- Ukwethulwa kwe-RTX Spark SoC ngokwakhiwa kwe-ARM, okuhlanganisa i-CPU enama-core angu-20 kanye ne-Blackwell GPU enamandla.
- Ukusekelwa kwendabuko kwe-Windows 11 ngokugxila kakhulu ku-AI yendawo, ama-ejenti ahlakaniphile, kanye nokudalwa kokuqukethwe kobungcweti.
- Ukuncintisana okuqondile ne-Apple Silicon ne-Qualcomm, okuvumela ama-laptop amancane kakhulu futhi asebenza kahle.
Isimo sekhompyutha yomuntu siqu sisanda kuguquka ngokuphelele. Kwa-Computex, i-Nvidia ikhiphe i-all-in-one chip ngokwethulwa kwe- RTX Spark , i-chip ehlanganisa konke ehlose ukususa ukwakheka kwendabuko kumakhompyutha womuntu siqu. Lokhu akuyona nje enye ingxenye; kuyisinyathelo esinamandla ohlelweni lwe-ARM ukuze kuthuthukiswe ukusebenza kwe-AI kanye nehluzo kumadivayisi eselula.
Nakuba amahemuhemu esakazeke izinyanga eziningi ukuthi igama lizoba yi-N1 noma i-N1X, inkampani ikhethe igama eliveza amandla nobuhlakani. Le superchip ye-petaflop ayigcini nje ngokuhlukana ne-inertia ye-Intel ne-AMD, kodwa futhi ibeka amehlo ayo ku-Apple ne-Qualcomm, ihlose ukwenza i-Windows ku-ARM ingabi yisithembiso futhi ibe ngokoqobo oluvelele kuma-laptop aphezulu.
Uma sibheka ngaphansi kwe-hood, sithola isilo sobuchwepheshe. Enhliziyweni yayo kukhona i-CPU eyenziwe ngokwezifiso esekelwe ku-ARM enama-core angu-20 , eyakhiwe ngokubambisana ne-MediaTek. Ngokusho kwe-Nvidia, ukusebenza kwale processor kuncintisana nokungcono kakhulu okutholakala emakethe ye-ARM PC, okushaya ibhalansi ephelele phakathi kwamandla aluhlaza kanye nokusetshenziswa kwamandla aphansi.
Kodwa lapho ikhanya khona ngempela kuse-GPU. I-RTX Spark ihlanganisa ukwakheka kwe-Blackwell okunama-cores angu-6.144 e-CUDA kanye nama-Tensor Cores esizukulwane sesihlanu. Kalula nje, ukusebenza kwehluzo kufana nokwe- laptop RTX 5070 , kodwa kuhlanganiswe ne-SoC. Konke lokhu kunikwe amandla ubuchwepheshe be -Nvidia NVLink-C2C , obuxhumanisa izingxenye ze-chip, ezikhiqizwe yi-TSMC kusetshenziswa inqubo yesimanje ye-nanometer engu-3.
Ngokuphathelene nememori, uhlelo lungaphatha kufika ku- 128 GB wememori ehlanganisiwe ye-LPDDR5X , kanye ne-bandwidth efinyelela ku-300 GB/s. Lokhu kubalulekile ngoba kuvumela i-GPU ukufinyelela idatha ngaphandle kwezithiyo, kususa isidingo sememori ezinikele futhi kuvumela izinhlelo eziningi ezincane.
I-AI kanye nobuciko: Inhloso yangempela
I-Nvidia ayikhiphi le chip ukuze nje yenze imidlalo isebenze ngokushesha, kodwa ukuze yenze i-AI ibe yinkanyezi ephelele. Ngenxa yamandla ayo okucubungula, i-RTX Spark iyisisekelo esifanele sokusebenzisa ama-ejenti endawo ku-Windows ngasese nangokuphephile, besebenzisana eduze ne-Microsoft ukuhlanganisa lezi zici kumathuluzi afana ne-OpenClaw ne-Hermes.
Kubadali bokuqukethwe, lezi yizindaba ezinhle kakhulu. Sikhuluma ngekhono lokuhlela ividiyo ngesisombululo se-12K 4:2:2 nokucubungula amamodeli ezilimi amakhulu anamapharamitha afinyelela ku-120 trillion. Eqinisweni, i-Adobe iklama kabusha umongo we-Photoshop ukuze ube uhlelo lokusebenza olusheshiswa yi-GPU oluyi-100% ngqo lwale hardware.
