Atom-Based JaguarBoard To Take On Raspberry Pi (hothardware.com)

MojoKid writes: The tiny single-board PC movement that's leading the Internet of Things (IoT) market is largely dominated by ARM-based processors, and for good reason — they're cheap, low power and capable. However, what if you prefer to work with the x86 architecture? JaguarBoard looks strikingly similar to Raspberry Pi, which is arguably the most popular single-board mini PC. But unlike Raspberry Pi, JaguarBoard allows users to develop for x86, courtesy of its Intel Atom Z3735G (Bay Trail) foundation. The chip is a quad-core part clocked at 1.33GHz to 1.83GHz with 2MB of L2 cache, offering a fair amount of horsepower for IoT applications. In addition to an Atom processor, JaguarBoard also boasts 1GB of DDR3L memory, 16GB of eMMC storage, three USB 2.0 ports, 10/100M LAN port, HDMI 1.4 output, SDIO 3.0 socket, two COM ports, four GPIO pins, and audio ports. It's an interesting device that you could use strictly as a mini PC for general purpose computing, as an embedded system, a learning or research tool, or for whatever DIY projects you can conjure up. It's not the only hobbyist-appropriate x86 board, but those specs are pretty good for $45.

Comments

Not a Raspberry pi competitor.

[goombah99]

4 gpio ports? this is not competing against a raspberry pi. And if I'm looking for something that is a computing device not a hacker board then I can take my $45 and get a Amazon tablet with USB IO for that whihc includes batteries, and a powersource. then I've got usb I/O or wifi I/o to a CHIP, Arduino or Raspberry pi $5. So it hits the sour spot between being under ported as a hacker board and over priced as a cheap computer.

Re:Not a Raspberry pi competitor.

[Anonymous Coward]

And do any of those solutions you bring up involve x86? That is the main point for this...

Re:Not a Raspberry pi competitor.

[LarryRiedel]

I don't know of a $45 Amazon tablet that can be a standalone off the shelf x86 Ubuntu system with performance comparable to a netbook which can host VMs running Ubuntu Snappy Core for IoT applications. Since the JaguarBoard also has I2C, COM and GPIO ports, it can in some cases be a replacement for an RPi, depending on the number of units to be deployed in production, and the profit margin and TCO of the target solution.

Re:Not a Raspberry pi competitor.

[TheRealMindChild]

Why would anyone care for x86 unless they are trying to run windows?

Re:Not a Raspberry pi competitor.

[Anonymous Coward]

What do you mean? Linux is already available on ARM, Windows isn't. If you just want Linux, you would be better off with one of the more common ARM boards.

Re:Not a Raspberry pi competitor.

[Gaygirlie]

4 gpio ports? this is not competing against a raspberry pi.

That's pretty much what I came in to say: this shit ain't an RPi - alternative, it's just a low-power PC. One of the major selling-points of RPi is the 40-pin GPIO-row, all useable natively, but with this shit you'd have to use I2C or USB GPIO-expanders and jump through extra hoops every single time you wanted to read or set a GPIO-pin state. Things get even worse if you wanted to use SPI.

Re:Not a Raspberry pi competitor.

[rbgaynor]

You can already do all of that on a Raspberry Pi running Linux - and run Wolfram's Mathematica.

Re: Not a Raspberry pi competitor.

[arglebargle_xiv]

For that price - it is useless. I would much rather get a kangaroo for $99

Oh sure, yeah, a 'roo is cute when it's young, but have you ever stopped to think about what happens when it gets older? There are countless millions of loving, adoptable single-board computers put to death every year, simply because people keep making more of them to try to make money, and other people reward them for this by paying them for these SBCs when they could instead save an innocent older SBC from death by adopting.

Many people who buy 'roos later decide it's an inconvenience and get rid of them, adding yet more innocent SBCs to the huge numbers already sent to China and Africa for ecotoxic reprocessing.

If nobody bought, and everyone adopted older SBCs, nobody would manufacture new ones, and eventually, no innocent, adoptable SBCs would die. Nobody wants to admit responsibility and wants to pass the blame, but everyone who abandons, buys or builds new SBCs while they are being burned for the gold in their connectors has blood on their hands.

Remember, a 'roo isn't just for Christmas.

The RPi's "secret weapon"

[petes_PoV]

Is its following, community and wide range of available software.

