{"id":42117,"date":"2019-03-18T00:00:00","date_gmt":"2019-03-17T21:00:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/blog_prohoster\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam"},"modified":"2020-02-18T13:44:55","modified_gmt":"2020-02-18T10:44:55","slug":"situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam","status":"publish","type":"post","link":"https:\/\/prohoster.info\/hu\/blog\/administrirovanie\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam","title":{"rendered":"A helyzet: a virtu\u00e1lis GPU-k teljes\u00edtm\u00e9nye nem marad el a hardveres megold\u00e1sokt\u00f3l","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"<p>Febru\u00e1rban Stanfordban egy konferenci\u00e1t tartottak a nagy teljes\u00edtm\u00e9ny\u0171 sz\u00e1m\u00edt\u00e1sok (HPC) t\u00e9m\u00e1j\u00e1ban. A VMware k\u00e9pvisel\u0151i elmondt\u00e1k, hogy GPU-val dolgozva a m\u00f3dos\u00edtott ESXi hipervizorral a teljes\u00edtm\u00e9ny nem marad el a bare metal megold\u00e1sokt\u00f3l.<\/p>\n<p>Azokr\u00f3l a technol\u00f3gi\u00e1kr\u00f3l besz\u00e9l\u00fcnk, amelyek lehet\u0151v\u00e9 tett\u00e9k ezt.<\/p>\n<p><noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/company\/it-grad\/blog\/443946\/\"><img decoding=\"async\" alt=\"A helyzet: a virtu\u00e1lis GPU-k teljes\u00edtm\u00e9nye nem marad el a hardveres megold\u00e1sokt\u00f3l\" src=\"\/wp-content\/uploads\/2019\/03\/5618dc11e259e374b3fcf9210bb12b59.jpeg\" style=\"display:block;margin: 0 auto;\" \/><\/a><\/noindex><noindex><a rel=\"nofollow\" name=\"habracut\"><\/a><\/noindex><br \/>\n<i>\/ \u0444\u043e\u0442\u043e <noindex><a rel=\"nofollow\" href=\"https:\/\/commons.wikimedia.org\/wiki\/File:Wikimedia_Foundation_Servers-8055_10.jpg\">Victorgrigas<\/a><\/noindex> <noindex><a rel=\"nofollow\" href=\"https:\/\/creativecommons.org\/licenses\/by-sa\/3.0\/deed.en\">CC BY-SA<\/a><\/noindex><\/i><\/p>\n<h2>Teljes\u00edtm\u00e9nyprobl\u00e9ma<\/h2>\n<p>\nElemz\u0151k becsl\u00e9se szerint a k\u00f6rnyezet adatk\u00f6zpontjaiban a munkaterhel\u00e9sek k\u00f6r\u00fclbel\u00fcl 70%-a <noindex><a rel=\"nofollow\" href=\"https:\/\/www.nextplatform.com\/2019\/02\/27\/making-performance-a-priority-with-virtualized-gpus\/\">virtualiz\u00e1lva van.<\/a><\/noindex>. A fennmarad\u00f3 30% azonban tov\u00e1bbra is bare metal k\u00f6rnyezetben m\u0171k\u00f6dik hipervizorok n\u00e9lk\u00fcl. Ezek a magas terhel\u00e9s\u0171 alkalmaz\u00e1sok nagyr\u00e9szt p\u00e9ld\u00e1ul a neur\u00e1lis h\u00e1l\u00f3zatok tan\u00edt\u00e1s\u00e1hoz kapcsol\u00f3dnak, \u00e9s grafikus processzorokat haszn\u00e1lnak.<\/p>\n<p>Egy ilyen trendet a szak\u00e9rt\u0151k azzal magyar\u00e1zz\u00e1k, hogy a hipervizor mint k\u00f6zb\u00fcls\u0151 absztrakci\u00f3s r\u00e9teg hat\u00e1ssal lehet a rendszer teljes\u00edtm\u00e9ny\u00e9re. \u00d6t \u00e9vvel ezel\u0151tti kutat\u00e1sokban <noindex><a rel=\"nofollow\" href=\"https:\/\/isi.edu\/~mkkang\/papers\/Cloud_2014.pdf\">adatokat tal\u00e1lhatunk<\/a><\/noindex> a sebess\u00e9g cs\u00f6kken\u00e9s\u00e9r\u0151l 10%-kal. Ez\u00e9rt a c\u00e9gek \u00e9s az adatk\u00f6zpontok \u00fczemeltet\u0151i nem sietnek az HPC terhel\u00e9sek virtu\u00e1lis k\u00f6rnyezetbe t\u00f6rt\u00e9n\u0151 \u00e1thelyez\u00e9s\u00e9vel.