Ukusetyenziswa kothotho lweNVIDIA Jetson kwiikhompyutha zemizi-mveliso ezifakiweyo

Isiqulatho

• Yintoni uthotho lweJetson lweNVIDIA?
• Iingenelo zobugcisa eziphambili zothotho lweJetson lweNVIDIA:
• Iingenelo zothotho lweJetson kwiikhompyutha zoshishino ezifakiweyo:
• Ukusetyenziswa kwiimeko zokwenyani:
• I-SINSMART inamava ekusombululeni izisombululo ezifanelekileyo zoshishino:

Yintoni uthotho lweJetson lweNVIDIA?

Okwangoku, i-AI itshintsha utshintsho lweshishini ngokusebenzisa ukwenziwa kwezinto ngokuzenzekelayo kunye nohlalutyo lwedatha. I-Jetsom ye-NVIDIA liqonga elenzelwe ngokukodwa iirobhothi kunye nezicelo ze-AI ezifakwe emphethweni. Le mveliso inokufakelwa kwiikhompyutha zoshishino kwaye inokubonelela ngamandla ekhompyutha afikelela kwi-275tops. Okwangoku, olu luhlu lweemveliso lunee-series ezininzi, kwaye uthotho ngalunye lunamandla ahlukeneyo ekhompyutha.

Iingenelo zobugcisa eziphambili zothotho lweJetson lweNVIDIA:

Uthotho lweJetson lweNVIDIA lube lukhetho olufanelekileyo kwiikhompyutha zoshishino ezifakwe ngaphakathi kunye namandla alo anamandla e-AI computing kunye neenzuzo ze-GPU edge computing. Ii-CUDA kunye nee-Tensor cores ezakhelwe ngaphakathi ziphucule kakhulu ukusebenza kakuhle kwengqiqo yokufunda nzulu, ngakumbi kwizicelo ezinje ngokucubungula imifanekiso kunye nombono wekhompyutha. Olu thotho luneempawu zokusetyenziswa kwamandla aphantsi kunye namandla aphezulu ekhompyutha, afanelekileyo ekusebenzeni ixesha elide kwezixhobo zoshishino;

Iingenelo zothotho lweJetson kwiikhompyutha zoshishino ezifakiweyo:

1. Izixhobo zochungechunge lweJetson zine-CUDA core yeNVIDIA eyakhelwe ngaphakathi, enamandla. Amandla okuqiqa e-AI ochungechunge lweJetson anokwenza imisebenzi yokufunda enzulu, ingakumbi xa isixhobo esisecaleni senza i-AI yokuqiqa ngexesha langempela ngokukhawuleza, kwaye senza ngcono kwaye sikrelekrele ekuboneni komatshini nasekufumaneni ngokuzenzekelayo;

2. Uthotho lweJetson luxhasa imisebenzi esebenzayo yokubala kunye nokukhawulezisa esebenza ngaxeshanye, ingakumbi ekucubunguleni idatha eninzi kunye neemeko ze-algorithm ezintsonkothileyo, ezinje ngohlalutyo lwevidiyo ngexesha langempela, ukufunyanwa kwezinto, njl.njl., zibonelela ngokusebenza okugqwesileyo ukuqinisekisa ukuba iikhompyutha zemizi-mveliso ezifakwe ngaphakathi zinokusingatha imisebenzi ngokufanelekileyo kwizicelo zoshishino;

Ukusetyenziswa kwiimeko zokwenyani:

Ukwenziwa kobukrelekrele: uhlolo olubonakalayo kunye nolawulo oluzenzekelayo kwimigca yemveliso

Kwimveliso ekrelekrele, uthotho lweJetson lusetyenziswa kakhulu ekuhlolweni okubonakalayo kunye nolawulo oluzenzekelayo lwemigca yemveliso. Ngobuchule bayo obunamandla bokuqiqa be-AI, iJetson inokucubungula idatha yemifanekiso kunye nevidiyo ngokukhawuleza, ifikelele ekufumaneni iziphene ngokuchanekileyo, ukuhlelwa kwemveliso kunye nolawulo lomgangatho, iphucule ukusebenza kakuhle kwemveliso kunye nokuchaneka, kwaye inciphise ukungenelela kwabantu.

Iirobhothi: Ukuhambahamba, ukucubungula okubonakalayo kunye nokulawula intshukumo

Kwicandelo leerobhothi, iJetson inika inkxaso esebenzayo yokucubungula okubonakalayo kunye nokulawula intshukumo kwiirobhothi. Amandla ayo okukhawulezisa i-GPU avumela iirobhothi ukuba zenze ngokukhawuleza ukuqonda okusingqongileyo, ukucwangcisa indlela kunye nokuqaphela izinto, ingakumbi kwimisebenzi yokuzimela yokuhamba kunye nokusebenza. Iimveliso zochungechunge lweJetson zifanelekile kwiimeko ezahlukeneyo ezifana neerobhothi zenkonzo kunye neerobhothi zoshishino.

I-Edge computing: Ukucutshungulwa kwedatha kunye nohlalutyo lwe-intanethi yezinto zemizi-mveliso ezisasazwe

Uthotho lweJetson lube sisixhobo esibalulekileyo sokubala idatha ngendlela ekhawulezileyo kunye nobuchule bokucubungula idatha obusebenzayo. Kwi-intanethi yezinto ezisasazwayo kwimizi-mveliso, iJetson inokucubungula idatha kwindawo evela kwinani elikhulu leenzwa kunye nezixhobo, ikwazi ukujonga ngexesha langempela, ukugcina izinto kwangaphambili kunye nohlalutyo olukrelekrele, inciphise ukuxhomekeka kwilifu, kwaye iphucule isantya sempendulo yenkqubo.

I-SINSMART inamava ekusombululeni izisombululo ezifanelekileyo zoshishino: