman Gambit Guide\" chez ChessBase. Sa collaboration avec Ban et Bushinsky s'est traduite notamment par une agressivit\u00e9 accrue des ouvertures \u00e0 partir de Junior 7.<\/p>\n <\/div>\n <\/div>\n\n \n <div class=\"section-inner\" style=\"padding-top: 0;\">\n <div class=\"quote-box\">\n <p>\"Le d\u00e9but de ce mill\u00e9naire a co\u00efncid\u00e9 avec une course brutale entre deux des meilleurs moteurs du monde : Shredder et Deep Junior. Le titre de champion du monde d'informatique d'\u00e9checs \u00e9tait pr\u00e9sent\u00e9 comme une affaire priv\u00e9e entre l'Allemand pr\u00e9cis et solide et l'Isra\u00e9lien plus audacieux et cr\u00e9atif.\"<\/p>\n <span class=\"quote-src\">\u2014 Shay Bushinsky, ChessBase 2006, apr\u00e8s la victoire \u00e0 Turin<\/span>\n <\/div>\n <div class=\"quote-box\">\n <p>\"Kasparov a soulign\u00e9 le style de jeu sp\u00e9culatif et 'intelligent' que Junior a montr\u00e9, en particulier le coup du match, 10...Bxh2+ dans la partie 5.\"<\/p>\n <span class=\"quote-src\">\u2014 ChessBase, rapport de fin de match New York, f\u00e9vrier 2003<\/span>\n <\/div>\n <\/div>\n\n <div class=\"footer-note\">\n Chroniques Software \u2014 Laboratoire 2026 \u2014 Deep Junior 7 \u00b7 8 threads \u00b7 AMD Ryzen 9 7950X3D \u00b7 Recalibration directe \u00b7 3031 \u00b111\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>Deep Junior 7 en configuration 8 CPU obtient une performance de 2996 Elo avec un score de 59,7% sur 72 parties. Il domine largement Zappa Mexico II (79,2%) et Rybka 2.3.1 (66,7%), confirmant l’apport massif du parall\u00e9lisme. Son seul \u00e9chec notable reste face \u00e0 Wasp 2 (25%), mais le r\u00e9sultat global valide sa sup\u00e9riorit\u00e9 face \u00e0 un pool d’opposition moyen \u00e0 2928 Elo.<\/p>\n","protected":false},"author":1,"featured_media":3411,"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_z7zdcvz7zdcvz7zd.png",1120,928,false],"landscape":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd.png",1120,928,false],"portraits":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd.png",1120,928,false],"thumbnail":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd-150x150.png",150,150,true],"medium":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd-300x249.png",300,249,true],"large":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd-1024x848.png",1024,848,true],"1536x1536":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd.png",1120,928,false],"2048x2048":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd.png",1120,928,false],"neve-blog":["https:\/\/www.chess.lumn.eu\/wordpress\/wp-content\/uploads\/2026\/03\/Gemini_Generated_Image_z7zdcvz7zdcvz7zd-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":"Deep Junior 7 en configuration 8 CPU obtient une performance de 2996 Elo avec un score de 59,7% sur 72 parties. Il domine largement Zappa Mexico II (79,2%) et Rybka 2.3.1 (66,7%), confirmant l'apport massif du parall\u00e9lisme. Son seul \u00e9chec notable reste face \u00e0 Wasp 2 (25%), mais le r\u00e9sultat global valide sa sup\u00e9riorit\u00e9 face…","_links":{"self":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3408"}],"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=3408"}],"version-history":[{"count":7,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3408\/revisions"}],"predecessor-version":[{"id":3806,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/posts\/3408\/revisions\/3806"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media\/3411"}],"wp:attachment":[{"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/media?parent=3408"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/categories?post=3408"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.chess.lumn.eu\/wordpress\/wp-json\/wp\/v2\/tags?post=3408"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}