Ikhithi yokuthuthukisa i-Jetson Xavier NX
I-NVIDIA Jetson Xavier NX developer suite iletha ukusebenza kwe-supercomputer emaphethelweni. I-suite ihlanganisa imojuli ye-Jetson XavierNX eyenza ukuthuthukiswa kwezinhlelo zokusebenza ze-AI zamamodeli amaningi kusetshenziswa izitaki zesofthiwe ye-NVIDIA ngaphansi kuka-10W. Ukusekelwa komdabu wamafu kwenza kube lula ukuthuthukisa isofthiwe ye-AI nokuyiphakela kumadivaysi asonqenqemeni. I-Developer Suite inaso sonke isitaki sesofthiwe ye-NVIDIA, okuhlanganisa ukusekelwa kwe-SDKS esheshisiwe namathuluzi amasha e-NVIDIA akhelwe ngokuqondile ukuthuthukiswa kohlelo lokusebenza nokuthuthukisa.
Jetson Xavier NX module yokuthuthukisa
Imojula ye-NVIDIA Jetson Xavier NX ingu-70x45mm kuphela ngosayizi futhi iletha ukusebenza kweseva okungu-21 TOPS (15W) noma kufika ku-14 TOPS (10W). Ingasebenzisa amanethiwekhi amaningi esimanje e-neural ngokuhambisana futhi icubungule idatha kusuka kuzinzwa eziningi ezinokulungiswa okuphezulu, ihlangabezana nezidingo zohlelo olugcwele lwe-AI. Ukusekela ubuchwepheshe bomdabu wamafu kwenza kube lula ukuthuthukisa isofthiwe ye-AI nokuyiphakela kumadivaysi asonqenqemeni. Ingasetshenziswa ekukhiqizeni ngobuningi futhi isekela zonke izinhlaka ze-AI ezidumile.
Jetson AGX Xavier development Kit
I-NVIDIA Jetson AGX Xavier inguqulo ethuthukisiwe ye-NVIDIA JetsonTX2 esebenza kangcono izikhathi ezingu-20 kanye nokusebenza kahle kwamandla okuphindwe ka-10 kune-TX2. Isekela i-NVIDIA JetPack ne-DeepStreamSDK kanye nemitapo yolwazi yesofthiwe ye-CUDAR, i-cuDNN, ne-TensorRT, futhi inikeza uhla lwamathuluzi alungele ukusetshenziswa akwenza kube lula futhi kusheshe kubasebenzisi ukudala nokusebenzisa izinhlelo zokusebenza zamarobhothi e-Al. . Okokukhiqiza, ukulethwa, ukudayisa, ezolimo, njll. Nge-Jetson AGX Xavier, ungakha imishini ezimelayo enamandla e-AI engasebenza ngokuncane njengo-10W kuyilapho uzuza kufika ku-32 TOPS. Ingxenye yenkundla yekhompiyutha ye-Al ehamba phambili embonini, i-Jetson AGX Xavier izuza kuhlelo olubanzi lwe-NVIDIA lwamathuluzi e-AI nokugeleza komsebenzi ukusiza onjiniyela baqeqeshe ngokushesha futhi basebenzise amanethiwekhi e-neural.