<\/p>\n<p>De a virtualiz\u00e1ci\u00f3s technol\u00f3gi\u00e1k fejl\u0151dnek \u00e9s javulnak. Egy h\u00f3nappal ezel\u0151tt a VMware konferenci\u00e1j\u00e1n elmondt\u00e1k, hogy az ESXi hipervizor nem gyakorol negat\u00edv hat\u00e1st a GPU teljes\u00edtm\u00e9ny\u00e9re. A sz\u00e1m\u00edt\u00e1si sebess\u00e9g ak\u00e1r h\u00e1rom sz\u00e1zal\u00e9kkal is cs\u00f6kkenhet, ami \u00f6sszehasonl\u00edthat\u00f3 a bare metal teljes\u00edtm\u00e9ny\u00e9vel.<\/p>\n<h2>Hogyan m\u0171k\u00f6dik ez<\/h2>\n<p>\nA HPC rendszerek grafikus processzorokkal val\u00f3 teljes\u00edtm\u00e9ny\u00e9nek fokoz\u00e1sa \u00e9rdek\u00e9ben a VMware v\u00e1ltoztat\u00e1sokat hajtott v\u00e9gre a hipervizor m\u0171k\u00f6d\u00e9s\u00e9ben. K\u00fcl\u00f6n\u00f6sen elt\u00e1vol\u00edtott\u00e1k bel\u0151le a vMotion funkci\u00f3t. Ez sz\u00fcks\u00e9ges a terhel\u00e9s kiegyens\u00falyoz\u00e1s\u00e1hoz, \u00e9s \u00e1ltal\u00e1ban virtu\u00e1lis g\u00e9pek (VM) \u00e1thelyez\u00e9s\u00e9re szolg\u00e1l serverrek vagy GPU-k k\u00f6z\u00f6tt. A vMotion kikapcsol\u00e1sa azt eredm\u00e9nyezte, hogy minden VM mostant\u00f3l egy adott grafikus processzorhoz van hozz\u00e1rendelve. Ez seg\u00edtett cs\u00f6kkenteni az adatcser\u00e9vel kapcsolatos k\u00f6lts\u00e9geket.<\/p>\n<p>A rendszer m\u00e1sik kulcsfontoss\u00e1g\u00fa komponense <noindex><a rel=\"nofollow\" href=\"https:\/\/docs.vmware.com\/en\/VMware-vSphere\/6.7\/com.vmware.vsphere.networking.doc\/GUID-BF2770C3-39ED-4BC5-A8EF-77D55EFE924C.html\">a<\/a><\/noindex> DirectPath I\/O technol\u00f3gia. Ez lehet\u0151v\u00e9 teszi, hogy a CUDA meghajt\u00f3 a p\u00e1rhuzamos sz\u00e1m\u00edt\u00e1sokhoz k\u00f6zvetlen\u00fcl l\u00e9pjen kapcsolatba a virtu\u00e1lis g\u00e9pekkel, megker\u00fclve a hipervizort. Amikor egy GPU-n t\u00f6bb VM-et kell futtatni, a GRID vGPU megold\u00e1s ker\u00fcl aktiv\u00e1l\u00e1sra. Ez a k\u00e1rtya mem\u00f3ri\u00e1j\u00e1t t\u00f6bb szegmensre osztja (de a sz\u00e1m\u00edt\u00e1si ciklusok nem oszlanak meg).<\/p>\n<p>A k\u00e9t virtu\u00e1lis g\u00e9p m\u0171k\u00f6d\u00e9si s\u00e9m\u00e1ja a k\u00f6vetkez\u0151k\u00e9ppen fog kin\u00e9zni:<\/p>\n<p><img decoding=\"async\" alt=\"A helyzet: a virtu\u00e1lis GPU-k teljes\u00edtm\u00e9nye nem marad el a hardveres megold\u00e1sokt\u00f3l\" src=\"\/wp-content\/uploads\/2019\/03\/5e1edef302889fc18b14fe097c6d560b.jpeg\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<h2>Eredm\u00e9nyek \u00e9s el\u0151rejelz\u00e9sek<\/h2>\n<p>\nC\u00e9g <noindex><a rel=\"nofollow\" href=\"http:\/\/on-demand.gputechconf.com\/gtc\/2018\/presentation\/s8249-machine-learning-on-vmware-vsphere-using-nvidias-virtualized-gpus.pdf\">teszteket v\u00e9gzett<\/a><\/noindex> hipervizort, a nyelvi modellt az alapon k\u00e9pzi <noindex><a rel=\"nofollow\" href=\"https:\/\/ru.wikipedia.org\/wiki\/TensorFlow\">TensorFlow<\/a><\/noindex>. A teljes\u00edtm\u00e9ny \"k\u00e1rosod\u00e1sa\" mind\u00f6ssze 3\u20134% volt a bare metalhoz k\u00e9pest. Cser\u00e9be a rendszer lehet\u0151s\u00e9get kapott az er\u0151forr\u00e1sok ig\u00e9ny szerinti eloszt\u00e1s\u00e1ra a pillanatnyi terhel\u00e9sek f\u00fcggv\u00e9ny\u00e9ben.