NVIDIA Jetson Xavier NX Developer Kit

The $399 Jetson Xavier NX Dev Kit is available now! Full article on JetsonHacks: wp.me/p7ZgI9-38w
0:16 Unboxing
5:14 Jetson Xavier NX Developer Kit Overview
5:31 Jetson Xavier NX Module
8:38 Carrier Board
11:06 Jetson Software
13:12 Dev Kit Setup - Includes Pro Tips!
15:51 Demo Time!
Xavier NX Developer Kit (NVIDIA Store - Jetson) click.linksynergy.com/fs-bin/...
Xavier NX Developer Kit (Amazon): amzn.to/3cyrZAP
Western Digital NVMe SSD: amzn.to/3fGWjeN
Samsung NVMe SSD (faster): amzn.to/2ZcdtLk
Samsung EVO 64 GB micro SD card: amzn.to/2Z39qBc
Dymo Label Maker: amzn.to/35XawQd
These are affiliate links.
Website: jetsonhacks.com
Github: github.com/jetsonhacksnano
Twitter: / jetsonhacks
#JetsonHacks #JetsonXavierNX #GTC2020
As an Amazon Associate I earn from qualifying purchases.

[@Goodwill345]

As always the best and honest review, sold! Thank You

[@JetsonHacks]

Wow, thank you for your kind words. Thanks for watching!

[@robodev6033]

Great job ! I follow you every video since came out this one is also the best quality never go down ! Keep up the great work !

[@JetsonHacks]

Thank you for the kind words, and thanks for watching!

[@SuperMakeSomething]

Excellent video and demo! Can't wait to receive mine to start experimenting with the platform!

[@JetsonHacks]

Thank you for the kind words. Looking forward to your video about it!

[@xjet]

The pace of the unboxing segment was fantastic... I've only just woken up! :-)

[@JetsonHacks]

Good one! Thanks for watching!

[@crazywazy5527]

it is just absolutely amazing that technology has come so far that we can build a computer as powerful as this, as small as this, as low-wattage as this, for as cheap as this. thanks for the informative video!

[@JetsonHacks]

When you compare this to the first Jetson that came out in 2014, it is quite amazing. It's shows you what companies now do with a 1000 people, 2 billion dollars, 3 years and 8 billion transistors. Thanks for watching!

[@90kchan]

thank you for useful videos and demos always.

[@JetsonHacks]

You are welcome! And thanks for watching!

[@ignaciocases]

Wonderful video as always, thanks Jim!!!

[@JetsonHacks]

Thank you for the kind words, and thanks for watching!

[@simonhc3]

Looks good, thanks for video and demo.

[@JetsonHacks]

You are welcome, and thanks for watching!

[@themip-medicalimageprocess8974]

That dude has got to be the most unintentional hilarious character I've ever seen on a tech channel! Great video - really enjoyed it!

[@JetsonHacks]

Glad you enjoyed it. Thanks for watching!

[@dexterpengji2118]

thank you for the great video!

[@JetsonHacks]

You are welcome, and thanks for watching!

[@fkxfkx]

Great Video, thank you. Could you provide a little more info about how you set the demo up and got it going? Was it a manual setup of 4 containers or was there an automated script that did this? TIA

[@JetsonHacks]

You are welcome! The demos are now up on the NVIDIA NGC portal ( ngc.nvidia.com ) and the scripts to run them are on Github on the NVIDIA-AI-IOT account in the jetson-cloudnative-demo repository. Full instructions are on the Github pages README. Thanks for watching!

[@vnagaravi]

How to power up the Xavier with battery

[@lakshay510]

Informative + Entertaining

[@JetsonHacks]

Thank you for the kind words, and thanks for watching!

[@kevin_delaney]

🤣 I love that you keep in the fails, "There's gunna be Riots" 😂🤣😂🤣

[@JetsonHacks]

Raider fans are very passionate, and have been known to be, how should we say, vocal at times. The KC Chiefs are an arch-enemy of the Raider nation. Thanks for watching!

[@kevin_delaney]

@@JetsonHacks Thank you for doing what you do! Your videos are invaluable!!!

