I-NVIDIA isandula kukhupha ikhadi lemizobo yeGeForce GTX 1660 Ti esekwe kwi-TU116 GPU entsha, kodwa intshukumo yoyilo lweTuring ukuya kwizixhobo zohlahlo lwabiwo-mali akukapheli. NgeGTX 1660 Ti, inkampani ithathe indawo yeGeForce GTX 1070 ngemodeli entsha nefikelelekayo ngokusetyenziswa kwamandla aphantsi, kodwa iGeForce GTX 1660 entsha ijongene nomnye umsebenzi: ukuvala umsantsa kwikhathalogu yeNVIDIA ekhoyo phakathi kweGeForce GTX. I-1060 kunye ne-GTX 1070 ekwindla yokugqibela, i-Radeon RX 590 yahlala kulo msantsa, kunye ne-Radeon RX 580, ngenxa yokuphucula umqhubi kunye nokutshintshwa kwemidlalo ukuya kwi-Direct3D 12, yaba yinto enye efanelekileyo kwi-GeForce GTX 1060. Kodwa. ngokukhululwa kwe-GTX 1660, ii-GPU "ezibomvu" zinembangi enzulu kudidi lwamakhadi evidiyo yabathengi, kuba imveliso entsha inexabiso eliphantsi kune-Radeon RX 590 kwaye inokusebenza okuphezulu.
Iimpawu zobugcisa, ixabiso
I-GeForce GTX 1660 isekwe kwiprosesa yemizobo ye-TU116 kunye neeyunithi zekhompyuter ezivaliweyo. Umahluko kuqwalaselo lwe-GPU phakathi kweGTX 1660 kunye ne-GTX 1660 Ti yehla kwi-multiprocessors ezimbini zostrimisho (SMs), ezidibene ziqulathe i-128 32-bit CUDA cores kunye ne-8 texture mappers. Ke, i-GeForce GTX 1660 throughput ihlupheke kuphela yi-8,3% kwimisebenzi yeendawo ezidadayo kunye nokuzaliswa kwe-texel ngaphandle kokulungelelanisa izantya zewotshi ye-GPU. Kwaye ii-frequencies, ngendlela, ziye zanda kuphela kwimodeli encinci: i-NVIDIA yandisa i-frequency yesiseko nge-30 MHz, kunye ne-Boost Clock nge-15 MHz.
Kodwa olo tshintsho lufihlakeleyo alunakwanela ukwahlula iGeForce GTX 1660 kunye neGTX 1660 Ti. Into ephambili eyahlula iimodeli ezimbini yayiluhlobo lwe-RAM. Ngelixa uhlengahlengiso lwe-Ti luxhotyiswe ngeetshiphusi ze-GDDR6 ezine-bandwidth ye-12 Gbps ngephini nganye, i-GeForce GTX 1660 ibuyele kumgangatho we-GDDR5. Ngaphezu koko, i-GTX 1660 ixhotyiswe ngeechips kunye ne-bandwidth ye-8 Gbps, oku kuthetha ukuba ngokubhekiselele kwi-bandwidth yememori epheleleyo, ikhadi elitsha levidiyo lihambelana ngokupheleleyo neenkcukacha zokuqala ze-GeForce GTX 1060 kunye ne-6 GB RAM, kunye neenguqulelo zamva. yeGTX 1060 ene-9 Gbps RAM nokuba igqwesile iGTX 1660 inale parameter. Nangona kunjalo, iprosesa yegraphics ye-TU116, ngenxa yokuphucula ukunyanzeliswa kombala, isebenza kunye ne-RAM ngokufanelekileyo.
Kuba isantya sewotshi okanye isantya se-GPU ye-GeForce GTX 1660 kunye ne-GTX 1660 Ti yahluke kakhulu, kwaye imodeli encinci ikwaphethe iichips ze-RAM ezinokusetyenziswa okuphezulu kwamandla (xa kuthelekiswa ne-GDDR6), i-accelerators ezimbini ezincinci zosapho lwe-Turing ziphawuliwe. ngogcino lwamandla afanayo - 120 W.
