Kwenzekeni
I-AWS kanye ne-NVIDIA bamemezele ukusebenzisana okunwetshiwe okugxile ekwandiseni umthamo wekhompuyutha we-AI kanye nokuhlanganisa okwengeziwe kwezingxenyekazi zekhompuyutha ze-NVIDIA nesofthiwe enezinsiza ze-AWS. Izinkampani zithi i-AWS ihlela ukufaka izigidi ezi-2 ezengeziwe ze-NVIDIA Blackwell Ultra, i-Rubin ne-Rubin Ultra GPU kuyo yonke ingqalasizinda yayo yomhlaba wonke ngo-2027 nango-2028, okuhlanganisa amafekthri e-AI namandla e-AI ye-ajenti, ukutholwa kwesayensi, i-automation yebhizinisi kanye nomsebenzi we-AI ongokoqobo.
I-AWS kanye ne-NVIDIA bathi banweba ukusebenzisana kweminyaka engu-16 ukuze bahlangabezane nalokho abakuchaza njengokusheshisa ukufunwa kwengqalasizinda ye-AI. Ukuzinikela okumaphakathi kuwuhlelo lwe-AWS lokusebenzisa izigidi ezimbili ze-NVIDIA GPU kuyo yonke ingqalasizinda yayo yomhlaba wonke phakathi no-2027 no-2028. Amasistimu ahleliwe ahlanganisa i-NVIDIA Blackwell Ultra, i-Rubin kanye ne-Rubin Ultra GPUs, enamandla ehloselwe umthwalo wokusebenza ohlanganisa ama-agent we-AI, ukutholwa kwesayensi okuzenzakalelayo, ne-Anterprise. Ngenxa yokuthi isimemezelo sisebenzisa ulimi lwesikhathi esizayo, sichaza ukuthunyelwa okuhleliwe kunokufakwa okuqediwe noma uhlu lwamagama olutholakalayo njengamanje.
Izinkampani ziphinde zathi i-AWS izonweba umthamo wayo we-NVIDIA Blackwell, okuhlanganisa ne-RTX PRO 4500 Blackwell Server Edition GPUs yezimo ze-Amazon EC2 G7. I-NVIDIA kanye ne-AWS zithi izimo ze-G7 ziletha izikhathi ezingu-4.6 zokusebenza okucatshangwayo kwe-AI kanye nokuphinda izikhathi ezingu-2.1 ukusebenza kwezithombe kwezenzakalo ze-G6 zesizukulwane sangaphambilini. Lezo yizimangalo ekukhululweni kwezinkampani; umthombo awuzihlonzi izimo zokuhlola, umthwalo wokusebenza, intengo noma ukuqinisekiswa okuzimele ngemuva kokuqhathanisa. I-AWS ichazwa njengomhlinzeki wokuqala omkhulu wamafu ukunikeza izimo ezisheshiswe yi-RTX PRO 4500, ngokusho kokukhishwa.
Ukusebenzisana okubanzi kuhlanganisa izendlalelo ezimbalwa zesitaki se-AI. I-AWS kanye ne-NVIDIA bathi basebenzela ukuletha ingqalasizinda esekwe ku-NVIDIA Vera CPU ku-AWS, banwebe i-NVLink Fusion ngobuchwepheshe benkumbulo obujwayelekile be-NVIDIA be-high-bandwidth ngokubambisana nabahlinzeki benkumbulo, futhi bahlanganise iplatifomu ye-ZXQAIUsU0 ne-System ZXZXX I-adaptha ye-Elastic Fabric. Bahlela futhi ukuqhubeka nokuhlinzeka ngezinhlobo ezivulekile ze-NVIDIA Nemotron nge-Amazon Bedrock kanye ne-Amazon SageMaker, ukusheshisa ukucutshungulwa kwedatha kanye ne-vector indexing ngokusebenzisa imitapo yolwazi ye-NVIDIA CUDA-X ku-Amazon EMR kanye ne-ZXQAIU1Q yomsebenzi nge-OpenSeoic Ukusetshenziswa kwamarobhothi e-NVIDIA kwe-AI engokoqobo ye-NVIDIA.
