NVIDIA Jetson Nano: duk game da kwamitin ci gaba ne

NVIDIA Jetson Nano

NVIDIA Jetson Nano Hukumar ci gaba ce ta musamman. Zai iya zama kamar naka ta hanyoyi da yawa Rasberi Pi, ko Arduino, amma an tsara shi musamman don takamaiman nau'in aikin. Kuma kamar waɗannan sauran allon talla, shi ma yana da ƙarancin farashi mai ƙanƙanci da ƙananan girma idan aka kwatanta da madadin kayan aiki.

Musamman, NVIDIA's Jetson Nano an keɓance musamman don ci gaban ayyukan fasaha na wucin gadi da ayyukan hanyoyin sadarwa na wucin gadi. Hanya mafi arha don farawa a wannan duniyar, koya yadda waɗannan tsarin mai hankali ke aiki, da ƙirƙirar rashin iyaka na ayyukan da zaku iya tunanin ...

Menene Jetson Nano?

NVIDIA Jetson Nano kwamiti ne na ci gaba, SBC wanda da shi za'a ƙirƙiri ayyuka da yawa dangane da hanyoyin yanar gizo, zurfafa karatu da AI. Tare da shi zaka iya kirkirar ayyuka daban-daban, daga kananan aikace-aikacen IoT mai hankali, zuwa robobi masu rikitarwa, tsarin hangen nesa na wucin gadi da kuma gane abu, na'urori wadanda suke yin aiki da hankali ta hanyar kimanta jerin sifofin firikwensin, kananan motoci masu sarrafa kansu, da dai sauransu

Amma duk tare da farantin fewan girma, kuma tare da farashi ƙwarai da gaske mai araha idan aka kwatanta da sauran tsarin ƙwararru tare da halaye iri ɗaya.

Kuma idan kayi mamaki me yasa zaka samu daya na waɗannan allon NVIDIA Jetson Nano, ya kamata ka tuna cewa waɗannan allon zasu ba ka damar ƙirƙirar ayyuka da yawa yayin koyo game da fasahar da ke taɓarɓarewa. Akwai kamfanoni da yawa da ke sha'awar mutane masu ilimin koyon na'ura, AI, zurfin ilmantarwa, da sauran nau'o'in horo iri daya, tunda fasaha ce ta gaba.

Halayen fasaha

SOM Jetson Nano

NVIDIA Jetson Nano tayi gaske m fasali don girmanta da farashinta. Da kyar ya wuce € 100, kuma tare da aan 'yan santimita cikin girma. Duk da wannan, yana iya haɓaka har zuwa 472 GFLOPs, isa don gudanar da yawancin algorithms na AI cikin sauri da aiwatar da hanyoyin sadarwa na wucin gadi da yawa a lokaci guda.

Kuma ba wai kawai ban sha'awa bane ga waɗannan adadi, amma kuma don ƙarancin amfani. Wannan allon na iya samun amfani da ke tsakanin 5 da 10W. Idan aka kwatanta da makamantan tsarin tabbas yana da ƙasa, don haka kuna fuskantar ingantaccen tsarin. Ba shi da alaƙa da wasu injunan da ke cinye ɗari ko dubban watts ...

Don ƙarin bayani, kuna iya ganin wannan cikakken cikakken bayani:

  • NVIDIA Maxwell GPU tare da mahimmin 128 CUDA
  • ARM Cortex-A57 QuadCore CPU
  • RAM 4GB LPDDR4
  • 16GB eMMC 5.1 ajiyar ajiya
  • Haɗuwa:
    • 12-hanyar haɗin kyamara (3 x 4 ko 4 x 2) MIPI CSI-2 DPHY 1.1 (18 Gbps)
    • Gigabit Ethernet na hanyar sadarwa (RJ-45)
    • HDMI 2.0 ko haɗin nuni na DP 1.2 | eDP 1.4 | DSI (1 x 2) 2 lokaci guda
    • Tashar jiragen ruwa 1/2/4 PCIE, 1 USB 3.0, 3 USB 2.0
    • Iarin I / O: 1 SDIO / 2 SPI / 4 I2C / 2 I2S / GPIO
    • 260-pin mai haɗawa
  • Girma: 69,6mm x 45mm
  • Amfani: 5-10w
  • Linux OS tare da kayan ci gaba