Sele sixoxe ngezinye iimpawu ze-chip ye-TU116 xa kuthelekiswa nabameli abapheleleyo be-Turing family (TU106, TU104 kunye ne-TU102) ekuhlaziyweni kwe-GeForce GTX 1660 Ti, kodwa kuyafaneleka ukugxila kwizinto ezininzi eziphambili ezenza I-TU116 efana ne-analogues yayo endala okanye, ngokuchaseneyo, izoba phakathi kwabo umda ongenakoyiswa. Iyonke, i-TU116 ifaka zonke izinto ezintsha ezenziwe yi-NVIDIA kwi-architecture ye-Turing, ngaphandle kwee-cores ezenza i-Ray Tracing kunye ne-tensor cores eyenza izibalo ze-FMA (i-Fused-Multiply Add) izibalo kwi-half-precision real matrices (FP16). ). Ezi zokugqibela zisetyenziswa ikakhulu kwimisebenzi yokufunda koomatshini, xa i-GPU idlula idatha kwinethiwekhi ye-neural eyenziwe kwangaphambili kwindawo okanye kwifama ekude. Ke, i-GeForce GTX 1660 kunye ne-GTX 1660 Ti ngaxeshanye yaphulukana nokuhambelana nazo zombini i-DXR (i-Direct3D 12 ulwandiso lokulandela umkhondo we-ray) kunye netekhnoloji ye-DLSS, evumela i-GPU ukuba inikezele ngesisombululo esincitshisiweyo kunye nesakhelo esilandelayo sokulinganisa usebenzisa inethiwekhi ye-neural.
Endaweni yeeyunithi ze-tensor, i-NVIDIA ixhobise i-TU116 ngoluhlu olwahlukileyo lwe-16-bit ye-CUDA cores - ayikhawulezi ngokwaneleyo ukuqhuba i-DLSS ngokufanelekileyo, kodwa sele kukho imidlalo esebenzisa i-half-precision operations kwizibalo ze-shader (umzekelo, uWolfenstein II. : I-Colossus eNtsha), ngenxa yokuba ukusebenza kwee-GPU ezifanelekileyo (ngoku iVega kunye neTuring chips) kwandisa kakhulu. Ngaphandle koko, kwakhona, i-TU116 iyahluka kwiitshiphusi ezindala zosapho lwayo kuphela ngendlela yobuninzi, inayo yonke imibhobho ephuculweyo ekwi-architecture ye-Turing, kwaye ixhasa imisebenzi yonikezelo yobunini efana ne-VRS (i-Variable Rate Shading).
| Umenzi | NVIDIA | |||||
|---|---|---|---|---|---|---|
| Umzekelo | GeForce GTX 1060 3 GB | GeForce GTX 1060 6 GB | GeForce GTX 1660 | GeForce GTX 1660 Ti | GeForce RTX 2060 | GeForce RTX 2070 |
| I-GPU | ||||||
| Isihloko | GP106 | GP106 | TU116 | TU116 | TU106 | TU106 |
| microarchitecture | Easter | Easter | Turing | Turing | Turing | Turing |
| Itekhnoloji yenkqubo, nm | I-16 nm FinFET | I-16 nm FinFET | 12 nm FFN | 12 nm FFN | 12 nm FFN | 12 nm FFN |
| Inani lee-transistors, izigidi | 4 400 | 4400 | 6 600 | 6 600 | 10 800 | 10 800 |
| Ikloko yekloko, MHz: Ikloko yeSiseko / iClock yoBoost | 1506/1708 | 1506/1708 | 1530/1785 | 1500/1770 | 1365/1680 | 1 / 410 (uHlelo lwabaseki: 1 / 620) |
| Inani le-shader ALUs | 1152 | 1280 | 1408 | 1536 | 1920 | 2304 |
| Inani lokwaleka kokuthungwa | 72 | 80 | 88 | 96 | 120 | 144 |
| Inani le-ROPs | 48 | 48 | 48 | 48 | 48 | 64 |
| Inani le-tensor cores | akukho | akukho | akukho | akukho | 240 | 288 |
| Inani lee-RT cores | akukho | akukho | akukho | akukho | 30 | 36 |
| RAM | ||||||
| Ububanzi bebhasi, kancinci | 192 | 192 | 192 | 192 | 192 | 256 |