Imininingwane yomthombo: nvidianews.nvidia.com ↗
Kungani kubalulekile
Lesi simemezelo sizonweba ingqalasizinda yefu etholakalayo ekuthuthukisweni nasekusetshenzisweni kwe-AI ngezinga elikhulu, kodwa ngokuyinhloko kuwuhlelo olubheke phambili kunobufakazi bokuthi umthamo usulethiwe. Iphinde ixhumanise ingqalasizinda ye-AI nemithwalo yemisebenzi kahulumeni, ukucutshungulwa kwedatha yebhizinisi namarobhothi, inwebe ubudlelwano ngaphandle kwezimo ze-GPU kuphela.
Uma ilethwa, umthamo owengeziwe ungathinta indlela izinhlangano ezihambisa ngokushesha ngayo amaphrojekthi e-AI kusukela ekuhlolweni kuya ekukhiqizeni. Ukukhishwa kuzimele ngokukhethekile ukunwetshwa okuzungeze amasistimu e-ejenti, okungase kudinge amakholi wamamodeli aphindaphindiwe nengqalasizinda yedatha esekelayo, kanye ne-AI ephathekayo, engadinga ukulingisa, ukuqeqeshwa nokuqinisekisa imithwalo yomsebenzi. Iphinde ikhombe ekusetshenzisweni kwesayensi nebhizinisi, iphakamisa ukuthi izethameli ezihlosiwe zidlulele ngale kwenani elincane lamalabhorethri asemngceleni. Umthombo awuqinisekisi ukuthi ingakanani umthamo ohleliwe ozotholakala kumakhasimende amafu ajwayelekile noma ukuthi ukufinyelela kuzokwabiwa kanjani.
Lesi simemezelo sibalulekile ngoba sihlanganisa ukunikezwa kwezingxenyekazi zekhompuyutha, ukuthunyelwa kwamafu kanye nokuhlanganiswa kwesofthiwe ohlelweni olulodwa lwezohwebo. Amakhasimende e-AWS, empeleni, angakwazi ukusebenzisa i-NVIDIA GPU eceleni kwe-silicon yangokwezifiso ye-AWS kanye nama-CPU e-NVIDIA Vera, kuyilapho i-NVLink Fusion ingaxhuma i-NVIDIA kanye nezingxenye ze-Trainium ngaphakathi kwesakhiwo esivamile se-rack-scale. Izinkampani zichaza lokhu njengendlela yokweseka ingqalasizinda ye-AI ehlukahlukene. Lokho kungase kunikeze onjiniyela izinketho ezengeziwe zokufanisa imithwalo yomsebenzi ehlukene kumaphrosesa ahlukene, kodwa umthombo awunikezi amabhentshimakhi azimele aqhathanisa ukulungiselelwa okuxubile nokunye noma ukuchaza imikhawulo yobunjiniyela yokuhambisa imithwalo yomsebenzi phakathi kwabo.
Uphiko lukahulumeni lukhuphula izinga lenzuzo yomphakathi. I-AWS kanye ne-NVIDIA bathi bahlela ukwakhela uhulumeni wase-US amafekthri e-AI, okuhlanganisa nama-GPU angu-100,000 engqalasizinda evikelekile ye-AWS yemithwalo yemisebenzi yezokuvikela kuzwelonke ku-Impact Level 6 nangaphezulu. Ukukhishwa akusho ama-ejensi, izinhlelo, amasayithi okuthunyelwa, amabhajethi, imigqa yesikhathi noma imisebenzi ethile lawa masistimu azoyenza. Ngakho-ke isungula ukuzibophezela kwengqalasizinda eshiwo, hhayi ubufakazi bokusetshenziswa okuthile kokuvikela kuzwelonke noma ukuhlolwa kwemiphumela yako. Ukusetshenziswa okuhleliwe kwenza futhi ezokuphepha, ukwengamela ukuthengwa kwempahla kanye nokuphathwa kwemithwalo ebucayi ibe yizindawo ezibalulekile ukuze kubikwe okwengeziwe.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
Ongakubuka ngokulandelayo
Imibuzo ebalulekile iwukuthi amandla e-GPU ahleliwe afika ngesikhathi esimisiwe, lapho asetshenziswa khona, yiziphi amakhasimende angafinyelela kuzo futhi ngaziphi izindleko, nokuthi izimangalo zokusebenza kwezinkampani nokusebenza kahle zibambe ekusetshenzisweni kokukhiqiza okuzimele. Ukukhishwa akuhlinzeki ngohlelo lwezikhathi lokusebenza olunemininingwane, izibopho zekhasimende, izidingo zamandla noma izinhlelo ezithile zenhlangano.