Without something comparable, all the SBCs in the world amount to very little. For example, consider the Orange Pi. It's based on a different architecture, it uses a different boot-up process. Sure, it runs Linux, it's probably hardware compatible up to a point, it's cheaper: $15 compared to what? $30 for a RPi (I'm not up to date on US dollar prices). Has it taken the world by storm? No. Can you buy it without sending your money to China and waiting 1 - 2 months? Definitely not.

What it, and all the other SBCs, lack is the ease of use. The wide range of almost-working software. The examples to create your own almost-working software. The documentation about what almost works and the "experts" (those people who can make TWO LEDs flash) who can and will answer questions - preferable with correct answers.

Re:The RPi's "secret weapon"

[goombah99]

Yep. I wasted $20 on an Orange Pi PC. Official distro would not boot, eventually found an unofficial on that would boot linux. Only ran on one of it's CPUs it looks like. And not much support for GPIO modes like SPI. it would not work with any of the bog standard monitors I owned. Did get it to work on an HDMI TV but only in certain modes. Problems with KEy boards. Would not recognize some SIM cards despite their meeting specs.

Sure I could make it work, but would I develop for it? no because if I came back a year later I'd probably fiund anything I created would not work on their unsupported releases and there would be a newer board out. With Rpi I can be productive imediately and know I have path forward for anything I make.

Re:The RPi's "secret weapon"

[Walter White]

Is its following, community and wide range of available software.

Really good point. To draw an Arduino analogy, there are other platforms that offer better bang for the buck (e.g. various STM Discovery boards, some of the Cypress boards) But the community and resulting accessibility make the Arduino much more popular.

Re:The RPi's "secret weapon"

[CajunArson]

It doesn't need "community support" when the "community" is the default version of practically every Linux distro in existence without even requiring the existence of a specialized "community" distro just to support one board. That's what makes it really compelling over a Raspberry Pi (and I was in the very first wave of people who ordered it too).

Re:The RPi's "secret weapon"

[mpthompson]

The problem with the Orange Pi and many other inexpensive systems is that they lack mainline Linux support. AllWinner just doesn't seem to be interested in investing the time, effort and money to make it happen.

With these types of systems you may be able to find and boot a relatively new kernel (ie. 3.8 or later), but even if you manage that you find yourself stuck in time as the rest of the Linux community and API's continue to progress. I have clients that have chosen to use hardware only supported by a hacked 2.6.35 Linux kernel and they fail to realize the enormous effort it takes to get newer applications and protocols working in such an environment -- if it is possible at all. Not to mention the security issues of not being able to track the latest software versions.

It's a shame because the hardware is perfectly fine, but with the systems I own and manage, I want to deal with manufacturers that have at least a 5 year horizon for software support with regards to the hardware they produce rather than 5 months which seems typical of the cheap systems out of China.

I'm intrigued by these Intel based systems if they can indeed run generic mainline Linux kernels. If so, it will be well worth the $30 or more price premium.

Re:The RPi's "secret weapon"

[amiga3D]

It looks nice. I notice that the $45 price was for early kickstarter pledges and later it went up to $65. I also notice shipment is delayed until middle February now. I don't see it as a competitor for the Pi though, more like a different niche. The Pi is a throwaway computer. I've purchased a dozen now and have them doing different things. Some A+ models hanging outside with camera mods attached for surveillance, a Pi2 running a media center and one for a file server. They run off a cell phone charger with no problems. The community support makes it almost trivial to set up anything as every time I look there are more projects posted. It's a hobby board, not a peecee competitor. I will say I might pick up a jaguarboard later on to play with if they actually start delivering them. It looks cool.

Re:The RPi's "secret weapon"

[DCFusor]

Yup - rather than mod up an already +5, thought I'd chime in myself. It's the community, and the existing popularity - the network effect, that makes the pi a win. I tasked myself with designing a "LAN of things" for my off-grid homestead, intended to last, well, forever - as long as I'll live. LAN because security, and my life depends on Neuman's little helpers. Will I be able to get a replacement for an also-ran that will go off market the instant it's not a huge success, as has happened many times from this source already? I'm not going to count on it.
Tell me again what advantage there is in X86 if I'm not going to use assembler? Anything running a modern opsys - any of them - will have "issues" doing real time hard-deadline stuff, because the opsys will preempt your code now and then to do its thing - which is why I hang an Arduino Uno or a Teensy off most of the Pi's in my setup - the hardware support for little fiddly bit-bang stuff stinks on all these class of boards, but linux (well, it's all there is, but I'd have chosen it anyway) - and its apps - from NGINX to MySQL, to...you name it, absolutely rocks on a pi.
It's a total no-brainer if I might want to have a hot spare available well down the road. If not new - so many have been sold I'd bet there's even a thriving used market by the time I'd need one.
I don't sell anything, I give what I develop away - GPL or just copyleft, I have no reason to care. Interested people might want to check out my forums under software to see some of what I've managed so far along the LAN of things lines. I like to say that surviving off-grid is actually the oldest profession - you have to be alive to do that other one that claims to be the oldest, after all ;~).
I have nothing against Intel - all my regular PC's are intel, and I like them - including the NUCs (I'm posting from a Haswell one with 2tb of spinner and 500gig of SSD right now). It's not the point, the point is - will they abandon this if it doesn't make money fast? Track record speaks for itself.