[@alexm4890]

Thx for the video

[@JetsonHacks]

You are welcome, and thanks for watching!

[@definty]

I wonder how quick av1 encoder would be? Real-time yet?

[@TheMode911]

I'd like to know graphics performance (opengl or others)

[@ram64man]

Could you use multiple Xavier nx boards in a cluster to run plex /Kodi and nas / thus improving playback and offer just a bunch of disks over usb 3

[@JetsonHacks]

I don't have an understanding as to why running a cluster of them makes a difference.

[@alialdabbagh3684]

Thank you very much for such nice demo, Could you show how to install Realsense D345i on NVIDIA Jetson AGX Xavier with ROS, please?

[@krukhlis]

Thank you! As always, hilarious and helpful! BTW, small question. Considering the fact they have added support for NVMe SSD -- is it now possible to install JetPack OS directly to SSD and forget about sluggish unreliable micro SD cards?

[@JetsonHacks]

Thank you for the kind words. Pricing is down in the joke market, so I am able to buy better quality ones now. I am not quite sure, but we should be able to copy the rootfs to the SSD, and then right after booting from the SD card pivot there. I'm not sure you can boot directly from the NVMe, but pivoting the root makes everything much faster. It's probably worth doing a video about it, me thinks. Thanks for watching!

[@paul_seiler]

@@JetsonHacks yes, pretty please :) As close to boot-from-NVME as possible :D

[@HitAndMissLab]

great intro. thanks

[@JetsonHacks]

You are welcome, and thanks for watching!

[@HitAndMissLab]

@@JetsonHacks Only dishartening thing is that Xavier NX is now over $2,000

[@JetsonHacks]

@@HitAndMissLab Yes, the parts shortages are hitting hard. However, you can buy a Jetson NX production module on a Seeed carrier board in a case (reComputer J2012 - seed studio.com) for ~$699. The carrier board is very much like the dev kit board.

[@alr6111]

@JetsonHacks , is all the inferencing for the voice demo performed in-house within Xavier's resources or is this all pushed out to the Triton Inference Server on another server somewhere else?

[@JetsonHacks]

Everything is running on board the Jetson Xavier NX. No network connection required. Thanks for watching!

[@69MrUsername69]

Is there support for Xavier NX as PCIe End Point? Probably a special carrier needs to be developed? I would appreciate a video on that as well as for AGX Xavier, thanks.

[@JetsonHacks]

Those sound like good questions for the official NVIDIA Jetson Xavier NX forum. Thanks for watching!

[@g.s.3389]

HI, one question, how can I run headless on a jatson board and viewing the results via web browser? if I disable the windows interface I get back important resources and more performance. any idea?

[@JetsonHacks]

Yes. Typically people will SSH into the board to use headless mode. Thanks for watching!

[@amortalbeing]

Thanks a lot. I always wanted to know their differences. and know which one I'd want to buy! So correct me if I'm wrong, it seems the developer kit is a better choice for robot stuff right?

[@JetsonHacks]

You are welcome! The developer kit is a good choice. Thanks for watching!

[@johnny_123b]

Is sd card slot uhs-II? Also how does booting work. Is it nvme accelerated or sdcard? And power modes, is 6core the fastest? What's the point in 2 core ones?

[@JetsonHacks]

This is beyond what can be answered here. You can read the L4T Documentation in the Jetson Download Center for the boot sequence, it uses CBoot. The power modes balance the number of CPUs with the amount of power going to the GPUs in the power budget. If you have CPU intensive tasks, more cores. You can ask specific questions on the official NVIDIA Jetson Xavier NX forum. Thanks for watching!

[@JBoy340a]

Great video. Love the sense of humor. But don't get the reference to not using black electrical tape. Just kidding.

[@JetsonHacks]

Thanks for the kind words, and thanks for watching!

[@EntropyOfTime]

hi was wondering if some could help me figure out how to get the fan on the jetson xavier to turn on ?

[@EntropyOfTime]

the fan*

[@JetsonHacks]

Please ask this question on the official NVIDIA Jetson Xavier forum, where a large group of developers and NVIDIA engineers share their experience.

