Ezweni elithuthukayo ngokushesha lamakhadi ezithombe, ukuhlala ungaphambi kwezitayela zemakethe kubalulekile kubathengi abangochwepheshe abafuna ukwenza izinqumo zokuthenga benolwazi. Njengoba imakethe ye-GPU ithuthuka ngezinto ezintsha ezifana nokusheshisa kwe-AI kanye nokulandela imisebe, ukuqonda lezi ntuthuko kanye nomthelela wazo ekusebenzeni komkhiqizo kubalulekile.
Lesi sihloko sidingida kuguquko lwemakethe lwakamuva, sigqamisa ubuchwepheshe obubalulekile namamodeli athengiswa kakhulu akha imboni. Ngokuhlola la mathrendi, ochwepheshe bangazulazula kangcono ezinkingeni zemakethe futhi baqondanise amasu abo nentuthuko enomthelela omkhulu. Izinqumo ezinolwazi namuhla zizobeka inkundla yempumelelo yesikhathi esizayo kulo mkhakha wokuncintisana.
Okuqukethwe
● Uhlolojikelele lwemakethe
● Ubuchwepheshe obuyinhloko nokuqanjwa kwemiklamo
● Amamodeli adayisa kakhulu ashayela amathrendi emakethe
● Isiphetho
Ukubukwa kwemakethe
Imakethe ye-GPU ikhula ngendlela engakaze ibonwe
Imakethe ye-GPU yomhlaba wonke ibhekene nokwanda okukhulu, imakethe inenani elikhulu $ 65.3 billion ku-2024 futhi kulindeleke ukuthi ifinyelele $ 274.2 billion nge-2029, ekhula ngezinga lokukhula lonyaka elihlanganisiwe (CAGR) lika 33.2%. Lokhu kukhula okudlondlobeleyo kuqhutshwa ukwanda kwesidingo sezithombe ezisebenza kahle kakhulu kwezemidlalo, ukubonwa kochwepheshe, i-AI, nezikhungo zedatha. Njengoba izimboni ezengeziwe zisebenzisa ubuchwepheshe obusheshiswa yi-GPU, imakethe iyaqhubeka nokukhula, ibeka ama-GPU njengengxenye ebalulekile kukhompyutha yesimanje.
Ukushintshashintsha kwezimakethe kanye nesimo sokuncintisana
Isimo sokuncintisana siyavela, nezinkampani ezinjenge-AMD, i-Nvidia, ne-Intel zilokhu zicindezela imingcele yokusebenza nokusebenza kahle. I-Nvidia ihola imakethe, ikakhulukazi ezingxenyeni ezifana ne-AI kanye nemidlalo ephezulu, kodwa i-AMD isizuze umhlabathi obalulekile ngama-GPU ayo angabizi kakhulu futhi anamandla, njengochungechunge lwe-Radeon RX 7900.
Ngaleso sikhathi, ukungena kwe-Intel emakethe ye-GPU ehlukile ngochungechunge lwayo lwe-Arc kwengeza esinye isendlalelo sokuncintisana, ikakhulukazi kusabelomali kanye nezigaba zebanga eliphakathi. Ukuguquguquka okuguquguqukayo kukhombisa imakethe enokuncintisana kakhulu lapho ukuqanjwa kabusha kanye namanani entengo kuzonquma amasheya emakethe eminyakeni ezayo.
Ubuchwepheshe obubalulekile kanye nemiklamo emisha
Ukulandelela imisebe nokuba ngokoqobo: Izinga elilandelayo lokuthembeka kokubuka
Ukulandelela uRay sekushintshe igeyimu embonini ye-GPU, okunika amandla ukunikezwa okungokoqobo ngokulingisa ukuziphatha kokukhanya. Le nqubo ilandelela izindlela zemisebe yokukhanya njengoba ihlangana nezinto ezibonakalayo, okuholela ekukhanyeni okunembile, ama-refractions, kanye nezithunzi. Ama-GPU anamuhla, njengalawo akuchungechunge lwe-RTX ye-Nvidia, ahlanganisa ama-RT cores azinikele ukwenza lezi zibalo eziyinkimbinkimbi ngesikhathi sangempela, ukuthuthukisa kakhulu ukwethembeka okubonakalayo kumageyimu.
Isibonelo, i-RTX 4090 ingafinyelela kufika ku-100% amafreyimu athuthukisiwe kumageyimu alandelelwe ama-ray uma kuqhathaniswa namamodeli adlule, okuyenza ibe ibhentshimakhi kumageyimu okusebenza okuphezulu. Ngaphezu kwalokho, I-Architecture ye-RDNA 3 ye-AMD inciphise igebe lokusebenza, inikezela ngamakhono okulandelela ama-ray aletha ukukhanya okungokoqobo kanye nemiphumela yethunzi, ngisho naku- 4K izinqumo.