Re:The RPi's "secret weapon"

[Gaygirlie]

I ordered an Orange Pi PC on eBay just recently and... boy-howdy do I regret it. It was only after ordering it that I found out that e.g. the CSI-connector isn't actually a CSI-connector and you need some expansion board to make use of it, and the expansion board isn't even supplied with the device itself -- and of course, none of this is mentioned either on the official website or their store! The forums are at the brink of self-immolation, no verification e-mails are ever received by people who try to register accounts there, the official website has a bunch of broken or outdated links, the Steven - guy behind the project has been missing for at least a week now and so on.

The hardware is real nice and personally I find their newly-announced OPi-Lite an exceedingly attractive device, but fuck if they're ever going to attract more than a few stragglers who end up horribly disappointed in the end when things are run like this. I wish I could get an OPi-Lite, but with RPi-level support.

Re:The RPi's "secret weapon"

[Dracos]

I've made 1024 LEDs flash with an rPi. To do that I had to buy a 32x32 LED matrix, power brick, and a proto hat; solder the necessary connections on the hat, then write a C++ application using a very nifty library I found. It showed the date and time, scrolled text, and displayed images.

So what's my level of expertise?

It seems your expectations for Ease of use are too high.

Missing Stuff

[Anonymous Coward]

Missing:

-No commercially available case
-No SPI port - [besides sdio, not accessible]
-No secondary i2c port
-All of "4" gpios - not nearly enough

this is a great XBMC box, but good for little else.

NanoPi2 does more for less

[Okian Warrior]

I just got a couple of NanoPi 2's, they're 1.4 GHz, and have embedded wireless on the board.

For $32, you don't need an octopus of wires to power your wifi USB dongle through a USB hub, both of which you need for the Raspberry PI. A NanoPi2, a $6 USB power supply, a 16GB memory card, and you're ready to go.

Of course, feel free to develop for the X86, because it's *such* an elegant architecture...

The client wanted a system to log (plastic injection molding) machine cycles, so I wrote a script to read the GPIO and make entries to a remote MySQL database. Everything except the glue script was off the shelf and open source. He can use any open source DB viewer and make whatever data views he needs.

You can make an IoT device in an afternoon with one of these.

The x86 SBC's "secret weapon"

[chihowa]

The advantage of an x86 based SBC is the ability to take advantage of the maturity and relative uniformity of the x86 platform. The arcane uboot process and the need for specific support for not only different ARM SoCs, but the specific machines built on them, leaves dozens of abandoned ARM based systems stuck on ancient custom-tweaked kernels (and Linux only). Almost all of the problems you list are inherent in any ARM-based system until the equivalent of a uniform and predictable BIOS-type system is implemented for ARM.

Re:The x86 SBC's "secret weapon"

[Anonymous Coward]

If you want that, it's very important to make sure the Atom-based system is actually a PC architecture system with a normal PC or EFI BIOS. Some of these Atom systems-on-chip have all the same drawbacks as ARM systems-on-chip, namely obscure bootloaders and binary drivers that were hacked into one version of the Linux kernel by some chip vendor. They will never be able to receive a normal Linux distribution to replace the firmware.

In a moment of carelessness I bought an Atom Z2480-based smartphone thinking the x86 chip was going to make tinkering easier in the future. I have no complaints about the performance of the device as a black-box appliance. However, it is so niche that there is no community really hacking firmware for it, and I don't have enough spare time to do all of that myself.

Power?

[stabiesoft]

No mention of power consumption. And as a guy who just dd a pool controller than used 7 GPIO + 4 analog ins, it would not fit the bill. And as a guy about to do an irrigation controller that needs 8 GPIO, again, no dice. I've been settling on beagleboard greens for my projects lately. Low power (2.5W) lots of GPIO, and analog in that is good enough. Raspberry would have worked for irrigation, but price is close, and had an extra beaglebone from initial order in case I blew one up when building the pool controller.