[@miklosbalogh1642]

I wonder where are the limitations of BERT... Is it possible to load a full E-book and get correct answers the same way ? Maybe to combine several of the same kind of books and still get the most relevant answers ? And what would it take to use all the english text of Wiki articles which is about 60GB uncompressed ?

[@JetsonHacks]

There is a lot of research underway in that area. A lot of data for your model to train on is good. Thanks for watching!

[@SanjayS-bo3ff]

hello i tried this but the display is black and the fan does not run. i even tried a new sd card but it does not boot up kindly help me

[@JetsonHacks]

I'm not sure what you tried. Please ask your question on the official NVIDIA Jetson Xavier NX forum, where people from NVIDIA can help you through any issues you might be having. Thanks for watching!

[@ivodamjanovic2717]

Can you add a google coral m.2 card and benchmark it? If m.2 not possible the usb version.

[@JetsonHacks]

Why would this be any different than using the coral with another computer? Have you tried it?

[@kaitlynracco1206]

Hi, can you recommend any case that can fit this

[@JetsonHacks]

There are a lot of options, from 3D printed DIY to industrial strength. Something like this maybe? amzn.to/2TuuiAW Geekworm makes something similar. You can look under 'Enclosures' on the Jetson Xavier NX wiki: elinux.org/Jetson_Xavier_NX It's probably worth an ask on the official NVIDIA Jetson Xavier NX forum, where a large group of developers and NVIDIA engineers share their experience. Thanks for watching!

[@terryphan7451]

Can you use an PCie 4 Nvme as boot instead of sd card?

[@JetsonHacks]

This question is worth asking in the official NVIDIA Jetson Xavier NX forum. I haven't look closely enough at this area to give a definitive answer. The boot sequence is the same as the Jetson AGX Xavier. Thanks for watching!

[@johndodd9973]

will you be doing a video on booting from an installed m.2 ssd?

[@JetsonHacks]

kzbin.info/www/bejne/kHyYd4yepK-kf8k

[@chetana9802]

super cool

[@JetsonHacks]

Thanks for watching!

[@dracleirbag5838]

Thanks for your great videos. Can you compare the different models for speed for face detect and object detect? Can you show difference in cores and memory and stuff? So far is the xavier agx the best one so far?

[@JetsonHacks]

Thank you for the kind words. I am not a good resource for the information that you are looking for. NVIDIA has published many articles on how fast different Jetson are. As far as machine learning goes, the AGX Xavier is almost 2x faster than the Xavier NX. The Xavier NX is about 10x faster than a TX2, 20x a Jetson Nano. Thanks for watching!

[@dracleirbag5838]

@@JetsonHacks thanks. So it seems like the nx is a good deal because it's cheaper than agx?

[@dracleirbag5838]

@@JetsonHacks I hope I'm not bothering you but do you know a resource for making a cluster out of them? It would be cool.to make a robot using a bunch of Nvidia single boards.

[@JetsonHacks]

@@dracleirbag5838 The NX is a good value in my opinion.You need to determine if you need the extra performance/memory of the AGX, and figure out if its worth the extra money.

[@JetsonHacks]

@@dracleirbag5838 Not sure what kind of resource you are thinking about. You can ask for advice on the official NVIDIA Jetson Xavier NX forum, where a large group of developers and NVIDIA engineers share their experience.

[@karabgut]

Could you explain what is 5 in SPE features, 5 is type of number of input outpus for sensor readings?

[@JetsonHacks]

The Sensor Processing Engine uses an ARM Cortex M5 processor. M5 is a model number. Thanks for watching!

[@karabgut]

@@JetsonHacks Ah, thanks a lot!So, I have to look at SPI,UART, I2C and other pins for sensor right? I can not take the sensor datas coming from any controller from GPIO pins?

[@JetsonHacks]

@@karabgut Not sure what you are asking. There are 40 GPIO pins. By default, the pins are configured with certain special functions (as you mention, SPI, UART, I2C, I2S). However, they can be configured as desired if you are a capable developer. Conceptually, the SPE is an additional micro controller which allows real time control over the pins if desired. In essence, it's like having a micro controller along with the SoC. However, I will note that you need to be a developer to access these functions.

