{"id":37370,"date":"2019-10-31T22:17:15","date_gmt":"2019-10-31T19:17:15","guid":{"rendered":"https:\/\/prohoster.info\/blog\/cerebras-protsessor-dlya-ii-neveroyatnyh-razmerov-i-vozmozhnostej\/"},"modified":"2019-10-31T22:17:15","modified_gmt":"2019-10-31T19:17:15","slug":"cerebras-protsessor-dlya-ii-neveroyatnyh-razmerov-i-vozmozhnostej","status":"publish","type":"post","link":"https:\/\/prohoster.info\/et\/blog\/news\/cerebras-protsessor-dlya-ii-neveroyatnyh-razmerov-i-vozmozhnostej","title":{"rendered":"Cerebras \u2014 ulatuslik ja v\u00f5imekas AI protsessor","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Cerebras protsessori kuulutus \u2015 Cerebras Wafer Scale Engine (WSE) v\u00f5i Cerebrase silikoonplaadi suuruse mootori kuulutamine \u2015 leidis aset<\/a><\/noindex> aastase Hot Chips 31 konverentsi raames. Selle silikooni monstri juures \u00fcllatab mitte ainult see, et see suudeti ellu viia. \u00dcllatav on idee julgust ja arendajate vaeva, kes riskisid 46 225 ruutmillimeetri suuruse kristalli arendamisega, mille k\u00fcljed on 21,5 cm. \u00dche protsessori valmistamiseks kulub terve 300 mm toorik. Isegi v\u00e4ikseima vea korral on tootmisdefekti m\u00e4\u00e4r 100 %, r\u00e4\u00e4kimata hinna kujunemisest, mida on isegi raske ette kujutada.<\/p>\n

\"Cerebras<\/p>\n

Cerebras WSE tootmisprotsessi juhib ettev\u00f5te TSMC. Tehnoloogia on 16 nm FinFET. See taivani tootja on samuti teeninud m\u00e4lestusm\u00e4rgi Cerebrase tootmise eest. Niisuguse kiibi valmistamine n\u00f5udis \u00e4\u00e4rmist meisterlikkust ja hulga probleemide lahendamist, kuid nagu arendajad usuvad, oli see seda v\u00e4\u00e4rt. Cerebrase kiip on praktikas superarvuti kiibil, millel on uskumatud andmeedastuse v\u00f5imed, minimaalne tarbimine ja fantastiline paralleelsus. Praegu on see ideaalne lahendus masin\u00f5ppe jaoks, mis v\u00f5imaldab teadlastel alustada \u00e4\u00e4rmiselt keeruliste probleemide lahendamisega.<\/p>\n

\"Cerebras
\n<\/center><\/p>\n

Iga Cerebras WSE kristall sisaldab 1,2 triljonit transistorit, mis on organiseeritud 400 000 AI-optimeeritud arvutustuuma ja 18 GB kohaliku jaotatud SRAM-m\u00e4lu vahel. K\u00f5ik see on \u00fchendatud rakustruktuuriga, mille kogutootlikkus on 100 petabit sekundis. M\u00e4lu l\u00e4bilaskev\u00f5ime ulatub 9 PB\/s. M\u00e4lu hierarhia on \u00fchekordne. Vahem\u00e4lu puudub, kattuvusi ei ole, juurdep\u00e4\u00e4sul on minimaalne latentsus. See on ideaalne arhitektuur AI-\u00fclesannete kiirendamiseks. Tootlikud numbrid: v\u00f5rreldes k\u00f5ige kaasaegsemate graafikatuumadega pakub Cerebrase kiip 3000 korda suuremat m\u00e4lu mahtu kristallil ja 10 000 korda suuremat m\u00e4lu vahetuskiirust.<\/p>\n

\"Cerebras<\/p>\n

Cerebras'i arvutus tuumad \u2014 SLAC (Sparse Linear Algebra Cores) \u2014 on t\u00e4ielikult programmeeritavad ja neid saab optimeerida t\u00f6\u00f6tamiseks mis tahes n\u00e4rviv\u00f5rkudega. Veelgi enam, tuumade arhitektuur filtreerib algselt andmeid, mis on esitatud nullide kaudu. See vabastab arvutusressursid vajadusest teostada tulutuid nulliga korrutamise operatsioone, mis andmete koormuse osas t\u00e4hendab arvutuste kiiruset\u00f5usu ja \u00e4\u00e4rmist energiaefektiivsust. Seega on Cerebras'i protsessor sadade v\u00f5i isegi tuhandete kordade v\u00f5rra t\u00f5husam masin\u00f5ppe osas, kui vaadata kristalli h\u00f5ivatud pinda ja selle tarbimist, v\u00f5rreldes t\u00e4nap\u00e4evaste lahendustega AI ja masin\u00f5ppe jaoks.<\/p>\n

\"Cerebras<\/p>\n

Sellise suurusega kiibi valmistamine n\u00f5udis<\/a><\/noindex> palju unikaalseid lahendusi. Selle paigutamine korpusesse pidi olema peaaegu k\u00e4sitsi teostatav. Ilmnesid probleemid, kuidas toita kristalli ja selle jahutada. Soojuse juhtimine oli v\u00f5imalik ainult vedelikuga ning see n\u00f5udis piirkondlikku tarnimist koos vertikaalse tsirkulatsiooniga. Sellegipoolest saadi k\u00f5ik probleemid lahendatud ja kiip t\u00f6\u00f6tas. Oleks huvitav teada selle praktilisest rakendusest.<\/p>\n