<\/p>\n<p>Az IT \u00f3ri\u00e1s szint\u00e9n <noindex><a rel=\"nofollow\" href=\"https:\/\/blogs.vmware.com\/apps\/2018\/08\/sharing-nvidia-gpus-in-singularity-hpc-containers-leveraging-vsphere.html\">teszteket v\u00e9gzett<\/a><\/noindex> kont\u00e9nerekkel. A c\u00e9g m\u00e9rn\u00f6kei neur\u00e1lis h\u00e1l\u00f3zatokat tan\u00edtottak meg k\u00e9pek felismer\u00e9s\u00e9re. Ennek sor\u00e1n egy GPU er\u0151forr\u00e1sait n\u00e9gy kont\u00e9neres VM k\u00f6z\u00f6tt osztott\u00e1k el. Ennek eredm\u00e9nyek\u00e9nt az egyes g\u00e9pek teljes\u00edtm\u00e9nye 17%-kal cs\u00f6kkent (\u00f6sszehasonl\u00edtva egy VM-mel, amely teljes hozz\u00e1f\u00e9r\u00e9ssel rendelkezik a GPU er\u0151forr\u00e1saihoz). Azonban a feldolgozott k\u00e9pek sz\u00e1ma m\u00e1sodpercenk\u00e9nt <noindex><a rel=\"nofollow\" href=\"http:\/\/3s81si1s5ygj3mzby34dq6qf-wpengine.netdna-ssl.com\/wp-content\/uploads\/2019\/02\/VMware-ML-throughput-results.png\">megn\u00f6vekedett<\/a><\/noindex> h\u00e1romszoros\u00e1ra. Olyan rendszerek v\u00e1rhat\u00f3k, amelyek <noindex><a rel=\"nofollow\" href=\"https:\/\/blogs.vmware.com\/vsphere\/2018\/10\/machine-learning-with-gpus-on-vsphere.html\">haszn\u00e1latban lesznek<\/a><\/noindex> az adatfeldolgoz\u00e1s \u00e9s a sz\u00e1m\u00edt\u00f3g\u00e9pes modellez\u00e9s ter\u00fclet\u00e9n.<\/p>\n<p>A VMware \u00e1ltal tal\u00e1lhat\u00f3 potenci\u00e1lis probl\u00e9m\u00e1k k\u00f6z\u00f6tt a szak\u00e9rt\u0151k <noindex><a rel=\"nofollow\" href=\"https:\/\/www.nextplatform.com\/2019\/02\/27\/making-performance-a-priority-with-virtualized-gpus\/\">kiemelik<\/a><\/noindex> a meglehet\u0151sen sz\u0171k c\u00e9lk\u00f6z\u00f6ns\u00e9get. A nagy teljes\u00edtm\u00e9ny\u0171 rendszerekkel m\u00e9g mindig csak n\u00e9h\u00e1ny c\u00e9g dolgozik. B\u00e1r a Statista <noindex><a rel=\"nofollow\" href=\"https:\/\/www.statista.com\/statistics\/227284\/share-of-global-cloud-data-center-workloads-development\/\">megjegyzi<\/a><\/noindex>, hogy 2021-re m\u00e1r 94%-ra n\u0151tt a virtualiz\u00e1lt glob\u00e1lis adatk\u00f6zpontok munkaterhei. A <noindex><a rel=\"nofollow\" href=\"https:\/\/www.marketsandmarkets.com\/PressReleases\/Quantum-High-Performance-Computing.asp\">el\u0151rejelz\u00e9sek szerint<\/a><\/noindex> az elemz\u0151k arra sz\u00e1m\u00edtanak, hogy a HPC-piac \u00e9rt\u00e9ke 32 milli\u00e1rd doll\u00e1rr\u00f3l 45 milli\u00e1rdra n\u0151 a 2017 \u00e9s 2022 k\u00f6z\u00f6tti id\u0151szakban.