[@karabgut]

@@JetsonHacks Thank you so much, in case of my location goodmorning! :)

[@JetsonHacks]

@@karabgut You're welcome and thanks for watching!

[@malathomas6141]

How does this compare to amd embedded V1605B and v1807b?

[@JetsonHacks]

I do not know. Thanks for watching!

[@opeope12]

I am just wondering if my jetson nano can be used in the jetson xavier carrier board

[@JetsonHacks]

I do not know, but it seems unlikely. Please ask this question on the official NVIDIA Jetson Nano developers forum where a large group of developers and NVIDIA engineers share their experience. Thanks for watching!

[@opeope12]

Thank you. I will be asking them shortly.

[@alr6111]

Is the speech recognition strictly available on the Xavier or can it run on a jetson nano as well.

[@JetsonHacks]

The demos here are available in NVIDIAs NGC, so you can try it out and see! Thanks for watching.

[@alr6111]

@@JetsonHacks This is for the crowd to see to answer my last question, the containers only work for the Xavier NX and AGX. Thanks for the great channel.

[@JetsonHacks]

@@alr6111 Thank you for the kind words, and thanks for the update!

[@screenvy]

Could you use the Nano carrier board with the NX?

[@JetsonHacks]

If you have a Nano Carrier board B1, the NX works with it. Thanks for watching!

[@billyram8582]

Can't thank you enough!

[@JetsonHacks]

You are welcome, and thanks for watching!

[@snwbrdn777]

So true: "Put on the labels slightly crooked so that it can trigger your OCD every time you look at it."

[@JetsonHacks]

I have found that useful. When there's a really important project with a deadline approaching, your mind will tell you that it is *really* time you fix that label above all else. Thanks for watching!

[@sy2532]

I wonder what this means for the TX2 pricing? The TX2 is 399? Why would I buy a TX2 if I could get an Xavier?

[@JetsonHacks]

Why indeed. Thanks for watching!

[@sy2532]

@@JetsonHacks The Nvidia Sign-in for purchasing does not work. 🤣 I can only signon through the developer pages.

[@wolfpackgaming2492]

Can you install steam on it and test out some games ?

[@MrWilliam932]

Is it compatible with the new Raspberry Pi High Quality Camera?

[@JetsonHacks]

There is no driver available yet for the new RPi camera on the Jetsons. Thanks for watching!

[@kerron_]

Nice

[@JetsonHacks]

Thanks for watching!

[@xacompany]

Nice presentation! May I ask where can I buy my own in the U.S.?

[@JetsonHacks]

Thank you for the kind words. Xavier NX Developer Kit (NVIDIA Store - Jetson) click.linksynergy.com/fs-bin/click?id=Y34Ss0fgjnU&offerid=676338.9&type=3&subid=0 Xavier NX Developer Kit (Amazon): amzn.to/3cyrZAP

[@xacompany]

@@JetsonHacks Thanks again!!!

[@jeremyshaw1]

The PoE and backpower headers are unpopulated, but still exist. Next to the RJ45 and power jack, respectively. Section 3.9 of the NX Devkit carrier board Spec pdf from Nvidia notes this and the design requirements for a PoE power solution.

[@JetsonHacks]

Thanks for the information! This is useful. Thanks for watching!

[@aravindhanravichandran1089]

Can we use this for AR, VR application??

[@JetsonHacks]

Depends on what you mean. Magic Leap built their AR headset around the less powerful Jetson TX2.

