{"id":109717,"date":"2025-10-30T20:08:13","date_gmt":"2025-10-30T20:08:13","guid":{"rendered":"https:\/\/ekamu.net\/?p=109717"},"modified":"2025-10-30T20:08:13","modified_gmt":"2025-10-30T20:08:13","slug":"geleneksel-cipleri-unutun-termodinamik-cipler-enerji-verimliliginde-devrim-yaratacak","status":"publish","type":"post","link":"https:\/\/ekamu.net\/index.php\/2025\/10\/30\/geleneksel-cipleri-unutun-termodinamik-cipler-enerji-verimliliginde-devrim-yaratacak\/","title":{"rendered":"Geleneksel \u00e7ipleri unutun: Termodinamik \u00e7ipler enerji verimlili\u011finde devrim yaratacak"},"content":{"rendered":"

<\/span> Yapay zeka ve bilimsel ara\u015ft\u0131rmalarda devrim yaratmay\u0131 ama\u00e7layan yeni bir teknoloji giri\u015fimi, geleneksel \u00e7ip \u00fcreticilerine meydan okumaya haz\u0131rlan\u0131yor. Extropic<\/strong>, olas\u0131l\u0131ksal bitler<\/strong> (p-bit) kullanan s\u0131ra d\u0131\u015f\u0131 bir \u00e7ip geli\u015ftirdi ve ilk \u00e7al\u0131\u015fan donan\u0131m\u0131n\u0131 \u00fcretmenin yan\u0131 s\u0131ra daha geli\u015fmi\u015f sistemlerin pratik g\u00f6revlerde kullan\u0131labilece\u011fini kan\u0131tlad\u0131. <\/figure>\n

Binlerce katl\u0131k enerji verimlili\u011fi<\/b><\/p>\n

Geleneksel CPU ve GPU<\/strong> mimarilerinden tamamen farkl\u0131 bir yakla\u015f\u0131m<\/strong> benimseyen Extropic\u2019in \u00e7ipleri, \u00f6l\u00e7eklendi\u011finde binlerce kat daha enerji verimli<\/strong> olmay\u0131 vaat ediyor. Bu, g\u00fcn\u00fcm\u00fczdeki s\u00fcrekli artan yapay zeka yat\u0131r\u0131mlar\u0131n\u0131n \u00f6n\u00fcndeki en b\u00fcy\u00fck zorlu\u011fun \u00e7\u00f6z\u00fclmesi anlam\u0131na geliyor. Elbette iddian\u0131n ger\u00e7e\u011fe d\u00f6n\u00fc\u015fmesi gerek ancak AI veri merkezlerinin enerji t\u00fcketimi giderek ciddi bir sorun haline geliyor.<\/p>\n

\u015eirket, bu yeni i\u015flemcilere termodinamik \u00f6rnekleme birimleri<\/strong> (TSU) ad\u0131n\u0131 veriyor. TSU\u2019lar, silikon bile\u015fenler arac\u0131l\u0131\u011f\u0131yla termodinamik elektron dalgalanmalar\u0131n\u0131 kullan\u0131yor ve bunlar\u0131, hava durumu tahmini gibi karma\u015f\u0131k sistemlerin olas\u0131l\u0131klar\u0131n\u0131 ya da g\u00f6r\u00fcnt\u00fc, metin ve video \u00fcretebilen yapay zeka modellerini sim\u00fcle etmek i\u00e7in \u015fekillendiriyor.<\/p>\n

Extropic, ilk \u00e7al\u0131\u015fan \u00e7iplerini s\u0131n\u0131rl\u0131 say\u0131da orta\u011f\u0131yla payla\u015ft\u0131. Bunlar aras\u0131nda ileri d\u00fczey yapay zeka laboratuvarlar\u0131, hava durumu modellemesi \u00fczerine \u00e7al\u0131\u015fan giri\u015fimler ve baz\u0131 h\u00fck\u00fcmet temsilcileri bulunuyor.<\/p>\n

Yeni donan\u0131m\u0131 test edenlerden biri de Atmo giri\u015fiminin CEO\u2019su Johan Mathe. Atmo, savunma bakanl\u0131\u011f\u0131 gibi m\u00fc\u015fteriler i\u00e7in normalden \u00e7ok daha y\u00fcksek \u00e7\u00f6z\u00fcn\u00fcrl\u00fckl\u00fc hava durumu tahminleri sunuyor. Mathe, Extropic\u2019in \u00e7iplerinin farkl\u0131 hava durumu olas\u0131l\u0131klar\u0131n\u0131 \u00e7ok daha verimli hesaplamay\u0131 m\u00fcmk\u00fcn k\u0131laca\u011f\u0131n\u0131<\/strong> vurguluyor. \u015eirket ayr\u0131ca TRHML ad\u0131n\u0131 verdi\u011fi bir yaz\u0131l\u0131m sunuyor. Bu yaz\u0131l\u0131m, Extropic \u00e7iplerinin davran\u0131\u015flar\u0131n\u0131 GPU \u00fczerinde sim\u00fcle etmeye imkan tan\u0131yor. Mathe, hem yaz\u0131l\u0131m hem de ger\u00e7ek \u00e7ipi kullanarak \u201cp-bit\u2019lerin beklenen \u015fekilde davrand\u0131\u011f\u0131n\u0131 g\u00f6zlemledi\u011fini\u201d aktar\u0131yor.<\/p>\n