| Uhlobo lwetshiphu | IGDDR5 SDRAM | IGDDR5 SDRAM | IGDDR5 SDRAM | IGDDR6 SDRAM | IGDDR6 SDRAM | IGDDR6 SDRAM |
| Ubuninzi bewotshi, MHz (i-bandwidth ngomfowunelwa, iMbps) | Ishumi (2000) 8000 (2250) | Ishumi (2000) 8000 (2250) | 2000 (8000) | I-1 500 (12 000) | I-1 750 (14 000) | I-1 750 (14 000) |
| Umthamo, MB | 3 096 | 6 144 | 6 144 | 6 144 | 6 144 | 8 192 |
| I/O ibhasi | I-PCI Express 3.0 x16 | I-PCI Express 3.0 x16 | I-PCI Express 3.0 x16 | I-PCI Express 3.0 x16 | I-PCI Express 3.0 x16 | I-PCI Express 3.0 x16 |
| Imveliso | ||||||
| Incopho yokwenziwa komsebenzi FP32, GFLOPS (ngokusekwe kubuninzi bezihlandlo ezixeliweyo) | 3935 | 4372 | 5027 | 5437 | 6451 | 7 / 465 (uHlelo lwabaseki) |
| Ukusebenza FP32/FP64 | 1/32 | 1/32 | 1/32 | 1/32 | 1/32 | 1/32 |
| Ukusebenza FP32/FP16 | 1/128 | 1/128 | 2/1 | 2/1 | 2/1 | 2/1 |
| RAM bandwidth, GB/s | 192/216 | 192/216 | 192 | 288 | 336 | 448 |
| Imveliso yomfanekiso | ||||||
| Ujongano lwemveliso yomfanekiso | I-DL DVI-D, i-DisplayPort 1.3/1.4, i-HDMI 2.0b | I-DL DVI-D, i-DisplayPort 1.3/1.4, i-HDMI 2.0b | I-DL DVI-D, i-DisplayPort 1.4a, i-HDMI 2.0b | I-DL DVI-D, i-DisplayPort 1.4a, i-HDMI 2.0b | I-DL DVI-D, i-DisplayPort 1.4a, i-HDMI 2.0b | I-DL DVI-D, i-DisplayPort 1.4a, i-HDMI 2.0b |
| TBP/TDP, W | 120 | 120 | 120 | 160 | 175/185 (Uhlelo lwabaseki) | |
| Ixabiso lokuthengisa (e-USA, ngaphandle kwerhafu), $ | 199 (icetyiswa ngexesha lokukhululwa) | 249 (ecetyiswayo ekukhululweni) / 299 (uHlelo lwabaseki, nvidia.com) | 229 (iyacetyiswa) | 279 (iyacetyiswa) | 349 (kucetyiswa) / 349 (uHlelo lwabaseki, nvidia.com) | 499 (kucetyiswa) / 599 (uHlelo lwabaseki, nvidia.com) |
| Ixabiso lokuthengisa (eRashiya), rub. | ND | ND (kucetyiswa ngexesha lokukhululwa) / 22 (uHlelo lwabaseki, nvidia.ru) | 17 (ecetyiswayo) | 22 (ecetyiswayo) | ND (kucetyiswa) / 31 (uHlelo lwabaseki, nvidia.ru) | ND (kucetyiswa) / 47 (uHlelo lwabaseki, nvidia.ru) |
I-GeForce GTX 1660 ingabizwa ngokuba ngowesithathu (emva kwe-RTX 2060 kunye ne-GTX 1660 Ti) umlandeli wekhadi levidiyo eliphakathi kwexabiso eliphakathi kwintsapho yasePascal - iGeForce GTX 1060. Kodwa ukuba uvala amehlo akho kwisixa se-RAM, ke ngokubhekiselele kwindawo yayo kumgca wemveliso entsha kufuneka ilingane ne-GeForce GTX 1060 version kunye ne-3 GB ye-RAM. Xa kuthelekiswa neyamva nje, i-GTX 1660 ayinayo kuphela i-buffer yefreyimu ephindwe kabini, kodwa ikwanayo ne-27% ngaphezulu kwe-shader throughput yethiyori kunemodeli yakudala, njenge-1660% ye-GTX 24 Ti igqithise i-GTX 1060 epheleleyo kunye ne-6GB ye-RAM. Ngelo xesha, i-NVIDIA ayilahli umgaqo-nkqubo wexabiso obekwe yi-GeForce RTX 20 intsapho yamakhadi evidiyo, apho zonke izixhobo ezitsha zibiza umthengi ngaphezu kwee-analogue zabo ngokuthe ngqo ngamanani emodeli ukusuka kwisizukulwana sangaphambili. Ngoko i-GeForce GTX 1660 iye yathengiswa ngexabiso elicetyiswayo le-$ 229, nangona i-GeForce GTX 1060 kunye ne-3 GB ye-RAM iqale kwi-$ 199.