Ukusheshisa kwe-AI: Ukuthuthukisa ukusebenza ngaphezu kokudlala
Ukusheshisa kwe-AI manje sekuyingxenye yokusebenza kwama-GPU anamuhla, anama-cores akhethekile aklanyelwe ukuphatha izimfuno ezishubile zekhompiyutha ze-AI nemisebenzi yokufunda komshini. I-Tensor Cores kaNvidia, etholakala ochungechungeni lwe-RTX 40, sebenzisa imisebenzi ye-matrix engavamile ukusheshisa imithwalo yemisebenzi ye-AI, efana nencazelo ejulile yokufunda kanye nokuqeqeshwa kwenethiwekhi ye-neural. Lawa ma-cores angaletha kufika izikhathi ezi-4 ukusebenza emisebenzini ye-AI uma kuqhathaniswa nezakhiwo ze-GPU zendabuko. Ngokufanayo, ama-accelerator e-AMD e-AMD ochungechungeni lwe-Radeon RX 7900 alungiselelwe ukuphindaphinda kwe-matrix, evumela ukubalwa kwe-AI okusheshayo okuthuthukisa ukuzizwisa kwemidlalo kanye nomsebenzi omningi ochwepheshe. Le ntuthuko ibalulekile hhayi kuphela kwezemidlalo kodwa nasezinhlelweni zokusebenza emikhakheni efana nezithombe zezokwelapha nezimoto ezizimele, lapho ukucubungula kwe-AI ngesikhathi sangempela kubalulekile.
I-Miniaturization nokusebenza kahle: Ikusasa lamandla ahlangene
Umkhuba obheke ekwenzeni okuncane ekwakhiweni kwe-GPU uqhutshwa isidingo sokusebenza okuphezulu kumadivayisi akhulayo aminyene futhi aphathekayo. Ama-GPU esimanje ayathuthukiswa ngezinqubo zokukhiqiza ezithuthukile, njenge Ubuchwepheshe benqubo ye-TSMC ye-5nm, okuvumela ukuminyana okukhulu kwe-transistor, okuholela kuma-chips anamandla kodwa amancane. Lokhu kulungiswa okuncane kuhambisana nokuthuthukiswa kokusebenza kahle kwamandla. Ngokwesibonelo, i I-Nvidia RTX 4070 Super isebenza nge I-TDP ye-220W nje, nokho iletha ukusebenza okuqhathaniswa nama-GPU amadala anezidingo zamandla eziphakeme kakhulu. Le ntuthuko ibalulekile ekuhlanganiseni ama-GPU ezinhlelweni eziphathwayo njengamakhompyutha aphathekayo kanye namadeskithophu anefomu elincane, lapho ukusetshenziswa kwamandla kanye nokuqedwa kokushisa kuyizinto ezikhathazayo ezibalulekile. Ngenxa yalokho, ama-GPU esizukulwane esilandelayo kulindeleke ukuthi angabi mancane nje kuphela kodwa futhi abe namandla kakhulu, awenze afanelekele uhla olubanzi lwezinhlelo zokusebenza kuyilapho enciphisa umthelela wawo kwezemvelo.
Amamodeli athengiswa kakhulu ashayela amathrendi emakethe
Izimanga ezimaphakathi: I-AMD Radeon RX 7900 GRE ihola iphakethe
I-AMD Radeon RX 7900 GRE isizimise ngokushesha njengomthengisi ongcono kakhulu emakethe ye-GPU ephakathi nendawo, enikeza ukuxutshwa okuphoqelela kokusebenza kanye nenani. Le modeli ifakwe I-16GB yememori ye-GDDR6 futhi isebenza nge-a Inkumbulo yememori ye-256-bit, okuyenza ilungele abadlali abafuna ukusebenza okuphezulu 1440p izinqumo. I-RX 7900 GRE's Ukukhipha amandla angu-300W isekela ukusebenza kwayo okuqinile, ukuletha ngaphezu kwama-TFLOPS angama-20 wamandla okucubungula, okuyivumela ukuthi ibambe imidlalo enzima kalula. Ukuduma kwayo kuqhutshwa futhi amanani ayo okuncintisana, okunikeza enye indlela engabizi kakhulu kuneminikelo kaNvidia ngenkathi isaletha okuhlangenwe nakho kwemidlalo yezinga eliphezulu.