\"Cerebras<\/p>\n

<\/p>\n

<\/center><\/p>\n

\t\t\t\t


\n\t\t\t\t\t\t\t\t<\/center>
Allikas: 3dnews.ru<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u0410\u043d\u043e\u043d\u0441 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440\u0430 Cerebras \u2015 Cerebras Wafer Scale Engine (WSE) \u0438\u043b\u0438 \u0434\u0432\u0438\u0436\u043a\u0430 Cerebras \u043c\u0430\u0441\u0448\u0442\u0430\u0431\u0430 \u043a\u0440\u0435\u043c\u043d\u0438\u0435\u0432\u043e\u0439 \u043f\u043b\u0430\u0441\u0442\u0438\u043d\u044b \u2015 \u0441\u043e\u0441\u0442\u043e\u044f\u043b\u0441\u044f \u0432 \u0440\u0430\u043c\u043a\u0430\u0445 \u0435\u0436\u0435\u0433\u043e\u0434\u043d\u043e\u0439 \u043a\u043e\u043d\u0444\u0435\u0440\u0435\u043d\u0446\u0438\u0438 Hot Chips 31. \u0413\u043b\u044f\u0434\u044f \u043d\u0430 \u044d\u0442\u043e\u0433\u043e \u043a\u0440\u0435\u043c\u043d\u0438\u0435\u0432\u043e\u0433\u043e \u043c\u043e\u043d\u0441\u0442\u0440\u0430 \u0443\u0434\u0438\u0432\u043b\u044f\u0435\u0442 \u0434\u0430\u0436\u0435 \u043d\u0435 \u0442\u043e, \u0447\u0442\u043e \u0435\u0433\u043e \u0441\u043c\u043e\u0433\u043b\u0438 \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u0442\u044c \u0432\u043e \u043f\u043b\u043e\u0442\u0438. \u0423\u0434\u0438\u0432\u043b\u044f\u0435\u0442 \u0441\u043c\u0435\u043b\u043e\u0441\u0442\u044c \u0437\u0430\u043c\u044b\u0441\u043b\u0430 \u0438 \u0442\u0440\u0443\u0434 \u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432, \u0440\u0438\u0441\u043a\u043d\u0443\u0432\u0448\u0438\u0445 \u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0430\u0442\u044c \u043a\u0440\u0438\u0441\u0442\u0430\u043b\u043b \u043f\u043b\u043e\u0449\u0430\u0434\u044c\u044e 46 225 \u043a\u0432\u0430\u0434\u0440\u0430\u0442\u043d\u044b\u0445 \u043c\u0438\u043b\u043b\u0438\u043c\u0435\u0442\u0440\u043e\u0432 \u0441\u043e \u0441\u0442\u043e\u0440\u043e\u043d\u0430\u043c\u0438 […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":28035,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-37370","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Cerebras \u2014 tohutu ja v\u00f5imekas AI protsessor | ProHoster","description":"Cerebras protsessori kuulutus \u2015 Cerebras Wafer Scale Engine (WSE) v\u00f5i Cerebrase silikoonplaadi suuruse mootori kuulutamine \u2015","canonical_url":"https:\/\/prohoster.info\/et\/blog\/news\/cerebras-protsessor-dlya-ii-neveroyatnyh-razmerov-i-vozmozhnostej","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"et_EE","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47Cerebras \u2015 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440 \u0434\u043b\u044f \u0418\u0418 \u043d\u0435\u0432\u0435\u0440\u043e\u044f\u0442\u043d\u044b\u0445 \u0440\u0430\u0437\u043c\u0435\u0440\u043e\u0432 \u0438 \u0432\u043e\u0437\u043c\u043e\u0436\u043d\u043e\u0441\u0442\u0435\u0439 | ProHoster","og:description":"\u0410\u043d\u043e\u043d\u0441 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440\u0430 Cerebras \u2015 Cerebras Wafer Scale Engine (WSE) \u0438\u043b\u0438 \u0434\u0432\u0438\u0436\u043a\u0430 Cerebras \u043c\u0430\u0441\u0448\u0442\u0430\u0431\u0430 \u043a\u0440\u0435\u043c\u043d\u0438\u0435\u0432\u043e\u0439 \u043f\u043b\u0430\u0441\u0442\u0438\u043d\u044b \u2015","og:url":"https:\/\/prohoster.info\/et\/blog\/news\/cerebras-protsessor-dlya-ii-neveroyatnyh-razmerov-i-vozmozhnostej","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-10-31T19:17:15+00:00","article:modified_time":"2019-10-31T19:17:15+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"37370","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-23 17:33:19","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 12:51:38","updated":"2026-01-23 17:33:19","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts\/37370","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/comments?post=37370"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts\/37370\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/media\/28035"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/media?parent=37370"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/categories?post=37370"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/tags?post=37370"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}