<\/p>\n<p><img decoding=\"async\" alt=\"A helyzet: a virtu\u00e1lis GPU-k teljes\u00edtm\u00e9nye nem marad el a hardveres megold\u00e1sokt\u00f3l\" src=\"\/wp-content\/uploads\/2019\/03\/d74d9df5b6e7957aa7b4990408d0eb7f.jpeg\" style=\"display:block;margin: 0 auto;\" \/><br \/>\n<i>\/ \u0444\u043e\u0442\u043e <noindex><a rel=\"nofollow\" href=\"https:\/\/commons.wikimedia.org\/wiki\/File:Inside_Suite.jpg\">Global Access Point<\/a><\/noindex> PD<\/i><\/p>\n<h2>Hasonl\u00f3 megold\u00e1sok<\/h2>\n<p>\nt\u00f6bb hasonl\u00f3, nagy IT c\u00e9gek \u00e1ltal kifejlesztett alternat\u00edv megold\u00e1s tal\u00e1lhat\u00f3 a piacon: AMD \u00e9s Intel.<\/p>\n<p>Az els\u0151 c\u00e9g grafikus feldolgoz\u00f3k virtualiz\u00e1l\u00e1s\u00e1ra szolg\u00e1l\u00f3 megold\u00e1st k\u00edn\u00e1l <noindex>aj\u00e1nlatot tesz<\/noindex> az SR-IOV (single-root input\/output virtualization) alap\u00fa megold\u00e1sra. Ez a technol\u00f3gia lehet\u0151v\u00e9 teszi a VM sz\u00e1m\u00e1ra, hogy a rendszer egyes hardveres lehet\u0151s\u00e9geihez hozz\u00e1f\u00e9rjen. A megold\u00e1s lehet\u0151v\u00e9 teszi, hogy egy GPU-t 16 felhaszn\u00e1l\u00f3 k\u00f6z\u00f6tt osszanak meg egyenletes teljes\u00edtm\u00e9nnyel a virtualiz\u00e1lt rendszerek k\u00f6z\u00f6tt.<\/p>\n<p>A m\u00e1sodik IT \u00f3ri\u00e1s eset\u00e9ben pedig <noindex><a rel=\"nofollow\" href=\"https:\/\/www.intel.com\/content\/www\/us\/en\/solution-providers\/documents\/intel-citrix-xendesktop-virtualized-graphics-brief.html\">a technol\u00f3gia alapja<\/a><\/noindex> a Citrix XenServer 7 hipervizor. Ez egyes\u00edti a standard GPU-illeszt\u0151program \u00e9s a virtu\u00e1lis g\u00e9p m\u0171k\u00f6d\u00e9s\u00e9t, lehet\u0151v\u00e9 t\u00e9ve, hogy a virtu\u00e1lis g\u00e9p 3D-alkalmaz\u00e1sokat \u00e9s asztalokat jelen\u00edtsen meg t\u00f6bb sz\u00e1z felhaszn\u00e1l\u00f3 eszk\u00f6zein.<\/p>\n<h2>A technol\u00f3gia j\u00f6v\u0151je<\/h2>\n<p>\nA virtu\u00e1lis grafikus processzorok fejleszt\u0151i <noindex><a rel=\"nofollow\" href=\"https:\/\/www.networkworld.com\/article\/3197846\/nvidia-gives-glimpse-of-the-future-at-its-gpu-technology-conference.html\">fogadnak<\/a><\/noindex> a mesters\u00e9ges intelligencia rendszerek bevezet\u00e9s\u00e9re \u00e9s a nagy teljes\u00edtm\u00e9ny\u0171 megold\u00e1sok ir\u00e1nti kereslet n\u00f6veked\u00e9s\u00e9re az \u00fczleti technol\u00f3gi\u00e1k piac\u00e1n. Rem\u00e9lik, hogy a nagy adatmennyis\u00e9gek feldolgoz\u00e1s\u00e1nak sz\u00fcks\u00e9gess\u00e9ge n\u00f6velni fogja a vGPU ir\u00e1nti keresletet.<\/p>\n<p>Jelenleg a gy\u00e1rt\u00f3k <noindex><a rel=\"nofollow\" href=\"https:\/\/searchservervirtualization.techtarget.com\/tip\/GPU-virtualization-evolves-with-new-chip-types-on-the-horizon\">m\u00f3dot keresnek<\/a><\/noindex> a CPU \u00e9s GPU funkci\u00f3k egyes\u00edt\u00e9s\u00e9re egy magban, hogy felgyors\u00edts\u00e1k a grafikai feladatok, matematikai sz\u00e1m\u00edt\u00e1sok, logikai m\u0171veletek \u00e9s adatfeldolgoz\u00e1s v\u00e9grehajt\u00e1s\u00e1t. Az ilyen magok megjelen\u00e9se a piacon a j\u00f6v\u0151ben meg fogja v\u00e1ltoztatni a forr\u00e1svirtualiz\u00e1ci\u00f3 \u00e9s az er\u0151forr\u00e1sok munkaterhel\u00e9sek k\u00f6z\u00f6tti eloszt\u00e1s\u00e1nak megk\u00f6zel\u00edt\u00e9s\u00e9t virtu\u00e1lis \u00e9s felh\u0151 alap\u00fa k\u00f6rnyezetben.