The Nvidia Jetson TX1: It’s Not For Everybody, But It Is Very Cool

Last week, the Nvidia Jetson TX1 was released. This credit card-sized module is a ‘supercomputer’ advertised as having more processing power than the latest Intel Core i7s, while running at under 10 Watts. This is supposedly the device that will power the next generation of things, using technologies unheard of in the embedded world.

A modern day smartphone could have been built 10 or 15 years ago. There’s no question the processing power was there with laptop CPUs, and the tiny mechanical hard drives in the original iPod was more than spacious enough to hold a library of Napster’d MP3s and all your phone contacts. The battery for this sesquidecadal smartphone, on the other hand, was impossible. The future depends on batteries and consequently low power computing. Is the Jetson TX1 the board that will deliver us into the future? It took a hands-on look to find out.

The Nvidia Jetson TX1 and Carrier Board

What is the TX1

The Jetson TX1 is a tiny module – 50x87mm – encased in a heat sink that brings the volume to about the same size as a pack of cigarettes. Underneath a block of aluminum is an Nvidia Tegra X1, a module that combines a 64-bit quad-core ARM Cortex-A57 CPU with a 256-core Maxwell GPU. The module is equipped with 4GB of LPDDR4-3200, 16GB of eMMC Flash, 802.11ac WiFi, and Bluetooth.

This module connects to the outside world through a 400-pin connector (from Samtec, a company quite liberal with product samples, by the way) that provides six CSI outputs for a half-dozen Raspberry Pi-style cameras, two DSI outputs, 1 eDP 1.4, 1 eDP 1.2, and HDMI 2.0 for displays. Storage is provided through either SD cards or SATA. Other ports include three USB 3.0, three USB 2.0, Gigabit Ethernet, a PCIe x1 and PCIe x4, and a host of GPIOs, UARTs, SPI and I2C busses.

The only way of getting at all these extra ports is, at the moment, the Jetson TX1 carrier board, a board that is effectively a MiniITX motherboard. Mount this carrier board in a case, modify a power supply and figure out how to wire up the front panel buttons, and you’ll have a respectable desktop computer.

This is not a desktop computer, though, and it’s not a replacement for a Raspberry Pi or Beaglebone. This is an engineering tool – a device built to handle the advanced robotics work of the future.


No tech review would be complete without benchmarks, and since this is an Nvidia board, that means a deep dive into the graphics performance.

The review unit Nvidia sent over came with an incredible amount of documentation, pointing me towards GFXBench 4.0 Manhattan 3.1 (and the T-rex one) to test the graphics performance.


In terms of graphics performance, the TX1 isn’t that much different from a run-of-the-mill mobile chipset from a few years ago. This is to be expected; it’s unreasonable to expect Nvidia to put a Titan in a 10 Watt module; the Titan itself sucks up about 250 Watts.

test suiteWhat about CPU performance? The ARM Cortex A57 isn’t seen very much in tiny credit-card sized dev boards, but there are a few actual products out there with it.  The TX1 isn’t a powerhouse by any means, but it does trounce the Raspberry Pi 2 Model B in testing by a factor of about three.

Compared to desktop/x86 performance, the best benchmarks again put the Nvidia TX1 in the same territory as a middling desktop from a few years ago. Still, that desktop probably draws about 300 W total, where the TX1 sips a meager 10 W.

This is not the board you want if you’re mining Bitcoins, and it’s not the board you should use if you need a powerful, portable device that can connect to anything. It’s for custom designs. The Nvidia TX1 is a module that’s meant to be integrated into products. It’s not a board for ‘makers’ and it’s not designed to be. It’s a board for engineers that need enough power in a reasonably small package that doesn’t drain batteries.

With an ARM Cortex A57 quad core running at almost 2 GHz, 4 GB of RAM, and a reasonably powerful graphics card for the power budget, the Nvidia TX1 is far beyond the usual tiny Linux boards. It’s far beyond the Raspi, the newest Beagleboard, and gives the Intel NUC boards a run for their money.

That huge and heavy heatsink is useful; while benchmarking the TX1, temperatures were only one or two degrees above ambient
That huge and heavy heatsink is useful; while benchmarking the TX1, temperatures were only one or two degrees above ambient

In terms of absolute power, the TX1 is about as powerful as a entry-level laptop from three or four years ago.

The Jetson TX1 is all about performance per Watt. That’s exceptional, new, and exciting; it’s something that simply hasn’t been done before. If you believe the reams of technical documents Nvidia granted me access to, it’s the first step to a world of truly smart embedded devices that have a grasp on computer vision, machine learning, and a bunch of other stuff that hasn’t really found its way into the embedded world yet.

Alexnex images processed per second per watt. No, Joules do not exist.
Alexnex images processed per second per Watt. No, Joules do not exist.

