Ibhodi Yezentuthuko I-ArduinoEzinguqulweni zayo ezahlukahlukene, kuyindlela enhle yabenzi nabathuthukisi abafuna ukudala amaphrojekthi abo we-DIY. Ngakolunye uhlangothi futhi une Ukusajingijolo Pi, i-SBC encane futhi eshibhile yokwakha inqwaba yamaphrojekthi. Kukho konke lokhu kungezwa inani elikhulu lezinye izindlela emakethe. Kepha mhlawumbe ufuna okuthile okucacile, okuthile ongasebenza nakho amanethiwekhi we-neural ne-AI. Ngemuva kwalokho ibhodi le-NVIDIA Jetson Nano yilokho okufunayo.
Uma uhlela ukwenza intuthuko ngokuya ngamanethiwekhi we-neural noma ufunde ngakho, ungathembela kuleli bhodi NVIDIA Jetson Nano ngakho. Konke kungaphezudlwana kwe- € 100, hhayi okuphezulu kakhulu uma kubhekwa amanani ezinye izinhlelo ezihlakaniphile ...
Uyini uJetson?
NVIDIA Jetson Nano Kungumsebenzi wenkampani eyaziwayo ye-chip chip eyenza ukuthuthukiswa kwezinhlelo eziningi ezintsha ze-AI ezincane. Ngaphezu kwalokho, ubheka amanani entengo amanye amaphrojekthi wamanethiwekhi we-neural, akwenze lokho ngentengo kanye nokusetshenziswa kwamandla okuphansi.
Ngaleli bhodi lentuthuko ungakha amaphrojekthi amaningi, njengezicelo ze-IoT, kusuka kumarobhothi amancane asekhaya, kuya kwezinye izinhlelo ezihlakaniphile ongazicabanga, ngokusebenzisa ama-network video recorders (NVRs), amasango ahlakaniphile, njll.
Konke kwimodyuli ye-PCB encane engaba ngu-70x45mm ngosayizi kunguqulo yayo yeNano, encane kunazo zonke. Eqinisweni, iyi- Uhlobo lwe-SOM module (System on Module) isilungele ukusebenza nayo, kanye nethuluzi lokuthuthukisa.
Izici zobuchwepheshe
Kepha izici zobuchwepheshe Kusuka ku-NVIDIA Jetson Nano, uzozithola unebhodi elinamandla okwazi ukwenza ama-GFLOPs angama-472 ukwenza ama-algorithm wesimanje we-AI ngokushesha okukhulu. Ingasebenzisa ngisho amanethiwekhi amaningi we-neural ngokufana futhi icubungule izinzwa eziningi ezinokulungiswa okuphezulu ngasikhathi sinye.
Futhi konke lokho kusetshenziswa kahle kwamandla. Ngaphandle kwamandla ayo, kuphela kudla phakathi kuka-5 no-10w. Amandla aphansi ngokwanele uma ucabanga ngezinzuzo ezinikezayo.
Ngemininingwane engaphezulu, ngikushiyela itafula eliphelele le- imininingwane:
| GPU | Ukwakhiwa kwe-NVIDIA Maxwell™ ngezinhlamvu ezingu-128 ze-NVIDIA CUDA® | |
| CPU | Iprosesa le-ARM® cortex®-A57 MPCore Quad Core | |
| Memoria | I-4GB 4-bit LPDDR64 | |
| Isitoreji | Isitoreji se-Flash esingu-16GB eMMC 5.1 | |
| Ukufaka ikhodi ngevidiyo | Amafreyimu angu-4K 30 (H.264 / H.265) | |
| Ukuhlelwa kwevidiyo | Amafreyimu angu-4K 60 (H.264 / H.265) | |
| Ikhamela | Indlela eyi-12 (3 x 4 noma 4 x 2) MIPI CSI-2 DPHY 1.1 (18 Gbps) | |
| Conectividad | I-Gigabit Ethernet | |
| Isikrini | I-HDMI 2.0 noma i-DP 1.2 | i-eDP 1.4 | I-DSI (1 x 2) 2 ngasikhathi sinye | |
| UPHY | 1 1/2/4 PCIE, 1 USB 3.0, 3 i-USB 2.0 | |
| INGABE | 1 I-SDIO / 2 SPI / 4 I2C / 2 I2S / GPIO | |
| Tamaño | 69,6 mm x 45 mm | |
| Imishini | Isixhumi sephini esingu-260 | |
Eminye imikhiqizo efanayo
I-NVIDIA nayo inikela neminye imikhiqizo efanayo kuJetson Nano wokuthuthukiswa kwe-AI ngamanethiwekhi we-neural wokufakelwa. Ezinye izibonelo yilezi:
- UJetson Xavier NX: imodyuli ye-SOM enikeza wonke amandla ekhompuyutha enkulu enobukhulu obuncane kakhulu. Ungakhuphukela kuma-TOP angama-21, okungukuthi, ama-21 Tera Operations ngemizuzwana. Amandla angaphezu kwamandla anele ukusebenzisa amanethiwekhi wamanje we-neural ngokufana futhi acubungule idatha kusuka kuzinzwa eziningi eziphakeme kakhulu ngasikhathi sinye.
- UJetson AGX Xavier: imodyuli entsha ephawula ingqophamlando ngokuya ngobuningi bekhompyutha nokusebenza kahle. Ye-AI, evumela ukwenziwa kwezizukulwane ezintsha zemishini ehlakaniphile.
- IJetson TX2- Elinye ibhodi lokuthuthukisa elisheshayo, elonga amandla lekhompiyutha ye-AI efakiwe. Ikhompiyutha enkulu kumodyuli esekwe ku-NVIDIA Pascal GPU. Ngafika ku-8GB ye-RAM kanye ne-bandwidth ye-59,7GB / s.
Vele, njengoba ubona, abafowabo abadala banaye amanani ephakeme kakhulu ...
Thola i-NVIDIA Jetson Nano
Uma uthatha isinqumo soku thenga i-NVIDIA Jetson Nano, unamathuba amaningi okukuthola. Enye yimikhiqizo enikezwa ngeplatifomu ye-Amarzon. Uzothola womabili ibhodi lezentuthuko ngaphandle kwamakhithi amaningi wokuthuthuka afaka phakathi i-adaptha yamandla, njll. Ngokwesibonelo:
- I-NVIDIA Jetson Nano Basic
- I-NVIDIA Jetson Nano Kit ene-adaptha yamandla, i-64GB microSD, i-USB
- Thenga kuphela imodyuli ye-SOM
Khumbula ukuthi ubuchwepheshe be ukufunda imishini, ubuhlakani bokufakelwa, ukufunda okujulile, njll., kuya ngokuya kuba "yimfashini", ngoba banenqwaba yezicelo ezithakazelisa kakhulu. Ngakho-ke, ukufunda ngazo kungaheha ukwenza amaphrojekthi amasha ngekusasa noma uthole imisebenzi ethokozisayo ezinkampanini ezidinga lolu hlobo lwamakhono ...