<\/p>\n<p>\nOlvasnival\u00f3 a t\u00e9m\u00e1ban a c\u00e9ges blogunkban:<\/p>\n<ul>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/iaas-blog.it-grad.ru\/tendencii\/pochemu-korporativnye-zakazchiki-ispolzuyut-virtualnye-mashiny-a-ne-kontejnery\/\">Mi\u00e9rt haszn\u00e1lnak a v\u00e1llalatok virtu\u00e1lis g\u00e9peket, nem pedig kont\u00e9nereket<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/iaas-blog.it-grad.ru\/novosti\/kak-dela-u-vmware-obzor-novyx-reshenij\/\">\u201eHogy \u00e1ll a VMware?\u201d: \u00faj megold\u00e1sok \u00e1ttekint\u00e9se<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/iaas-blog.it-grad.ru\/funkcionalnost\/novaya-funkcionalnost-v-vmware-vsphere-6-7\/\">VMware vSphere 6.7: mik\u00e9nt hasznosak az \u00faj hipervizori funkci\u00f3k<\/a><\/noindex><\/li>\n<\/ul>\n<p>\nP\u00e1r poszt a Telegram-csatorn\u00e1nkr\u00f3l:<\/p>\n<ul>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/t.me\/iaasblog\/187\">A j\u00f6v\u0151 informatikai architekt\u00far\u00e1ja \u2013 ki \u00e9s hogyan k\u00e9sz\u00fcl r\u00e1<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/t.me\/iaasblog\/195\">5 eszk\u00f6z a felh\u0151 kezel\u00e9s\u00e9re<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/t.me\/iaasblog\/194\">Hogyan v\u00e9d egy IaaS-szolg\u00e1ltat\u00f3 az \u00fcgyfelek adatait<\/a><\/noindex><\/li>\n<\/ul>\n<p>Forr\u00e1s: <a content=\"nofollow\" rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/company\/it-grad\/blog\/443946\/\">habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"<p>\u0412 \u0444\u0435\u0432\u0440\u0430\u043b\u0435 \u0432 \u0421\u0442\u044d\u043d\u0444\u043e\u0440\u0434\u0435 \u043f\u0440\u043e\u0448\u043b\u0430 \u043a\u043e\u043d\u0444\u0435\u0440\u0435\u043d\u0446\u0438\u044f, \u043f\u043e\u0441\u0432\u044f\u0449\u0451\u043d\u043d\u0430\u044f \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u043c \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u043c (HPC). \u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u0442\u0435\u043b\u0438 VMware \u0440\u0430\u0441\u0441\u043a\u0430\u0437\u0430\u043b\u0438, \u0447\u0442\u043e \u043f\u0440\u0438 \u0440\u0430\u0431\u043e\u0442\u0435 \u0441 GPU \u0441\u0438\u0441\u0442\u0435\u043c\u0430 \u043d\u0430 \u0431\u0430\u0437\u0435 \u043c\u043e\u0434\u0438\u0444\u0438\u0446\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u043e\u0433\u043e \u0433\u0438\u043f\u0435\u0440\u0432\u0438\u0437\u043e\u0440\u0430 ESXi \u043f\u043e \u0441\u043a\u043e\u0440\u043e\u0441\u0442\u0438 \u043d\u0435 \u0443\u0441\u0442\u0443\u043f\u0430\u0435\u0442 bare metal \u0440\u0435\u0448\u0435\u043d\u0438\u044f\u043c. \u0420\u0430\u0441\u0441\u043a\u0430\u0437\u044b\u0432\u0430\u0435\u043c \u043e \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u044f\u0445, \u043a\u043e\u0442\u043e\u0440\u044b\u0435 \u043f\u043e\u0437\u0432\u043e\u043b\u0438\u043b\u0438 \u044d\u0442\u043e\u0433\u043e \u0434\u043e\u0431\u0438\u0442\u044c\u0441\u044f. \/ \u0444\u043e\u0442\u043e Victorgrigas CC BY-SA \u041f\u0440\u043e\u0431\u043b\u0435\u043c\u0430 \u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u0438 \u041f\u043e \u043e\u0446\u0435\u043d\u043a\u0430\u043c \u0430\u043d\u0430\u043b\u0438\u0442\u0438\u043a\u043e\u0432, \u043f\u043e\u0440\u044f\u0434\u043a\u0430 70% \u0440\u0430\u0431\u043e\u0447\u0438\u0445 \u043d\u0430\u0433\u0440\u0443\u0437\u043e\u043a \u0432 \u0434\u0430\u0442\u0430-\u0446\u0435\u043d\u0442\u0440\u0430\u0445 \u0432\u0438\u0440\u0442\u0443\u0430\u043b\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u044b. [&hellip;]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[688],"tags":[],"class_list":["post-42117","post","type-post","status-publish","format-standard","hentry","category-administrirovanie"],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 5.0.4 - aioseo.com -->\n\t<meta name=\"description\" content=\"\u0412 \u0444\u0435\u0432\u0440\u0430\u043b\u0435 \u0432 \u0421\u0442\u044d\u043d\u0444\u043e\u0440\u0434\u0435 \u043f\u0440\u043e\u0448\u043b\u0430 \u043a\u043e\u043d\u0444\u0435\u0440\u0435\u043d\u0446\u0438\u044f, \u043f\u043e\u0441\u0432\u044f\u0449\u0451\u043d\u043d\u0430\u044f \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u043c \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u043c (HPC).\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"Yuri Gagarin\"\/>\n\t<link rel=\"canonical\" href=\"https:\/\/prohoster.info\/hu\/blog\/administrirovanie\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam\" \/>\n\t\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.4\" \/>\n\t\t<meta property=\"og:locale\" content=\"hu_HU\" \/>\n\t\t<meta property=\"og:site_name\" content=\"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\" \/>\n\t\t<meta property=\"og:type\" content=\"article\" \/>\n\t\t<meta property=\"og:title\" content=\"\ud83e\udd47\u0421\u0438\u0442\u0443\u0430\u0446\u0438\u044f: \u0432\u0438\u0440\u0442\u0443\u0430\u043b\u044c\u043d\u044b\u0435 GPU \u043d\u0435 \u0443\u0441\u0442\u0443\u043f\u0430\u044e\u0442 \u043f\u043e \u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u0438 \u0436\u0435\u043b\u0435\u0437\u043d\u044b\u043c \u0440\u0435\u0448\u0435\u043d\u0438\u044f\u043c | ProHoster\" \/>\n\t\t<meta property=\"og:description\" content=\"\u0412 \u0444\u0435\u0432\u0440\u0430\u043b\u0435 \u0432 \u0421\u0442\u044d\u043d\u0444\u043e\u0440\u0434\u0435 \u043f\u0440\u043e\u0448\u043b\u0430 \u043a\u043e\u043d\u0444\u0435\u0440\u0435\u043d\u0446\u0438\u044f, \u043f\u043e\u0441\u0432\u044f\u0449\u0451\u043d\u043d\u0430\u044f \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u043c \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u043c (HPC).