And here lies the problem with the Jetson TX1; because a platform like this hasn’t been available before, the development stack, examples, and community of users simply isn’t there yet. The number of people contributing to the Nvidia embedded systems forum is tiny – our Hackaday articles get more comments than a thread on the Nvidia forums. Like all new platforms, the only thing missing is the community, putting Nvidia in a chicken and egg scenario.

This a platform for engineers. Specifically, engineers who are building autonomous golf carts and cars, quadcopters that follow you around, and robots that could pass a Turing test for at least 30 seconds. It’s an incredible piece of hardware, but not one designed to be a computer that sits next to a TV. The TX1 is an engineering tool that’s meant to go into other devices.

Alternative Applications, Like Gamecube

With that said, there are a few very interesting applications I could see the TX1 being used for. My car needs a new head unit, and building one with the TX1 would future proof it for at least another 200,000 miles. For the very highly skilled amateur engineers, the TX1 module opens a lot of doors. Six webcams is something a lot of artists would probably like to experiment with, and two DSI outputs – and a graphics card – would allow for some very interesting user interfaces.

That said, the TX1 carrier board is not the breakout board for these applications. I’d like to see something like what Sparkfun put together for the Intel Edison – dozens of breakout boards for every imaginable use case. The PCB files for the TX1 carrier board are available through the Nvidia developer’s portal (hope you like OrCAD), and Samtec, the supplier for the 400-pin connector used for the module, is exceedingly easy to work with. It’s not unreasonable for someone with a reflow toaster oven to create a breakout for the TX1 that’s far more convenient than a Mini-ITX motherboard.

Right now there aren’t many computers with ARM processors and this amount of horsepower out now. Impressively powerful ARM boards, such as the new BeagleBoard X15 and those that follow the 96Boards specification exist, but these do not have a modern graphics card baked into the module.

Without someone out there doing the grunt work of making applications with mass appeal work with the TX1, it’s impossible to say how well this board performs at emulating a GameCube, or any other general purpose application. The hardware is probably there, but the reviewers for the TX1 have been given less than a week to StackOverflow their way through a compatible build for the most demanding applications this board wasn’t designed for.

It’s all about efficiency

Is the TX1 a ‘supercomputer on a module’? Yes, and no. While it does perform reasonably well at machine learning tasks compared to the latest core-i7 CPUs, the Alexnet machine learning tasks are a task best suited for GPUs. It’s like asking which flies better: a Cessna 172 or a Bugatti Veyron? The Cessna is by far the better flying machine, but if you’re looking for a ‘supercomputer’, you might want to look at a 747 or C-5 Galaxy.

On the other hand, there aren’t many boards or modules out there at the intersection of high-powered ARM boards with a GPU and on a 10 Watt power budget. It’s something that’s needed to build the machines, robots, and autonomous devices of the future. But even then it’s still a niche product.

I can’t wait to see a community pop up around the TX1. With a few phone calls to Samtec, a few hours in KiCad, and a group buy for the module itself ($299 USD in 1000 unit quantities), this could be the start of something very, very interesting.

An Improvement to Floating Point Numbers

On February 25, 1991, during the eve of the of an Iraqi invasion of Saudi Arabia, a Scud missile fired from Iraqi positions hit a US Army barracks in Dhahran, Saudi Arabia. A defense was available – Patriot missiles had intercepted Iraqi Scuds earlier in the year, but not on this day.

The computer controlling the Patriot missile in Dhahran had been operating for over 100 hours when it was launched. The internal clock of this computer was multiplied by 1/10th, and then shoved into a 24-bit register. The binary representation of 1/10th is non-terminating, and after chopping this down to 24 bits, a small error was introduced. This error increased slightly every second, and after 100 hours, the system clock of the Patriot missile system was 0.34 seconds off.

A Scud missile travels at about 1,600 meters per second. In one third of a second, it travels half a kilometer, and well outside the “range gate” that the Patriot tracked. On February 25, 1991, a Patriot missile would fail to intercept a Scud launched at a US Army barracks, killing 28 and wounding 100 others. It was the first time a floating point error had killed a person, and it certainly won’t be the last.

Continue reading “An Improvement to Floating Point Numbers”

$50 Multimeter Comparison and Teardown

We remember when buying even a modest digital multimeter was a big investment. These days, you can find tool stores giving away cheap meters and if you are willing to spend even a little money, you can buy a meter with tons of features like capacitance, temperature, and other measurements.

Like most things, though, you can pay a little money for a bargain, or you can overpay for a dud. To help you pick, [TechnologyCatalyst] decided to do an extensive video review of 15 different meters in the under $50 price category.