Extropic\u2019in XTR-0<\/strong> adl\u0131 donan\u0131m\u0131, farkl\u0131 g\u00f6revler i\u00e7in yeniden yap\u0131land\u0131r\u0131labilen bir FPGA \u00e7ip ile i\u00e7inde birka\u00e7 k\u00fcbit<\/strong> bulunan iki adet olas\u0131l\u0131ksal \u00e7ip olan X-0<\/strong>\u2019dan olu\u015fuyor. Geleneksel 1 veya 0<\/strong> de\u011ferine sahip bitler yerine, p-bitler belirsizli\u011fi modelleyebiliyor<\/strong>. \u00d6l\u00e7ek s\u0131n\u0131rl\u0131 olsa da bu \u00e7ip, \u015firketin yakla\u015f\u0131m\u0131n\u0131n potansiyelini g\u00f6steriyor.<\/p>\n

Geleneksel \u00e7iplerden fark\u0131 ne?<\/b><\/p>\n

<\/span> Geleneksel CPU\u2019lar ve GPU\u2019lar, hesaplamalar\u0131 ikili mant\u0131k (binary logic) \u00fczerine kurar. Her bit yaln\u0131zca iki durumdan birinde bulunabilir: 1 (a\u00e7\u0131k) veya 0 (kapal\u0131). Bu yap\u0131, dijital bilgisayarlar\u0131n temelidir ve i\u015flemler mant\u0131ksal do\u011fruluk \u00fczerinden y\u00fcr\u00fcr. Ancak bu sistemler, her i\u015flemi kesin bir sonuca indirgedi\u011fi i\u00e7in \u00f6zellikle yapay zeka gibi olas\u0131l\u0131ksal veya belirsizlik i\u00e7eren s\u00fcre\u00e7leri modellemede do\u011fal bir s\u0131n\u0131ra sahip. <\/figure>\n

Extropic\u2019in geli\u015ftirdi\u011fi TSU ise tamamen farkl\u0131 bir fiziksel ve matematiksel temele dayan\u0131yor. Bu birimlerdeki p-bit (probabilistic bit), klasik bit gibi sabit bir 1 veya 0 de\u011fil. Bu birimler her an iki durum aras\u0131nda olas\u0131l\u0131ksal olarak sal\u0131n\u0131r. Yani bir p-bit, belirli bir anda \u201c%70 olas\u0131l\u0131kla 1, %30 olas\u0131l\u0131kla 0\u201d olabilir. Bu, t\u0131pk\u0131 do\u011fadaki termodinamik dalgalanmalar gibi rastlant\u0131sal ama kontrol edilebilir bir davran\u0131\u015ft\u0131r. TSU\u2019lar bu rastlant\u0131sal hareketleri istatistiksel modeller \u00fcretmek i\u00e7in kullan\u0131r. B\u00f6ylece yapay zeka modelleri, hava durumu tahminleri veya molek\u00fcler sim\u00fclasyonlar gibi belirsizlik temelli sistemleri, klasik i\u015flemcilere g\u00f6re \u00e7ok daha do\u011fal ve enerji verimli bi\u00e7imde modelleyebilir.<\/p>\n

Extropic, arXiv \u00fczerinde yay\u0131nlad\u0131\u011f\u0131 bir makalede, binlerce p-bit i\u00e7eren daha b\u00fcy\u00fck bir \u00e7ipin (Z-1) gelecek y\u0131l haz\u0131r olaca\u011f\u0131n\u0131 ve bunun yeni t\u00fcr bir dif\u00fczyon modeli olu\u015fturmak i\u00e7in kullan\u0131labilece\u011fini duyurdu. 250.000 p-bit<\/strong>‘e sahip olaca\u011f\u0131 s\u00f6ylenen Z-1<\/strong>, g\u00f6r\u00fcnt\u00fc ve videolar olu\u015fturmak ve robotlar\u0131n hareketlerini y\u00f6nlendirmek i\u00e7in kullan\u0131lan AI modellerini olu\u015fturmak i\u00e7in kullan\u0131labilir.<\/p>\n\n

https:\/\/www.youtube.com\/embed\/Y28JQzS6TlE<\/span><\/p>\n

Kaynak :\u00a0https:\/\/www.donanimhaber.com\/geleneksel-cipleri-unutun-termodinamik-cipler-geliyor–197951<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

Yapay zeka ve bilimsel ara\u015ft\u0131rmalarda devrim yaratmay\u0131 ama\u00e7layan yeni bir teknoloji giri\u015fimi, geleneksel \u00e7ip \u00fcreticilerine meydan okumaya haz\u0131rlan\u0131yor. Extropic, olas\u0131l\u0131ksal bitler (p-bit) kullanan s\u0131ra d\u0131\u015f\u0131 bir \u00e7ip geli\u015ftirdi ve ilk \u00e7al\u0131\u015fan …<\/p>\n","protected":false},"author":1,"featured_media":109718,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[4245,498,2134,639,421],"class_list":["post-109717","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji","tag-bit","tag-cip","tag-geleneksel","tag-veri","tag-yapay-zeka"],"_links":{"self":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/109717","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=109717"}],"version-history":[{"count":1,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/109717\/revisions"}],"predecessor-version":[{"id":109721,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/109717\/revisions\/109721"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media\/109718"}],"wp:attachment":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media?parent=109717"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/categories?post=109717"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/tags?post=109717"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}