Ukujonga kwithegi yexabiso lemveliso entsha, umntu unokuphinda acatshukiswe kukubawa kwe-NVIDIA, ukuba akunjalo ngenxa yobuthathaka beminikelo yangoku ye-AMD, ethi, ngokufika kwe-architecture ye-Turing, isasazeke ukusuka phezulu ukuya kwinqanaba lexabiso eliphakathi. Ke, olona hlengahlengiso lufikelelekayo lweRadeon RX 590 (ixabiso ukusuka kwi-240 yeedola kwindawo entsha ye-newegg.com) okwangoku zibiza kakhulu kuneGeForce GTX 1660, kwaye kwimarike yaseRussia, isincomo se-NVIDIA (i-17 ruble) ibeke iGTX 990 inxalenye ephantsi yoluhlu, oluhlala kwimveliso ye-AMD (ukusuka kwi-ruble ye-1660 ngokwe-market.yandex.ru).
Ngokungafaniyo nezinye ii-accelerators kwi-Turing chips, kuquka i-GeForce GTX 1660 Ti, i-GTX 1660 ayinabaphi bakhuphisana ngokuthe ngqo kwinkampu yayo. Iimodeli ezisondeleyo ze-10-series ngokwemigqaliselo kunye nokusebenza - i-GeForce GTX 1060 6 GB kunye ne-GeForce GTX 1070 - ikude nemveliso entsha kwixabiso, nangona eyokuqala (kunye ne-GTX 1060 ngoku ithengiswa ngamaxabiso aqala kwi-$ 209 okanye i-14 iiruble) zisaya kulibaziseka ukuthatha abanye abathengi abanokubakho de oovimba bezinto zakudala ze-NVIDIA eziqokelelwe ngexesha le-cryptocurrency boom iphelile.
IGIGABYTE GeForce GTX 1660 OC: uyilo
Umbono wokuqala wekhadi elitsha lemizobo kudidi oluphantsi kunye nexabiso eliphakathi (kwaye, eneneni, malunga neemodeli ezibiza kakhulu) lenziwa ngcono kusetyenziswa uhlengahlengiso olulula njengomzekelo, kuba ezi zezona zifunwa kakhulu - ngokungafaniyo " premiumβ iinguqulelo ezisekwe kwi-GPU efanayo, ethi ngexabiso ihlala ingena kuluhlu lweyona modeli yakudala ikufutshane. Ngaloo ndlela, sinethamsanqa kwakhona, kuba iGeForce GTX 1660 imele isixhobo seGIGABYTE esikhutshwe ngaphaya kwe-WINDFORCE kunye ne-AORUS series. Awuyi kuphosisa ukuba uyaqaphela kwiifoto ikhadi elifanayo levidiyo esilivavanye kwiiveki ezintathu ngaphambili kwi-GeForce GTX 1660 Ti uphononongo - isebenzisa ibhodi yesekethe efanayo kunye nokupholisa, kodwa nge-GPU eyahlukileyo kunye ne-GDDR5 chips endaweni ye-GDDR6. .
I-GPU kwi-GIGABYTE GeForce GTX 1660 OC ibhodi i-over-clocked yangaphambili. Nangona singeke siyazi idatha echanekileyo malunga nokuphindaphinda kwayo kude kube kupapasho lwenqaku, xa umenzi ethumela inkcazo yeemveliso ezintsha kwiwebhusayithi yakhe, sele icacile kwinkqubo yokupholisa ethobekileyo ukuba ukugqithiswa kwe-overclocking apha kungumfuziselo kuphela. . Kwaye iGIGABYTE igqithise imodeli endala nge-30 MHz kuphela.