If you are looking for a quick video to watch, you might want to move along. The review is in nine videos ranging from an introduction, to a comparison of build quality, discussion about the displays on each meter, and, of course, the measurement capability of each meter. There’s even a video that shows tear downs so you can see inside the instruments.

Continue reading “$50 Multimeter Comparison and Teardown”

Review: Single Board 65C02 and 65C816 Computers

The 6502 is a classic piece of computing history. Versions of this CPU were found in everything from the Apple ][, to the Nintendo Entertainment System, and the Commodore 64. The history of the 6502 doesn’t end with video games; for the last forty years, this CPU has found its way into industrial equipment, medical devices, and everything else that doesn’t need to be redesigned every two years. Combine the longevity of the 6502 with the fact an entire generation of developers first cut their teeth on 6502 assembly, and you have the makings of a classic microprocessor that will, I’m sure, still be relevant in another forty years.

The cathedral of The 6502 is Western Design Center. For more than 35 years, WDC has been the home of 6502-related designs. Recently, WDC has been interested in the educational aspects of the 6502, with one of the VPs, [David Cramer], lending his time to an after-school club teaching opcodes.

The folks at WDC recently contacted me to see if I would give their hardware a close look, and after providing a few boards, this hardware proved to be both excellent. They’re great for educators adventurous enough to deviate from the Arduino, Processing, and Fritzing zeitgeist, and for anyone who wants to dip their toes into the world of 65xx development.

Continue reading “Review: Single Board 65C02 and 65C816 Computers”

Review: Stickvise Needs a Place on Your Bench

Stickvise is a simple device for a simple problem. It holds a work piece while you work on it. Most obviously this means a PCB for soldering, but there’s a twist of versatility that will make it work for a wide range of needs. Being someone who has often used the roll-of-solder-to-hold-a-circuit-board-down trick, only to upset the apple cart when I run out solder, this is a great little tool to have within arm’s reach. For those that already have a PCB vise, how often do you need more than one? How rarely do you need something that large? And if you’re lucky enough to have a microscope for soldering this is a perfect fixture for moving a board to and from without adjusting the focus.

Details of the Design

Simplicity. This is three pieces of aluminum bar-stock, some steel rod, nylon jaws, two springs, and some fasteners. It all works extremely well. To load up a new circuit board I loosen the wing nut and squeeze the clamp shut. Hand tightening the nut doesn’t take much force and it hasn’t slipped for me at all despite moving it around the bench for several days. Once set, the board can be taken out and flipped over easily thanks to the springs.

The extensibility here is key. As it stands, the nylon jaws have a V-groove to hold a board. If you need to support much taller boards you can always put some standoffs between the aluminum and the nylon jaws.

stickvise-custom-jawsBetter yet is the ability to design jaws for your own needs. [Alex Rich], Stickvise’s creator, already has a number of STL files available so that you may print out your own. The “fingers” on the custom jaw shown here interlock with the ones on the opposite side. But my favorite is an articulated set of “third-hand” style jaws based off of the PCB probe jig [Anool] covered back in May. There are even plans to make a parametric STL file so that printing larger or taller jaws doesn’t require a CAD modeling session.

If the range of the vise is too narrow you can simply replace the center bar with a longer one (source yourself or purchase from [Alex]) — the fixed aluminum end is secured with a set screw. This can even be used as a type of stretcher by reversing the spring jaw. I couldn’t think of an application in my own shop for this but you never know.

Stickvise Roots

stickvise-hackaday-approvedIf you have an eagle eye you’ll have noticed the Jolly Wrencher with “Hackaday Approved” next to it on the Stickvise. When [Alex Rich] started refining his original design he posted about it as a project on It didn’t take long to grab our attention and, after tossing around the idea a bit we approached [Alex] about his plans for manufacturing and how Hackaday might figure into that. I love seeing hardware come to life like this; it puts an artisanal spin on the things I choose to have in my lab.



It’s so simple you could build it, but for me the production quality is well worth buying it instead. It’s simple and durable, with the ability to be specialized for a number of different purposes. I wish I had had it when populating the board I’ve been showing off in these pictures (the LayerOne Badge from this year). If you do any work with circuit boards at the bench the stickvise is a solid entry on your must-have-tools list.

The Stickvise is available in the Hackaday Store.

Review: DSLogic Logic Analyzer

Logic analyzers historically have been the heavy artillery in an engineer’s arsenal. For many of us, the name invokes mental images of large HP and Tektronix iron with real CRT screens. Logic connections were made through pods, with hundreds of leads weaving their way back to the test equipment. The logic analyzer came out when all else failed, when even a four channel scope wasn’t enough to figure out your problems. Setting them up was a pain – if you were lucky, the analyzer had a PC keyboard interface. If not, you were stuck typing your signal names into the front panel keyboard. Once setup though, logic analyzers were great at finding bugs. You can see things you’d never see with another tool – like a data bus slowly settling out after the read or write strobe.