NVIDIA Jetson Kayayyakin Iyali

NVIDIA tana da yawancin waɗannan kayayyakin ci gaban AI tare da hanyoyin sadarwar neuroanal na wucin gadi. Wasu daga cikin shahararrun samfuran sune:

  • Jetson Xavier NX: ita ce SOM, ma'ana, System On Module, ko kuma cikakken tsarin da aka shigar cikin tsari guda ɗaya. Duk da kamanninta da girmanta, tana ba da ƙarfin ikon sarrafa kwamfuta, tare da har zuwa 21 TOPs, ma'ana, 21 Tera Operations a sakan ɗaya. Wannan ya isa ya gudanar da hanyoyin sadarwar wucin gadi da yawa a hankali kuma a lokaci guda.
  • Jetson AGX Xavier: wani mahimmin tsari mai karfin gaske dangane da yawan lissafi da inganci kuma hakan yazo ne bayan Jetson Nano, yana bada damar kirkirar sabbin tsararun injina masu fasaha.
  • Jetson TX2: wani madadin Jetson Nano, kuma daga dangi ɗaya. Ya yi fice domin tsananin gudu da ingancinsa. Musamman tsara don aikace-aikacen AI da aka saka, inda girman da amfani suke. A wannan yanayin, ya dogara ne akan gine-ginen NVIDIA Pascal, wanda aka ba da shi ta 8GB na RAM da kuma faɗin bandwidth har zuwa 59,7GB / s.

Sayi NVIDIA Jetson Nano

Idan kun kasance kuna son farawa a cikin mahaliccin ko duniyar DIY tare da ayyukan cibiyar sadarwar wucin gadi, zaku iya sayi wannan jirgin NVIDIA Jetson Nano a cikin shaguna na musamman ko a dandamali kamar Amazon, inda ake siyar dasu daban ko tare da kayan haɓaka don farawa da sauri tare da duk abin da kuke buƙata:

A halin yanzu an ƙaddamar da hukumar NVIDIA Jetson Nano tare da rage farashin kimanin $ 59 kuma ga abin da suka kara WiFi. Babban labari, abin kawai shine sun rage mahimman ƙwaƙwalwar ajiya zuwa 2GB. Idan kuna so shi dole ku jira, don yanzu yana cikin kawai presale don abokan aiki ...

Madadin NVIDIA Jetson Nano

Google Coral

Idan kuna sha'awar ilimin inji, AI da cibiyoyin sadarwar wucin gadi, ya kamata ku san wasu madadin NVIDIA Jetson Nano, tunda ba ita kadai bane farantin wadannan dalilai. Kuna iya samun wasu SBC waɗanda aka tsara musamman don waɗannan ayyukan kamar waɗannan masu zuwa:

Google Coral

Google ya ɓullo da lamba, Google Coral, tare da sauran kayan haɗi da kayayyaki da ake buƙata don ƙirƙirar ayyukan AI. Daga cikin abubuwan da ke cikin wannan dandalin kuna da:

Google Coral yana da wasu halaye na fasaha flashy, kamar:

  • NXP i.MX 8M CPU tare da Quad Core Cortex-A53 da Cortex-M4F
  • GC7000 Lite Graphics GPU,
  • Google Edge TPU coprocessor tare da har zuwa 4 TOPS ko 2 TOPS / w.
  • Ya hada da 1GB LPDDR4 RAM
  • Adana har flash na 8GB eMMC da yiwuwar faɗaɗa shi ta amfani da katunan microSD.
  • Yana da haɗin WiFi, USB, Bluetooth, Ethernet, jack jack, HDMI, MIPI-DSI, da iko akan USB-C 5v.