\" \/>\n\t\t<meta property=\"og:url\" content=\"https:\/\/prohoster.info\/hu\/blog\/administrirovanie\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam\" \/>\n\t\t<meta property=\"og:image\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:secure_url\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:width\" content=\"350\" \/>\n\t\t<meta property=\"og:image:height\" content=\"350\" \/>\n\t\t<meta property=\"article:published_time\" content=\"2019-03-17T21:00:00+00:00\" \/>\n\t\t<meta property=\"article:modified_time\" content=\"2020-02-18T10:44:55+00:00\" \/>\n\t\t<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<meta property=\"article:author\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"\ud83e\udd47 Helyzet: a virtu\u00e1lis GPU-k teljes\u00edtm\u00e9nyben nem maradnak el a hardveres megold\u00e1sokt\u00f3l | ProHoster","description":"Febru\u00e1rban konferenci\u00e1t tartottak a Stanfordon, amely a nagy teljes\u00edtm\u00e9ny\u0171 sz\u00e1m\u00edt\u00e1sokkal (HPC) foglalkozott.","canonical_url":"https:\/\/prohoster.info\/hu\/blog\/administrirovanie\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"hu_HU","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\udd47\u0421\u0438\u0442\u0443\u0430\u0446\u0438\u044f: \u0432\u0438\u0440\u0442\u0443\u0430\u043b\u044c\u043d\u044b\u0435 GPU \u043d\u0435 \u0443\u0441\u0442\u0443\u043f\u0430\u044e\u0442 \u043f\u043e \u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u0438 \u0436\u0435\u043b\u0435\u0437\u043d\u044b\u043c \u0440\u0435\u0448\u0435\u043d\u0438\u044f\u043c | ProHoster","og:description":"\u0412 \u0444\u0435\u0432\u0440\u0430\u043b\u0435 \u0432 \u0421\u0442\u044d\u043d\u0444\u043e\u0440\u0434\u0435 \u043f\u0440\u043e\u0448\u043b\u0430 \u043a\u043e\u043d\u0444\u0435\u0440\u0435\u043d\u0446\u0438\u044f, \u043f\u043e\u0441\u0432\u044f\u0449\u0451\u043d\u043d\u0430\u044f \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u043c \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u043c (HPC).","og:url":"https:\/\/prohoster.info\/hu\/blog\/administrirovanie\/situatsiya-virtualnye-gpu-ne-ustupayut-po-proizvoditelnosti-zheleznym-resheniyam","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":"2019-03-17T21:00:00+00:00","article:modified_time":"2020-02-18T10:44:55+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"42117","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":"2026-01-22 07:19:19","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 23:45:28","updated":"2026-01-22 07:19:19","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/posts\/42117","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/comments?post=42117"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/posts\/42117\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/media?parent=42117"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/categories?post=42117"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/hu\/wp-json\/wp\/v2\/tags?post=42117"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}