[@Blipstream]

Hi Jim, thanks for the great content! Could you please recommend a method for remote managing a Jetson product? I can setup a Raspberry Pi with teamviewer for example on a LTE router but the Jetson/Ubuntu environment only lets me remote manage on a local VNC network. The goal of our pilot is to record video through a jetson/ip camera setup at a shopping mall, train an algorithm and do tweaks remotely from home :) In case you have a tutorial or a suggestion please share. Best, Jake

[@Blipstream]

Ps, maybe this is a network setup issue I’m having and it’ll only work through a fixed internet connection + static IP (since LTE addresses always change)

[@JetsonHacks]

Here's a pretty good talk on how you might accomplish a task like that. You can also ask specific questions on the official NVIDIA Jetson Developer forums, where a large group of developers and NVIDIA engineers share their experience. Here's the talk: kzbin.info/www/bejne/foKloo17mLJ8qK8

[@nasalimbu3078]

Complete set

[@JetsonHacks]

Thanks for watching!

[@whothefoxcares]

Nvidia Xavier AGX uses UFS SD card. It UFS as good as RootonUSB or RootonNVME?

[@JetsonHacks]

USB Drives are about 10x faster than SD cards, NVMe about 20x. Thanks for watching!

[@whothefoxcares]

@@JetsonHacks www.samsung.com/us/computing/memory-storage/memory-cards/mb-fa256g-am-mb-fa256g-am/ *UFS micro SD cards are faster than external USB drives* Thanks for reading comment.

[@JetsonHacks]

@@whothefoxcares it sounds like you are more comfortable with that. Good luck on your project!

[@jawadhaidar3931]

Thanks for the video. I am trying to buy one but the price now on amazon is 1500$, where can I find it at 399$

[@JetsonHacks]

Like many other electronic devices currently, the microchip shortage has tightened the Xavier NX supply. NVIDIA is working on it, but has no timeline yet of when stock will be available. Thanks for watching!

[@jawadhaidar3931]

@@JetsonHacks okay thanks a lot

[@johnmoore4248]

Can I use that module on nano b1 board?

[@JetsonHacks]

I do not know. This is a good question to ask on the official NVIDIA Jetson Xavier NX forum, where a large group of developers and engineers share their experience. Thanks for watching!

[@johnmoore4248]

@@JetsonHacks hi thanks you for your reply. I just found on the product page and said it will support the xavier product that will launch on May. I hope this is the product that they said.

[@jorgemiar]

I thought the spec sheet originally said power supply was 5V?

[@JetsonHacks]

The Xavier NX module works on the Jetson Nano B1 board, which is 5V. The carrier board for the Jetson Xavier NX Developer kit works on 9V - 19V. Thanks for watching!

[@krukhlis]

@@JetsonHacks So am I right I can buy Xavier NX module only, grab one of my Jetson Nanos, kick out Jetson Nano module and put Xavier NX module into Jetson Nano mother board instead? If yes -- Is there anything else beside NVMe that I will be missing with such setup( e.g. it will work slower comparing to full blown Jetson Xavier NX , because of underpowered power input)?

[@mattizzle81]

Hmm, my Jetson Nano and Raspberry Pi 4 are both collecting dust because they were too slow and underpowered for me. Maybe, just maybe this wouldn't be an expensive dust collector for me? Not sure.

[@JetsonHacks]

It depends on your use case, but this is a different class of machine entirely. It's much faster than a Jetson TX2, and when running from a SSD feels nearly as fast as an Jetson AGX Xavier. Thanks for watching!

[@energyideas]

Try updating your Rpi OS much improved.

[@whothefoxcares]

slow automobiles become collectors items. slow computers become unemployed, in the gutter, without a source of power or knowledge. some turn to mining bitcoin.

[@JetsonHacks]

@@whothefoxcares Yes, though few consumer items actually get to be old. Usually they are destroyed or recycled when they are reach the end of their "usefulness". Ford Model Ts are collector items now, but millions were relegated to the scrap heap as just old, worn out cars. Some survive. There are some original Apple Macintoshes people collected, but most went to the trash or recycle bins.

[@josephtremblant2173]

Try booting from SSD with RPI 4, get a good aluminium case with dual fans and quality heatsinks so you can OC to 2.147Ghz and last but not least, use preferably a drive like Samsung T5 external SSD or a Samsung evo 860 SSD drive with the right usb to Sata adapter(w/ UAS and Trim support on raspbian) to boot up from RPI 4. FYI, the difference in reading and writing speeds between a Samsung T5, Samsung Evo 860 and Crucial MX500 are abysmal compare to generic SSD's. For Example. Samsung T5 delivers 315 MBps reading/writing speeds on average and a generic SSD drive goes between (30-120 MBps)

[@AltMarc]

They still didn't move from micro usb to usb c (PD), even when the AGX have it.