Okokuqala, bukela ukubulawa kuqhathaniswa nethagethi yango-2027–2028. Umthombo awuwephuli ama-GPU ayizigidi ezingu-2 ukwenza izenzakalo zaminyaka yonke, izifunda, amasayithi esikhungo sedatha noma inani elithile lomkhiqizo ngale kokuqamba imindeni ye-Blackwell Ultra, Rubin kanye ne-Rubin Ultra. Futhi ayisho ukuthi ungakanani umthamo o-odwe, oxhaswe ngemali, ofakiwe noma ubekelwe amakhasimende. Ukukhiqiza, ukupakisha, inethiwekhi, ukutholakala kwamandla kanye nezinguquko zokulawula noma zokuthunyelwa ngaphandle kungase kuthinte isikali sokugcina, kodwa ukukhishwa akunikezi ubufakazi bezinga lephrojekthi kulezo zinkinga.
Okwesibili, hlola ukuthi ukusebenza kahle okufunwayo kanye nokusebenza kuhumusha kube yimithwalo yangempela yomsebenzi. I-AWS ne-NVIDIA zibika izikhathi ezingafika kwezingu-3.7 zokucubungula idatha ngokushesha kanye nokusebenza kwentengo okungcono okungu-30% kokucushwa okuthile kwe-GPU esheshisa i-Amazon EMR, kanye nenkomba ye-vector esheshayo efika izikhathi ezingu-9 ngekota eyodwa yezindleko ku-Amazon OpenSearch Umthombo awuwacacisi amasethi edatha, izisekelo, umumo womthwalo womsebenzi, izilinganiso zokusetshenziswa noma izindleko zokusebenza eziphelele. Ukuhlolwa okuzimele kuzodingeka ukuze kunqunywe ukuthi lezo zibalo zimelela kangakanani kumakhasimende asebenzisa ukubuyisa-ukubuyisa, ukusesha kwe-semantic, izibalo ezinkulu noma ezinye izinhlelo zokusebenza zokukhiqiza.
Okokugcina, bukela ukukhishwa okungokoqobo kwemodeli, izingcezu zezokuphepha nezirobhothi. Ukukhishwa kuthi amamodeli e-Nemotron azotholakala nge-Bedrock kanye ne-SageMaker, kodwa ayinikezi amagama amamodeli amasha, izinsuku zokukhishwa, intengo noma ukutholakala kwesifunda. Ithi i-Amazon Amarobhothi kanye ne-NVIDIA izosebenza kukho konke ukulingisa, ukukhiqizwa kwedatha yokwenziwa, ukuqeqeshwa kwerobhothi, ukuthuthukiswa komzila, ukuphepha okusebenzayo nokuqinisekiswa kwe-real-to-sim, kodwa ayimemezeli irobhothi elithile, indawo, umphumela wokuphakelwa noma ukuhlolwa kokuphepha. Okubaluleke kakhulu okungaziwa ukuthi lokhu kuhlanganiswa kukhiqiza izinzuzo zekhasimende ezilinganisekayo kanye nezivikelo eziqinisekiswa ngezinga elichazwa yizinkampani.