{"id":111110,"date":"2025-11-18T20:28:31","date_gmt":"2025-11-18T20:28:31","guid":{"rendered":"https:\/\/ekamu.net\/?p=111110"},"modified":"2025-11-18T20:28:31","modified_gmt":"2025-11-18T20:28:31","slug":"nvidia-nvqlink-ile-superbilgisayarlarla-kuantum-bilgisayarlari-bir-araya-getiriyor","status":"publish","type":"post","link":"https:\/\/ekamu.net\/index.php\/2025\/11\/18\/nvidia-nvqlink-ile-superbilgisayarlarla-kuantum-bilgisayarlari-bir-araya-getiriyor\/","title":{"rendered":"Nvidia, NVQLink ile s\u00fcperbilgisayarlarla kuantum bilgisayarlar\u0131 bir araya getiriyor"},"content":{"rendered":"

<\/span> Kuantum bilgisayarlar son on y\u0131lda b\u00fcy\u00fck bir ilerleme kaydetmi\u015f olsalar da pratik kullan\u0131m alanlar\u0131 bug\u00fcn h\u00e2l\u00e2 olduk\u00e7a s\u0131n\u0131rl\u0131. Di\u011fer yandan klasik s\u00fcper bilgisayarlar ise devasa veri k\u00fcmelerini i\u015fleyebilme g\u00fcc\u00fcne ra\u011fmen kuantum avantaj\u0131 gerektiren baz\u0131 problemlerle ba\u015f etmekte zorlan\u0131yorlar. Bu da ikisinin g\u00fcc\u00fcn\u00fc bir araya getirecek hibrit mimarilere<\/strong> y\u00f6nelik bir ihtiya\u00e7 ortaya \u00e7\u0131kar\u0131yor. Bu ihtiyac\u0131 kar\u015f\u0131layacak firma ise Nvidia olacak gibi g\u00f6r\u00fcn\u00fcyor.\u00a0NVIDIA\u2019n\u0131n Uluslararas\u0131 Y\u00fcksek Performans Bilgisayar Fuar\u0131’nda duyurdu\u011fu yeni\u00a0NVQLink <\/strong>platformu, bu hibrit mimari aray\u0131\u015f\u0131na \u00e7\u00f6z\u00fcm olmay\u0131 hedefliyor. <\/figure>\n

Nvidia CEO\u2019su Jensen Huang<\/strong>\u2019\u0131n \u201ckuantum ve klasik s\u00fcper bilgisayarlar\u0131 birbirine ba\u011flayan Rosetta Ta\u015f\u0131\u201d diye nitelendirdi\u011fi NVQLink, kuantum i\u015flemcileri y\u00fcksek performansl\u0131 GPU\u2019larla do\u011frudan konu\u015fturmay\u0131 sa\u011flayan<\/strong> a\u00e7\u0131k bir sistem mimarisi olarak tasarland\u0131. \u015eimdiden d\u00fcnya \u00e7ap\u0131nda bir d\u00fczineden fazla s\u00fcper bilgisayar merkezi bu mimariyi benimsemeye haz\u0131rlan\u0131yor. Platformun en dikkat \u00e7ekici entegrasyonlar\u0131ndan biri ise Quantinuum<\/strong>\u2019un yeni duyurdu\u011fu Helios<\/strong> kuantum bilgisayar\u0131yla olacak.\u00a0Quantinuum, gelecekteki t\u00fcm sistemlerinde NVIDIA GPU\u2019lar\u0131yla NVQLink \u00fczerinden hibrit i\u015f ak\u0131\u015flar\u0131n\u0131 destekleyece\u011fini a\u00e7\u0131klad\u0131.<\/p>\n

Nvidia-Quantinuum \u0130\u015f Birli\u011fi, Hata D\u00fczeltmede Rekor H\u0131z Sa\u011flad\u0131<\/b><\/p>\n

Nvidia ve Quantinuum yaln\u0131zca yeni bir ortakl\u0131\u011fa imza atmakla kalmad\u0131; ayn\u0131 zamanda kuantum hata d\u00fczeltme konusunda bug\u00fcne kadarki en etkileyici teknik sonu\u00e7lardan birini de payla\u015ft\u0131. \u0130ki ekip, NVIDIA GH200 Grace Hopper i\u015flemcilerini Helios sistemine ba\u011flayarak<\/strong> \u00f6l\u00e7eklenebilir kuantum LDPC (d\u00fc\u015f\u00fck yo\u011funluklu e\u015flik kontrol\u00fc)\u00a0kodlar\u0131n\u0131n ger\u00e7ek zamanl\u0131 \u00e7\u00f6z\u00fcmlemesini ger\u00e7ekle\u015ftirdi. Bu i\u015flem s\u0131ras\u0131nda hata d\u00fczeltme d\u00f6ng\u00fcs\u00fcnde gereksinim olan 2 milisaniyelik e\u015fi\u011fin \u00e7ok alt\u0131na inilerek 67 mikrosaniyelik<\/strong> tepki s\u00fcresi elde edildi. Bu, \u015fimdiye kadar alan\u0131n ula\u015ft\u0131\u011f\u0131 en h\u0131zl\u0131 ger\u00e7ek zamanl\u0131 hata d\u00fczeltme performans\u0131 olarak kayda ge\u00e7ti.<\/strong><\/p>\n