[@JetsonHacks]

I'm not sure USBC PD can provide enough power. The dev kit can draw 5A @ 19V. My understanding is that USB-C PD only delivers that in 20V mode. Thanks for watching!

[@AltMarc]

@@JetsonHacks Usb C-PD works for the AGX, tested with a power bank, but there you only get 3.42A 19V from the power supply. Did the NX have a bigger PS ?

[@LilSarge]

Got a like for the RAIIIDERSSSS

[@JetsonHacks]

The Raider Nation!

[@citysnap6136]

playback at X 1.5 speed

[@JetsonHacks]

Always a good idea. Thanks for watching!

[@jorgemiar]

Could you try and set up realsense? 😁😁See if the D435 works

[@JetsonHacks]

Here's a review of someone using it with the Xavier NX: kzbin.info/www/bejne/gXe2mJ6gg9argq8

[@jorgemiar]

@@JetsonHacks Great timing! Thanks!

[@patrickpoirier1877]

@@JetsonHacks it is bare n=minimum installation with no kernel module patch of the UVC driver so the camera is not working properly

[@mehdiozel517]

So, as Nvidia DLA works we get much better FPS. WoW.

[@ubacow7109]

I wish they were a lot cheaper, I wanna get one to work on but its so exp~

[@neliasoares6217]

ARM IS THE FUTURE

[@JetsonHacks]

Thanks for watching!

[@memeplayer1923]

Would you try build some pure-Nvidia builds by shoving GeForce graphics cards onto Jetson Nano/Xavier NX boards with use of EXP GDC docks? A Xavier NX with RTX 2060 would be insanely dope. But a Jetson Nano with GTX 1080 Ti would be a bottleneck show.

[@JetsonHacks]

I think there are two issues. First, how do you power the graphics card, and second (and the show stopper) is the graphics driver for the card. I don't believe the graphics cards have ARM drivers. Thanks for watching!

[@memeplayer1923]

@@JetsonHacks How to power graphics cards? You can try have PSU or laptop power supplies to connect to them, if your GPU is connected via EXP GDC or something similar. And for the graphics card driver... You may try install Windows 10 for Jetson devices. There were already some Jetson devices installed with Windows 10.

[@josephtremblant2173]

LattePanda Delta 432(4Gb Ram and 32Gb storage) for $188 and Odroid-H2($111 without Ram and storage) are much better options if you are looking for budget friendly options.

[@JetsonHacks]

Not quite sure how you do your comparison, other than $$$. Thank you for sharing your opinion, and thanks for watching!

[@AdmV0rl0n]

399... ouch. At this price I am sad to say they need to be shipping a case. Too rich for my blood.

[@JetsonHacks]

Thank you for sharing your opinion, and thanks for watching!

[@crazywazy5527]

when compared to the relatively weak performance and specs of the Jetson Nano, it's well worth the 4x cost increase. can't say I don't relate though lol

[@energyideas]

Buy twelve Rpi4 2gigs for equivalent price at Microcenter.

[@sethloh6060]

Rofl that fire extinguisher, have ye' got no faith in newfangled electronics

[@JetsonHacks]

I trust the electronics, but not myself. My high school chemistry teacher told me that the way I do experiments, I should always keep a fire extinguisher beside me. He was not wrong. Thanks for watching!

[@cyberfication]

I've the Jetson Nano. it is a great dev kit. Now it big brother on steroid came out. I am very disappointed, not by Xavier NX Developer Kit. I couldn't be able to buy it, you know due to the covid-19 austerity measure put in place here in Ethiopia. I think I'll never be able to have it a years time. That is to bad. I'm jealose of anyone who have Xavier NX Developer Kit 😞😟😧 anyway, I like you channel, keep it up, thank you

[@JetsonHacks]

You are welcome, and thanks for watching!