Ubucwebe bomqhele kaNvidia: RTX 4070 Super kanye ne-RTX 4080 Super
I-RTX 4070 Super kaNvidia isiphenduke imodeli ethengiswa kakhulu, ikakhulukazi phakathi kwabadlali abafuna ibhalansi phakathi kwentengo nokusebenza. Le GPU inezici I-12GB yememori ye-GDDR6X futhi 2,475 MHz iwashi lokuthuthukisa, enikeza ukusebenza okumangazayo kukho kokubili 1440p kanye ne-4K izinqumo. I-RTX 4070 Super yaziwa kakhulu ngayo ukwazi ukulandela i-ray, inikezwe amandla yi Isizukulwane sesibili se-RT cores okuthuthukisa kakhulu ukukhanya kwegeyimu nemininingwane yesithunzi. Ngaphezu kwalokho, i- I-DLSS 3.0 isici sithuthukisa izilinganiso zozimele ngokusebenzisa i-AI ukuze kukhiqizwe ozimele abengeziwe, okuyenza kube ukukhetha okuncanyelwayo kubadlali bamageyimu ababeka kuqala kokubili ikhwalithi yokubuka nokudlala igeyimu okushelelayo.
I-RTX 4080 Super, enye into evelele ohlwini lwe-Nvidia, isiza abathanda abafuna ukusebenza okusezingeni eliphezulu ngaphandle kwethegi yenani eliphakeme le-RTX 4090. Nge I-16GB yememori ye-GDDR6X futhi I-320W TDP, izipesheli ze-RTX 4080 Super cishe ama-TFLOPS angama-30 wamandla okucubungula, okwenza kube inketho yokuya kuyo 4K imidlalo. Ikhono layo lokuphatha real-time ray ukulandelela kanye nezithombe ezithuthukisiwe ze-AI ezinqumweni eziphakeme kakhulu zikwenze kwaba intandokazi kubadlali bamageyimu abafuna okungcono kakhulu ku-hardware yabo.
Oshampeni besabelomali: AMD Radeon RX 7600 kanye neNvidia RTX 4060
Kubadlali abakwisabelomali, i-AMD Radeon RX 7600 kanye ne-Nvidia RTX 4060 yizinqumo ezihamba phambili. I-Radeon RX 7600, nayo I-8GB yememori ye-GDDR6 futhi I-165W TDP, inikeza ukusebenza okuhle kakhulu ku Imidlalo ye-1080p izimo. Isakhiwo sayo, esakhelwe phezu kwayo I-RDNA 3, ivumela ukusetshenziswa kahle kwamandla kuyilapho iletha ukusebenza okuqinile, ikakhulukazi ezihlokweni ze-AAA. Isikhalazo se-RX 7600 sisemandleni ayo okunikeza izici ezithuthukile ezifana I-FidelityFX Super Resolution endaweni yenani elifinyeleleka kubabukeli abaningi.
Ngokufanayo, i-Nvidia RTX 4060 idwebe i-niche njengenketho evumelana nesabelomali engayekethisi ezicini. Nge I-8GB yememori ye-GDDR6 futhi Ibhasi lememori le-128-bit, i-RTX 4060 isekela ukulandelela ama-ray futhi I-DLSS 3.0, okwenza kube imbangi enamandla engxenyeni yesabelomali. Iklanyelwe ukuvelela ku 1080p kanye ne-1440p yokudlala, ehlinzeka ngokusebenza okushelelayo ngisho nasemidlalweni enezithombe eziqinile. I-RTX 4060's I-115W TDP futhi iyenza ibe enye yezinketho ezonga amandla kakhudlwana emakethe, futhi yengeza ekuthandeni kwayo abadlali abaqaphela ibhajethi.
Isiphetho
Imakethe yamakhadi wezithombe ithuthuka ngokushesha, iqhutshwa ukuthuthuka okubalulekile kwezobuchwepheshe kanye noshintsho oluthandwayo lwabathengi. Ngamamodeli afana ne-AMD Radeon RX 7900 GRE kanye ne-Nvidia RTX 4070 Super ehamba phambili, imboni isethelwe ukuthi iqhubeke nomkhondo wayo okhuphukayo, inikeza izixazululo ezinamandla nezisebenza kahle kuzo zonke izingxenye. Njengoba amabhizinisi ezulazula kulo mkhakha wezwe oguquguqukayo, ukuhlala unolwazi mayelana nalezi zinguquko ezibalulekile kuzobaluleka ekwenzeni izinqumo zokuthenga zamasu ezihambisana namathrendi wesikhathi esizayo.