Uma kukhulunywa ngemidlalo, isithembiso sizimisele kakhulu: ukudlala izihloko ze-AAA ngesinqumo se-1440p nge-FPS engaphezu kwe-100 , ngenxa yokulandelela imisebe kanye nomlingo we-DLSS 4.5. Nakuba ukuhambisana kwemidlalo yemvelo ku-ARM kusalokhu kuyinselele, ungqimba lokulingisa lwe-Prism lwe-Microsoft kanye nokusebenzisana nabahlinzeki besofthiwe kuqinisekisa ushintsho olubushelelezi kakhulu, okuvumela i-Windows 11 ukuthi ilungiselelwe imidlalo futhi ithole ukusebenza.
Impi ye-laptop: Sala kahle ngokuqina
Umthelela ongokoqobo wale chip uyamangalisa. I-Nvidia ithi i-RTX Spark ivumela ukwakhiwa kwama-laptops angu- 14 mm ubukhulu futhi anesisindo esingaba ngu-1,3 kg. Uma kuqhathaniswa ne-MacBook Pro engu-14 intshi, ekhuluphele kancane futhi esindayo, kusobala ukuthi i-Nvidia ihlose ukunqoba impi yokuthwala.
Uhlu lwabalingani lukhulu. Imikhiqizo efana ne -ASUS, i-Dell, i-HP, i-Lenovo, kanye ne-MSI isivele iqinisekisile ukuthi izokwethula amadivayisi ngale SoC. Ngisho ne-Microsoft izoletha ubuchwepheshe kumugqa wayo we-Surface, ikakhulukazi i- Surface Laptop Ultra , idivayisi enesibonisi se-MiniLED esingu-15-intshi kanye nokukhanya kwe-HDR okungu-2000 okumangalisayo.
Elinye iphuzu elibalulekile ukusebenza kahle kwamandla. Ukusetshenziswa kwamandla kuphansi kakhulu kunanoma iyiphi inhlanganisela ye-x86 CPU kanye ne-GPU ezinikele. I-TDP ingasukela kuma-watts angu-80 iye phansi kuma-watts angu-10 , okuholela empilweni yebhethri ende kakhulu kanye namazinga okushisa alawulwayo ngaphandle kwesidingo samafeni anomsindo.
I-ARM vs x86: Kwenzekani ngempela?
Ukuze uqonde umongo, kubalulekile ukwazi ukuthi i-ARM isebenzisa ukwakheka kwe -RISC (Reduced Instruction Set Computing) , kuyilapho i-Intel ne-AMD zisebenzisa i-CISC. Inzuzo ye-ARM ukuthi isebenza kahle kakhulu ngokwemvelo, yize impumelelo enkulu ye-Apple ibangelwe ukutshalwa kwezimali okukhulu ekwakhiweni. Nge-RTX Spark, i-Nvidia izama ukuvala leso sikhala ohlelweni lwe-Windows.
- Rendimiento: Nakuba i-Intel Core Ultra isalokhu iyinkosi emisebenzini esindayo yesikhathi esidlule, i-ARM ibusa ekusebenzeni kahle nasekulawuleni ukushisa.
- Ukuhambisana: Isithende esikhulu se-Achilles se-ARM kube yisofthiwe, kodwa ngokulingisa nokusekelwa okuvela ezikhulwini ezinkulu ezifana ne-Adobe ne-Blackmagic, inkinga iyaxazululwa.
- IA: Ama-NPU ahlanganiswe ne-ARM anikeza izikhathi zokuphendula ezisheshayo kakhulu zokucabanga kwendawo.
Naphezu kwakho konke, indlela ebheke phambili ayinazo izinkinga zayo. Kusenezinkinga zokuhambisana nabashayeli bezinga le-kernel kanye nezinye izinhlelo zokulwa nokukhohlisa emidlalweni ethandwayo ezingaba yinkinga, yize iNvidia isivele isebenza nabathuthukisi ukugwema amaphutha ahlupha isizukulwane sokuqala samaprosesa e-Snapdragon.