<\/span> Ek olarak, Helios kontrol motoruna entegre edilen NVIDIA GPU tabanl\u0131 \u00e7\u00f6z\u00fcmleyici, ger\u00e7ekle\u015ftirilen deneylerde i\u015flemlerin mant\u0131ksal do\u011frulu\u011funu %3\u2019\u00fcn \u00fczerinde<\/strong> art\u0131rd\u0131. \u00d6rne\u011fin Bring\u2019s kodu kullan\u0131larak sekiz mant\u0131ksal kubitin 30 fiziksel kubite kodland\u0131\u011f\u0131 senaryoda, \u00fc\u00e7 d\u00fczeltme turu sonunda hata oran\u0131 %4,95\u2019ten %0,925\u2019e<\/strong> d\u00fc\u015ft\u00fc. Bu da yakla\u015f\u0131k 5,4 katl\u0131k bir iyile\u015fme<\/strong> anlam\u0131na geliyor ve kuantum bilgisayarlar\u0131n pratik uygulamalara yakla\u015fmas\u0131 a\u00e7\u0131s\u0131ndan \u00f6nemli bir e\u015fik olarak de\u011ferlendiriliyor. <\/figure>\n

Elde edilen performans kazan\u0131mlar\u0131n\u0131n arkas\u0131nda NVQLink\u2019in sundu\u011fu teknik altyap\u0131 yat\u0131yor. Sistem, yapay zek\u00e2 i\u015flemlerinde kullan\u0131lan FP4 hassasiyetinde toplam 40 petaflop i\u015flem g\u00fcc\u00fc sunuyor; bu, saniyede 40 katrilyon i\u015flem<\/strong> anlam\u0131na geliyor ve GPU\u2019lar\u0131n kuantum bilgisayardan gelen veriyi neredeyse an\u0131nda i\u015fleyebilmesini sa\u011fl\u0131yor.<\/strong><\/p>\n

Ayr\u0131ca GPU ile kuantum i\u015flemci (QPU) aras\u0131ndaki veri hatt\u0131<\/strong>, 400 gigabit\/saniyelik<\/strong> olduk\u00e7a y\u00fcksek bir bant geni\u015fli\u011fine sahip. Bu sayede iki sistem aras\u0131nda devasa miktarda verinin \u00e7ok h\u0131zl\u0131 aktar\u0131lmas\u0131 m\u00fcmk\u00fcn oluyor. Veri iletimindeki gecikme s\u00fcresi 4 mikrosaniyenin alt\u0131nda<\/strong>, yani milyonda d\u00f6rt saniyeden daha k\u0131sa. Bu d\u00fczeyde d\u00fc\u015f\u00fck gecikme, ger\u00e7ek zamanl\u0131 hata d\u00fczeltme gibi milisaniyelik hassasiyet isteyen uygulamalar i\u00e7in kritik \u00f6neme sahip.<\/p>\n

Bunun yan\u0131nda NVQLink\u2019in RDMA over Converged Ethernet<\/strong> (bir t\u00fcr geli\u015fmi\u015f, do\u011frudan bellek eri\u015fimi sa\u011flayan a\u011f teknolojisi) kullanmas\u0131, ara\u015ft\u0131rmac\u0131lar\u0131n klasik GPU kaynaklar\u0131n\u0131 sistem b\u00fcy\u00fcd\u00fck\u00e7e kolayca \u00f6l\u00e7ekleyebilmesine imk\u00e2n tan\u0131yor. Yani kuantum i\u015flemciler daha fazla k\u00fcbit ve daha karma\u015f\u0131k devrelerle geni\u015fledik\u00e7e, klasik bilgisayar taraf\u0131ndaki GPU g\u00fcc\u00fc de ayn\u0131 h\u0131zda art\u0131r\u0131labiliyor. B\u00f6ylece hibrit kuantum\u2013klasik mimari, donan\u0131m b\u00fcy\u00fcd\u00fck\u00e7e performans kaybetmeden \u00e7al\u0131\u015fmaya devam ediyor.<\/p>\n

Nvidia ve Quantinuum\u2019un ortaya koydu\u011fu sonu\u00e7lar, uzun zamanda teori a\u015famas\u0131nda kalan bu hibrit yakla\u015f\u0131m\u0131n art\u0131k laboratuvar deneylerinden \u00e7\u0131k\u0131p kurumsal ve bilimsel altyap\u0131lara entegre olmaya haz\u0131r h\u00e2le geldi\u011fini g\u00f6steriyor. E\u011fer bu e\u011filim ayn\u0131 h\u0131zla devam ederse, kuantum bilgisayarlar\u0131n pratik de\u011fer \u00fcretme s\u00fcreci beklenenden \u00e7ok daha erken ger\u00e7ekle\u015febilir.<\/p>\n\n

<\/span><\/p>\n

Kaynak :\u00a0https:\/\/www.donanimhaber.com\/nvidia-nvqlink-kuantum-bilgisayarlara-gpu-entegrasyonu-sagliyor–198701<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

Kuantum bilgisayarlar son on y\u0131lda b\u00fcy\u00fck bir ilerleme kaydetmi\u015f olsalar da pratik kullan\u0131m alanlar\u0131 bug\u00fcn h\u00e2l\u00e2 olduk\u00e7a s\u0131n\u0131rl\u0131. Di\u011fer yandan klasik s\u00fcper bilgisayarlar ise devasa veri k\u00fcmelerini i\u015fleyebilme g\u00fcc\u00fcne ra\u011fmen kuantum avantaj\u0131 …<\/p>\n","protected":false},"author":1,"featured_media":111111,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[1923,523,90,2840,641],"class_list":["post-111110","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji","tag-gercek","tag-gpu","tag-is","tag-kuantum","tag-sistem"],"_links":{"self":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111110","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/comments?post=111110"}],"version-history":[{"count":1,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111110\/revisions"}],"predecessor-version":[{"id":111114,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111110\/revisions\/111114"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media\/111111"}],"wp:attachment":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media?parent=111110"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/categories?post=111110"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/tags?post=111110"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}