Ukuthuthukiswa okushumekiwe
I-NVIDIA Jetson TX2 iletha isivinini kanye nokusebenza kahle kwamandla kumadivayisi ekhompyutha e-AI ashumekiwe. Le moduli ye-supercomputer ifakwe i-NVIDIA PascalGPU, ifika ku-8GB yememori, i-59.7GB /s yomkhawulokudonsa wememori yevidiyo, inikeza izinhlobonhlobo ze-interfaces ezijwayelekile zehadiwe, ukuzivumelanisa nemikhiqizo ehlukahlukene kanye nokucaciswa kwefomu, futhi ifinyelele umuzwa weqiniso we-AI computing terminal.
I-Al Artificial Intelligence
I-NVIDIA Jetson TX2 ingasebenzisa amanethiwekhi amaningi athuthukile e-neural afana ne-TensorFlow, PyTorch, Caffe/Caffe2, Keras.MXNet, nokuningi. Ngokunika amandla amakhono afana nokuqashelwa kwesithombe, ukutholwa kwento nokubeka endaweni, ukuhlukaniswa kwezwi, ukuthuthukiswa kwevidiyo, nokuhlaziya okuhlakaniphile, lawa manethiwekhi angasetshenziswa ukwakha amarobhothi azimele kanye nezinhlelo ezihlakaniphile ze-AI.
Ikhithi yokuthuthukisa iJetson TX2
I-NVIDIA Jetson TX2 iyikhithi yokuthuthukisa ye-AI eyonga amandla futhi enamandla, efakwe i-quad-core ARM A57 processor kanye ne-dual-core Denver2 processor, 256-core NVIDIA Pascal architecture GPU, super Al computing power, Ifanele imishini ekhaliphile. njengamarobhothi, ama-drones, amakhamera ahlakaniphile kanye nemishini yezokwelapha ephathekayo.
Ikhithi yokuthuthukisa ye-NVIDIA Jetson TX2 inikwa amandla yibhodi lokuthuthukisa i-Jetson TX2 futhi iza nezinhlobonhlobo ze-Hardware interface ezisekela i-NVIDIA JetPack, okuhlanganisa nemitapo yolwazi yesofthiwe efana ne-BSP, ukufunda okujulile, umbono wekhompyutha, i-GPU computing, ukucubungula i-multimedia, i-CUDA, i-cuDNN, kanye ne-TensorRT. Ezinye izinhlaka ze-Al ezidumile nama-algorithms nazo ziyasekelwa, njenge-TensorFlow, PyTorch, Caffe/Caffe2, Keras,MXNet, njll.
Uma kuqhathaniswa ne-Jetson TX1, i-Jetson TX2 iletha ukusebenza kwekhompuyutha kabili kanye nohhafu wokusetshenziswa kwamandla, ihlinzeka ngokusebenza okungcono nokunemba kwezinhlelo zokusebenza zedivayisi ezifana namadolobha ahlakaniphile, amafekthri ahlakaniphile, amarobhothi nama-prototypes okukhiqiza. Iphinde isekele zonke izici zemojula ye-Jetson TX1, kuyilapho inika amandla amanethiwekhi amakhulu we-neural ajulile nayinkimbinkimbi.
Izimiso zokucaciswa:
CPU: Dual-core Denver 264 bit CPU + quad-core ARM Cortex-A57 MPCore
I-GPU: 256 core Pascal GPU
Imemori :8GB 128-bit LPDDR4 Imemori yememori :32GB eMMC 5.1
Bonisa :HDMI 2.0 / eDP 1.4/2x DSl
Bonisa :HDMI 2.0 / eDP 1.4/2x DSl/ 2x DP 1.2
I-USB: I-USB 3.0 +USB 2.0(I-Micro USB)
Abanye :GPIO, l2C, 12S,SPI,UART
Ukuhlinzekwa kwamandla: DC jack (19V)
I-Ethernet :10/100/100OBASE-T iyaguquguquka
Ikhamera: 12-channel MIPI CSI-2 D-PHY 1.2 (30 Gbps)
Ikhadi elingenantambo :802.11ac WIFI + Bluetooth
Ukufaka ikhodi yevidiyo :4K x 2K 60Hz (HEVC)
Ukukhipha ikhodi kwevidiyo :4K x 2K 60Hz (usekelo lwamabhithi angu-12)