Uyilo lweGIGABYTE GeForce GTX 1660 OC lubonisa iimpawu zoqoqosho kulo lonke. Ikhadi levidiyo alinayo neyona nto ilula yokubuyela ngasemva, singasathethi ke ngee-RGB ze-LED ezine-tint eyenziwe ngokwezifiso kunye nokukwazi ukuqhagamshela imicu ye-LED. Iqhiya, eyenziwe ngeplastiki ngokupheleleyo, ivalela iPCB kumacala amathathu, ifihla imilinganiselo yePCB.
Inkqubo yokupholisa nayo ilula kakhulu: ukushisa kwe-GPU kunye ne-RAM chips zichithwa yi-radiator ye-aluminium, kwaye inxalenye yobhedu kuphela ngumbhobho wokushisa ogqithise kwisiseko sayo. Nangona kunjalo, i-GIGABYTE GeForce GTX 1660 OC epholileyo ayikho ngaphandle kophuculo oluthile. Ngaloo ndlela, i-radiator ine-protrusion edibeneyo kunye ne-transistors ye-field-effect kunye ne-voltage regulator chokes, kunye nabalandeli ababini abanobubanzi obuyi-87 mm bajikeleza kwicala elichaseneyo - ngaloo ndlela kuncitshiswe ukuphazamiseka kokuhamba komoya.
|
|
|
Iphakheji yephakheji ye-GIGABYTE GeForce GTX 1660 OC i-ascetic ngokusemandleni: ngaphezu kwekhadi levidiyo ngokwayo, ibhokisi iqulethe imiyalelo yephepha kuphela kunye ne-software disc.
IGIGABYTE GeForce GTX 1660 OC: PCB
Ngokusekelwe kwi-PCB esetyenziswe kwi-GeForce GTX 1660, i-GIGABYTE sele ivelise inqwaba yezinye izixhobo, ukusuka kwi-GeForce GTX 1660 Ti ukuya kwi-GeForce RTX 2070. Olu luhlu lweemodeli lubandakanya ii-GPU ezahlukeneyo (TU116, TU106) kunye neentlobo ezimbini ze-RAM. iitshiphusi (i-GDDR5 kunye ne-GDDR6) zihambelana nombane, kwaye imilinganiselo emincinci ye-PCB yenza ukuba kube lula ukuvelisa izixhobo zobukhulu obuqhelekileyo kunye namakhadi evidiyo ahlangeneyo e-Mini ITX form factor.
Le PCB inokwamkela amacandelo ezigaba ezisibhozo ze-voltage regulator, kodwa ukusetyenziswa kwamandla kwezixhobo ezisekelwe kwi-TU116 kunye ne-TU106 ukusuka kwi-120 ukuya kwi-175 W (ngokweenkcukacha zereferensi), ngoko i-accelerator ephantsi inelisekile ngesigaba esithandathu. I-VRM: izigaba ezine zikhonza i-GPU kunye ne-microcircuits imemori yokufikelela ngokungahleliwe. Ngenxa yobudlelwane bayo kunye neemodeli ezindala zosapho lweTuring, imveliso entsha ixhotyiswe nge-transistors esebenza kwintsimi kunye nomqhubi odibeneyo (obizwa ngokuba yiDrMOS okanye "izigaba zamandla" - izigaba zamandla), ezibonelela ngokusebenza okuphezulu kunye nokuvumela ukuba Umlawuli we-VRM we-PWM ukurekhoda ngokuchanekileyo i-voltage kwi-drain ye-transistors.
Nangona i-TU116's display controller ihambelana ne-DVI, i-GIGABYTE ikhethe izixhumi ezintathu ze-DisplayPort kunye nemveliso enye ye-HDMI. Kodwa i-GeForce GTX 3.1 kunye ne-GTX 2 Ti zivinjwe ngokusisiseko i-USB 1660 Gen 1660 interface ngenkxaso ye-DisplayLink protocol. Iipads zoqhagamshelwano zokubeka iliso kwi-voltages kunye ne-hardware voltmode, i-chip ye-BIOS egciniweyo kunye nobunye ubunewunewu obufanayo nazo, ngokuqinisekileyo, azikho apha.
umthombo: 3dnews.ru