$65 now, how's TDP?

[4wdloop]

The $45 and later $50 was an early kickstarter deal. It's $65 now (or $500 for 10).
1GB ram and 16GB flash makes it a non-windows worthy at the moment (don't even bring up the Win10 IoT gimmick).

The 4 core Atom is a good CPU with a decent GPU (for a small SoC).
But how does this board's TDP compare to Pi or BBB?

Re:$65 now, how's TDP?

[DCFusor]

A killer-important question for my apps. I use a decent (currently around 5) number of pi's on my off-grid solar powered homestead, and they need to be on 24/7/365 to do things like control my power and water systems, collect weather data (internal and external) from buildings on my campus, do security, handle anticipatory HVAC controls, and in general make my life easier more quickly than I get weaker from old age. TDP is huge when there are times like right now - solar panels covered with snow and main batteries getting low.
Being able to segment and prioritize power use when it's a limit - yet utilize all that nature happens to be providing at the moment (be that water, heat, power) is critical to my life. Can't do that if the thing doing it is itself a big load, or too expensive to partition into lots of separate functional units that CAN be shut down selectively when power is low - though it's better to never have to shut data collection down.

.
While mips per watt is really important - plain old watts are too. A setup that can handle a decent peak computational loading had better be good at idle too. Just like solar power - the need for computation is a famine or flood situation. Sometimes you need a lot - sometimes not. With solar, assuming a decently sized system - a system that can handle Feburary in "conserve" mode is dumping so much power on you the rest of the year you might as well be using for an electic car (which I do). A good systems design handles things like this that we can't really change, gracefully.

So close and yet.....

[Lumpy]

Honestly this is not even competition for a RasPi. 4 io ports... that's a fail.
what they should have done is put 2 separate Ethernet ports on this and they would have utterly OWNED the home brew networking device market.

Re:So close and yet.....

[erapert]

Agreed

No WiFi and only 10/100 ethernet... It's like they TRIED to make the device suck. If they're not going to provide either WiFi or 1Gb ethernet then this is just speaks volumes of the competence of the team behind Jaguar

USB 2.0... It's 2016 and this is an x86 device. USB 3.0 really should be standard especially if you're not going to provide SATA or 1Gb ethernet or WiFi. Come on, put something nice on the board.

4 IO ports... Holy moley! They completely missed the point of the IoT movement and SBCs in general.

No SATA... sure, these are hard to find on SBCs, but wouldn't it make sense on an x86 machine?

HDMI port... great job, Jaguar team, you managed to hit one of the the bare minimum requirements.

$65 -- You gotta be kidding me. There's cheaper boards out there with more support, more features, and more performance.

Here, for only ten dollars more you can have a machine that curb-stomps the Jaguar: It's called an ODROID XU4. You're welcome.

The price isn't bad, it's just not great

[drinkypoo]

In addition to an Atom processor, JaguarBoard also boasts 1GB of DDR3L memory, 16GB of eMMC storage, three USB 2.0 ports, 10/100M LAN port, HDMI 1.4 output, SDIO 3.0 socket, two COM ports, four GPIO pins, and audio ports. [...]those specs are pretty good for $45.

Well, they aren't horrible. The onboard eMMC is a boon if it's fast, but it probably isn't. If the CPU is faster and you need that, OK. But two com ports don't make up for only having four GPIO pins, and PineA64+ is not only cheaper (starts at $15 + shipping with 512MB, it's $26 shipped with 1GB the 2GB model is $36 with shipping) but it's got a R-Pi 2 connector and another expansion connector, and GigE, plus some other nice connectors besides. If what you want is the computer, then yeah, that's a cool price. If what you want is something like a R-Pi, then you don't want this.

10/100 still? no, thanks!

[dltaylor]

Although my switches can handle the low-speed devices, I just don't see the point. GbE is has been in SoCs for at least a decade, and the only device in my house not GbE is a rarely-used Wii with a 10/100 USB dongle (if I used it often, it would have the same 10/100/1000 dongle as the Wii U that gets most current use).

I live in a higher-density building and don't run any wireless, except odd, and temporary, occasions for the 'phone, and wish that IT could run USB networking as a simple device, rather than only as a bridge for tethered devices, because my Linux box(es) could easily "tether" it to the home network over USB.