There have been a number of USB based logic analyzers introduced in recent years, but they didn’t really catch on until Saleae released their “Logic” line of devices. Low cost, high-speed, and easy to use – these devices were perfect. They also inspired an army of clone devices based upon the same Cypress Semiconductor parts. DSLogic designed by DreamSource Labs, can be thought of as an open source evolution of the original Saleae device.

DSLogic appeared in 2013 as a Kickstarter campaign for an open source logic analyzer with an optional oscilloscope extension. I think it’s safe to say that they did well, raising $111,497 USD, more than 10 times their initial goal of $10,000 USD. These days both the DSLogic and the oscilloscope extension are available at The Hackaday Store. In this review we’re focusing on the logic analyzer portion of the tool. 

Click past the break for the full story!

Continue reading “Review: DSLogic Logic Analyzer”

Review: Transistor Tester

Amazon has been getting creepier and creepier lately with their recommendations.  Every time I log in, I’m presented with a list of new Blinky LEDs, Raspberry Pi accessories, Arduino shields, and the like. It’s as if they know me. Their customer database paid off when it recommended a $22 transistor / component tester. I’ve been seeing those testers around quite a bit lately. Curiosity got the better of me and my mouse found its way to the “Buy it now with one click” button. Two days later I had a “SainSmart Mega328 Transistor Tester Diode Triode Capacitance ESR Meter MOS/PNP/NPN L/C/R” in my hands.

I’m going to get the obvious out of the way. This thing is built cheap – as cheap as the factories can make it. My particular unit arrived with the LCD flapping in the breeze, hanging on by its flex cable. Fitting the LCD back into the acrylic backlight frame revealed a slightly worrisome twist in that same flex. Thankfully nothing was actually damaged, though I do want to give the flex cable some protection in the future. More on that later. The circuitry was open for all the world to see on the bottom of the tester. The heart of the unit is an ATmega328. Supporting it are a few transistors and a handful of passives.

I didn’t have huge expectations for the tester, but I hoped it would at least power up.  Hooking up a 9 volt battery and pressing the magic button brought the tester to life. Since I didn’t have anything in the socket, it quickly lit up and displayed its maker information – “”, and “By Efan & HaoQixin”, then it informed me that I had “No, unknown, or damaged part”.

I had a few resistors lying around the bench (doesn’t everyone?) so I put one in. The tester read it as 9881 ohms. Sure enough, it was a 10K 5% resistor.  Capacitors – ceramic disc, electrolytic, and surface mount all worked as well. The tester even provided ESR values. The real test would be a transistor. I pulled an old  2N2222 in a TO-18 metal can, and popped it in the tester. The damn thing worked – it showed the schematic symbol for an NPN transistor with Collector, Base, and Emitter connected to Pins 1,2,and 3 respectively. Flipping the pins around and re-testing worked as well. The tester showed hFe as 216, and forward voltage as 692 mV, both reasonable numbers for a 2N2222.

triacThe tester worked surprisingly well – it was able to correctly identify BJTs, FETs, even esoteric parts. The only thing it balked on was a linear voltage regulator, which showed up as two diodes. Regulators are a bit more than a simple device though, so I can’t blame the tester there.  The values returned were all reasonable as well. While I don’t have a calibrated lab to check against, the numbers lined up with my Fluke meter.

So what exactly is driving this little tester? There are about 20 versions of it on the market, all of them from China. 91make is a seller on, often referred to as “China’s ebay.” 91make’s front page features no less than 7 versions of the transistor tester, with various cases and LCDs. Some digging turned up the history on this device. It turns out the transistor tester is an open source hardware project (translated) originally created by [Markus Frejek], and built upon by [Karl-Heinz Kubbeler] and a number of others. The Subversion repository  for the project shows it is quite active, with the most recent check-in only a few hours ago. The project is also well documented. The English PDF is 103 pages, explaining theory of operation, the circuit itself, and the software. The document even explains some of the shortcomings of the Chinese versions of the tester, including using a zener diode where the original schematic calls for a precision 2.5V reference. Yes, it will work, but it won’t be as accurate as the original.

The devs also don’t officially support the clones which I can understand, considering the quality and changes in design each manufacturer is baking in to their own version. There is  a huge thread on the EEVblog forum covering these testers. Some can be modified to be closer to the official version. In fact, with an ISP tool the intrepid hacker can update the firmware to the current rev from [Karl-Heinz’s] repository.

So the final verdict on this tester is that it is a thumbs up with a small caveat. These testers are built down to a cost (and that cost is as close to zero as possible). They’re great for sorting parts, but they’re no substitute for a higher quality measuring device. I’d also love to see a version that supports the original developers.