[@danam579]

Fire extenguisher?

[@JetsonHacks]

Yes. My HS science teacher brought over a fire extinguisher to my lab desk one day and said, "The way that you do experiments, you should always keep one of these next to you". He was not wrong. Thanks for watching!

[@lashlarue59]

Is the attraction of this the ability to have a complete self contained prepackaged machine learning environment? I'm just trying to understand where something like this fits for development verses a PC with a RTX GPU in it.

[@JetsonHacks]

Apples and oranges. The Jetsons in general are for what is called 'edge' devices, where low power consumption and form factor are important. Think of devices that run off of batteries, robots for example. There are a myriad of devices which require machine learning, but don't have space for an entire PC/laptop. The general gist is that you would train you machine learning models on a desktop like you mention or in the cloud, and then deploy on a Jetson, which specialize in inferencing. Thanks for watching!

[@lashlarue59]

@@JetsonHacks Thanks a lot! That answers my question.

[@notlessgrossman163]

Kansas city Chiefs.. no it's 42..

[@danelectr]

I see switch pro

[@weerobot]

****Warning Bad Dancing****

[@sampletext69420]

i thought its 512 cuda cores

[@JetsonHacks]

That is not my understanding. There are 384 CUDA cores in the Xavier NX (along with 48 tensor cores, and 2 Deep Learning Accelerators). The device you are probably thinking about is the Jetson AGX Xavier, which has 512 CUDA cores, 64 tensor cores and 2 DLAs. Thanks for watching!

[@sampletext69420]

@@JetsonHacks should i buy it for self-driving develop?

[@energyideas]

Don't use Jetson. They don't update OS. 20.04 should be usable, but they stay way behind.

[@JetsonHacks]

Thank you for sharing your opinion. You should use another product which has this level of performance. Thanks for watching!

[@DDoseOfInspiration]

Why they made it so expensive ?..

[@memeplayer1923]

As expensive as RTX 2080? Yeah, this board has workstation-based Volta architecture to make it more expensive than a consumer-based Turing architecture. The Nano used outdated Maxwell architecture that was last used on GeForce 900 series. Wonder why the NX don't use Turing, though.

[@JetsonHacks]

8 Billion transistors ain't cheap. Not sure what you are comparing it to.

[@DDoseOfInspiration]

@@JetsonHacks I understand but this product could have been used in many projects if they could make it 120-150$ range there is one nano it's 99$ but in the mid range there is nothing as far as I know ...

[@brianbai9385]

OCD label lol

[@JetsonHacks]

Thanks for watching!

[@standardtechno-salesengine4498]

Time pass parson Slower than tortise

[@JetsonHacks]

Thank you for sharing your opinion, and thanks for watching!

[@josephtremblant2173]

Impressive SBC board by Nvidia. However, IMMO it doesn't justify a price tag of $399 USD. Getting a case, 512 Gb NvMe drive, taxes and shipping it could easily climb to $600 USD. I could buy 3 lattepanda SBC's for that price. Way overpriced like everything nvidia does, specially video cards.

[@JetsonHacks]

Thank you for sharing your opinion. It sounds like you should get 3 LattePanda for your projects since they meet your needs. To me it appears that the least expensive LP with 8GB of memory is $379, so I am not quite sure what you are comparing. Thanks for watching!

[@sonukushwah221]

your video very slow you are taking whole day for making video

[@JetsonHacks]

Thank you for sharing your opinion. I don't know what it means, but I appreciate you taking the time to share it. Thanks for watching!

[@ericandrews4861]

You have to be stupid to buy that thing. Way over priced. Just buy a full sized computer.

[@JetsonHacks]

I do not follow your point. How do you fit a "full sized computer" on a palm sized drone, or in a small, smart camera? This is a development kit for the same types of chips that many automobile manufacturers use for their infotainment and ADAS systems. These systems do not have room for 'full size computers'. It might just be that you are not a member of the intended market for the device. Your single variate analysis based on cost obviously does not match everyones product development needs. BTW, there are world class smart people using these, so your thoughtful analysis may not be correct.