Amapharamitha we-Jetson Xavier NX suite | |
I-GPU | Isakhiwo se-NVIDIA Volta ene-384 NVIDIA I-CUDA cores kanye nama-Tensor cores angama-48 |
CPU | 6-core NVIDIA Carmel ARM v8.264-bit CPU 6 MB L2+4 MB L36MB L2+4MB L3 |
I-DL accelerator | 2x NVDLA Izinjini |
I-accelerator yombono | I-7-Way VLIW Vision Processor |
Inkumbulo yangaphakathi | 8 GB 128-bit LPDDR4x @51.2GB/s |
Indawo yokugcina | I-Micro SD iyadingeka |
Ukubhala amakhodi kwevidiyo | 2x4K @30|6x 1080p @60|14x 1080p @ 30(H.265/H.264) |
Ukuqoshwa kwevidiyo | 2x4K @60|4x 4K @30|12x 1080p @60 32x1080p @30(H.265)2x 4K @30|6x 1080p @60|16x 1080p @30(H.264) |
Ikhamera | 2x MIP|CSl-2 DPHY imizila |
Inethiwekhi | I-Gigabit Ethernet,M.2 Key E(WiFi/BT kufakiwe),M.2 Key M(NVMe) |
Bonisa isixhumi esibonakalayo | I-HDMI nembobo yokubonisa |
I-USB | 4x USB 3.1,USB 2.0 Micro-B |
Okunye | GPIO,I2 C,I 2 S,SPI,UART |
Ukucaciswa nosayizi | 103x90.5x34.66 mm |
Amapharamitha wemojula ye-Jetson Xavier NX | ||
Igama | 10 W | 15 W |
Al performance | 14 TOPS(INT8) | 21 TOPS (INT8) |
I-GPU | I-384-core NVIDIA Volta GPU ene-48 Tensor Ama-cores | |
I-GPU Max I-Freq | 800 MHz | 1100 MHz |
CPU | 6-core NVIDIA Carmel ARM v8.264-bit CPU 6MB L2+4MB L3 | |
I-CPU Max I-Freq | 2-core @1500MHz 4-core @1200MHz | 2-core @1900MHz 4/6-core @1400Mhz |
Inkumbulo yangaphakathi | 8 GB 128-bit LPDDR4x @1600 MHz 51.2GB/s | |
Indawo yokugcina | 16 GB eMMC 5.1 | |
Amandla | 10W|15W | |
I-PCle | 1x1+1x4 (PCle Gen3,Root Port &Endpoint) | |
Ikhamera ye-CSI | Kufika kumakhamera ayi-6 (36 ngeziteshi ezibonakalayo) 12 imizila MIPI CSI-2 I-D-PHY 1.2(kufika ku-30 Gbps) | |
Ukubhala amakhodi kwevidiyo | 2x464MP/sec(HEVC),2x 4K @30(HEVC) 6x 1080p @60(HEVC) 14x1080p @30(HEVC) | |
Ukuqoshwa kwevidiyo | 2x690MP/sec(HEVC),2x 4K @60(HEVC) 4x 4K @30(HEVC),12x 1080p @60(HEVC) 32x 1080p @30(HEVC) 16x1080p @30(H.264) | |
Bonisa | 2 imodi eningi ye-DP 1.4/eDP 1.4/HDMI 2.0 | |
I-DL accelerator | 2x NVDLA Izinjini | |
I-accelerator yombono | I-7-Way VLIW Vision Processor | |
Inethiwekhi | 10/100/1000 BASE-T Ethernet | |
Ukucaciswa nosayizi | 45 mmx69.6 mm 260-pin SO-DIMM isixhumi |
I-Developer suite I/O | Jetson AGX Xavier |
PCle X16 | I-PCle X16X8 PCle Gen4/x8 SLVS-EC |
RJ45 | I-Gigabit Ethernet |
I-USB-C | Izimbobo ezimbili ze-USB 3.1, izimbobo ze-DP (uma uthanda), kanye nezimbobo ze-PD Ongakukhetha) Sekela ukulungisa iphutha kwesistimu evaliwe futhi ubhale ngembobo efanayo |
Isixhumi esibonakalayo sekhamera | (16)CSI-2 iziteshi |
M.2 Ukhiye M | I-NVMe |
M.2 Ukhiye E | I-PCle x1+USB 2.0+UART (ye-Wi-Fi/LTE)/ 2S+DMIC +GPIOs |
40 pin ehlangene | UART+SPI+CAN+I2C+I2S+DMIC +GPIOs |
Umsindo we-Hd | Isixhumi somsindo we-Hd |
eSTATp+USB 3.0 Uhlobo A | Isixhumi esibonakalayo se-SATA +USB 3.0 ne-PCle x1 bridge (PD+ yedatha yesixhumi esibonakalayo engu-2.5-inch SATA) |
I-HDMI Uhlobo A | I-HDMI 2.0 |
μSD/UFS ikhadi | I-SD/UFS |