Farashin VIM3

Farashin VM3 Wani madadin ne don ayyukan AI, kodayake bashi da wasu halaye na manya, yana da madaidaiciyar hukuma da zata iya zama kyakkyawar dama don farawa:

  • CPU A311D x4 Cortex-A73 2.2Ghz da x2 Cortex-A53 a 1.8Ghz.
  • Tare da NPU a 5 TOPS
  • Har zuwa 4GB na RAM
  • 16-32GB eMMC Samsung
  • MIPI-DIS, HDMI, WiFi, Ethernet, microSD, USB, haɗin PCIe, da dai sauransu.

HiSilicon HiKey 970 (Huawei)

HiSilicon shine kamfanin da ke ƙasa Huawei wanda ke kera kwakwalwan. Da kyau, a ƙarƙashin wannan alamar zaku sami wani madadin don haɓaka ayyukan hanyoyin sadarwar hanyoyi kamar HiKey 970, dace da Huawei SDK. Bugu da kari, yana da wasu fasali masu kayatarwa:

  • ARM Kirin tare da Cortex A73 QuadCore + Cortex-A53 QuaCore
  • Mali G72 MP12 GPU
  • NPUs da aka keɓe
  • 6GB na LPDDR4
  • 64GB flash memory
  • WiFi, microSD, HDMI, USB, haɗin PCIe, da dai sauransu.
  • UEFI

Sophon BM1880 (matasan ARM + RISC-V)

Farashin BM1880 Yana da madadin kwamiti wanda Sophon.ia ya haɓaka. Idan ka yanke shawarar siyan daya, zaka sami wasu fasali kamar:

  • 2x Cortex-A53 CPU a 1.5Ghz + RISC-V a 1Ghz
  • 1 TPUs @ INT8 godiya ga mai sarrafa Tensor
  • 4GB LPDDR4
  • 32GB eMMC walƙiya
  • Haɗuwa Ethernet, WiFi, USB, microSD, Jack, da dai sauransu

Intel Neural Stick

Wani aikin kwatankwacin na baya shine wannan Intel Neural Stick. Shafin 2 yana yanzu, kuma abin da ke cikin wannan yanayin shine cewa sandar USB ce wacce zaka iya haɗa kai da komputa cikin nutsuwa don fara ayyukanka, kodayake yana da ƙarancin aiki fiye da allon da suka gabata. Hakanan, idan kuna buƙatar ƙarin ƙarfi, zaku iya amfani da su da yawa a cikin kebul na USB don ƙara iyawa ...

Si cin kasuwa wannan Neural Stick, farashinsa ya kai kimanin € 100, kuma ya dace da Linux da Windows. Bugu da kari, yana ba da damar aiki tare da OpenVINO azaman kayan aikin ci gaba.

Bayani na RK3399

Rockchip kuna da wannan ingantaccen kayan haɓaka-haɓaka ingantaccen kayan haɓaka ilmantarwa wanda da shi zaku ƙirƙiri abubuwa masu ban sha'awa da bambance bambancen ayyuka da shi. Yana tallafawa TensorFlow Caffe har zuwa 3 TOPS, da kuma tsarin Android da GNU / Linux.

Idan kanaso ka siya, to akwai shi a ciki iri daban-daban (an umarce shi daga mafi ƙasƙanci zuwa mafi girman farashi):


Abubuwan da ke cikin labarin suna bin ka'idodinmu na ka'idojin edita. Don yin rahoton kuskure danna a nan.

Kasance na farko don yin sharhi

Bar tsokaci

Your email address ba za a buga. Bukata filayen suna alama da *

*

*

  1. Mai alhakin bayanan: Miguel Ángel Gatón
  2. Dalilin bayanan: Gudanar da SPAM, gudanar da sharhi.
  3. Halacci: Yarda da yarda
  4. Sadarwar bayanan: Ba za a sanar da wasu bayanan ga wasu kamfanoni ba sai ta hanyar wajibcin doka.
  5. Ajiye bayanai: Bayanin yanar gizo wanda Occentus Networks (EU) suka dauki nauyi
  6. Hakkoki: A kowane lokaci zaka iyakance, dawo da share bayanan ka.