Hacker Board Survey Results: More Raspberry Pi, Please
The results are in for the 2017 Hacker Board survey. A total of 1,705 Linux.com and LinuxGizmos readers voted for their favorite Linux-driven, community-backed SBCs under $200 out of a catalog of 98. As with last year’s jointly sponsored survey, as well as the 2015 and 2014 polls, a Raspberry Pi single board computer came out on top.
What was remarkable this time around was the huge 4-to-1 gap between the Raspberry Pi 3 and the nearest competitor, which for the first time was also a Raspberry Pi: the Raspberry Pi Zero W. Third place went to the revamped, Cortex-A53 based Raspberry Pi 2.
Our official totals are based on Borda Count scoring, in which we tripled the number of first choices for an SBC, then doubled the number of second place selections, and added the two results to the unadjusted third-choice amount. When looking only at first-choice selections, the Raspberry Pi 3 outshined the second-place UDOO X86 by a factor of 7-to-1.
Before the quad-core Raspberry Pi 2 arrived in 2015, it was easy to consider the possibility that some Odroid, UDOO, Banana Pi, BeagleBone, or other contender might come along to give the Pi a run for its money. More powerful, and often more affordable competitors arrived, and most, such as the BeagleBone, offered far better open source hardware support. Yet, the Pi 2 continued to thrive, and the 64-bit RPi 3 took it to the next level.
Even the very competitive Raspberry Pi 3 has been overshadowed by some faster, cheaper, and more feature rich boards, some of which provide the same 40-pin expansion connector. But none offer the guaranteed compatibility of expansion boards, especially the newer HAT add-ons, nor can they match the project’s software support. Perhaps because the Raspberry Pi Foundation is at heart an education-focused effort, the community is also deeper, broader and more grassroots than many of the more vendor-driven projects.
X86 boards gain traction
I’ll take a further look at the Pi phenomenon below, but first let’s examine some other trends reflected in the results. First is the rising popularity of x86 entries, which this year totaled eight boards out of 98 mostly ARM-based designs.
Earlier x86 entries have had middling scores in our surveys, but newer models like Seco’s UDOO X86, Intel’s MinnowBoard Turbot Quad, and Aaeon’s UP Squared and UP Board have drawn considerable interest. Prices are still higher than with ARM boards, but they offer powerful quad-core Intel Atom SoCs, as well as features you don’t usually find on ARM SBCs such as SATA and USB 3.0.
The Intel Edison Kit for Arduino advanced to 18th from last year’s 35th, and the Quark-based Intel Galileo Gen 2 ranked #43. However, both products, including the Intel Edison module itself, are being discontinued, according to a June 19 Hackaday story. Although the discontinuation of these older boards is unsurprising, the story also said that the newer, Atom-based Intel Joule module is also being discontinued.
Another surprise this year was the rebounding popularity of official Arduino boards that also run Linux. Most of these, such as the number 10 ranked Arduino Industrial 101, were in last year’s survey, but didn’t score as well. This also represents a win for the beleaguered MIPS architecture, which forms the foundation of the companion chips on the MCU-based Arduino boards that run OpenWrt or Linino Linux.
Despite the gains for x86 and MIPS boards, ARM boards still dominated the contest. The first non Raspberry flavored SBC on the list is the fourth ranked Odroid-XU4, which is based on an octa-core Samsung Exynos SoC. Other top 10 winners include the industrial-oriented BeagleBone Black, the 96Boards compatible DragonBoard 410c, the pseudo Pi compatible Odroid-C2, and the ninth-ranked Raspberry Pi Zero, which has been eclipsed by the almost identical, but wireless enabled RPi Zero W.
The 11th ranked BeagleBone Black Wireless is one of several BeagleBone clones that did well, including the new robotics-targeted BeagleBone Blue. Other Top 20 models include the Pine A64, the Asus Tinker Board, and Banana Pi BPI-M64, all of which have RPi 40-pin expansion. Also in the 10-20 range are the Arduino Yun and Arduino Tian, the UP Squared and UP Board, and the old Intel Edison Kit for Arduino. Number 19 goes to the Chip Pro Dev Kit -- Next Thing’s sandwich style alternative to the Chip SBC, which is currently out of stock.
The 1,705-person sample, as well as the 90 different countries of origin, suggest the survey is a fairly accurate indicator of consumer SBC popularity. However, SBCs that are more commonly available and popular in east Asia, such as the many Orange Pi and NanoPi boards in our catalog, may have been undercounted. It turns out that SurveyMonkey is blocked in China, which resulted in only eight Chinese respondents, three of which are from Hong Kong. Yet content firewalls probably don’t explain why there were similarly miniscule totals cast from Japan, Korea, and Russia. Readership totals suggest that the numbers should be considerably higher. We’re looking into it.
The Raspberry Pi vs. the IBM PC
Even if there were more voters from China, Japan, Korea, and Russia, it’s hard to imagine that the Raspberry Pi would not have dominated. The many Raspberry Pi pseudo clones with 40-pin connectors, including the Orange Pi, NanoPi, Banana Pi, Odroids, and others did well overall but did not score quite as high as in last year’s survey. This would suggest that the compatibles competition has been neutralized at least for now.
It is difficult to find an exact parallel in computing history to the Raspberry Pi and the consumer SBC market – in part due to the novel open source, hobbyist, and educational nature of the genre. The closest we can think of is the IBM PC market back in the 1980s, which was also marked by a comparatively open platform. Yet while the Raspberry Pi dominates the consumer SBC market, IBM has long been a relatively minor PC vendor, and sold off its ThinkPad line to Lenovo in 2005.
Despite establishing the PC standard, IBM was quickly besieged by competition from cheaper PC compatibles such as the Compaq, and it was ultimately eclipsed by those vendors. There were many reasons for this, including the IBM PC’s high price, IBM’s underestimation of the PC market it created, and the very openness of the platform it created and which was the prime factor behind the PC standard’s success. The openness was not entirely voluntary, as full PC clones were not possible until Compaq reverse-engineered the closed IBM PC BIOS, as fictionally dramatized in the first season of Halt and Catch Fire.
None of these were issues with the Raspberry Pi, which wasn’t even the first community-backed Linux hacker board, having followed the BeagleBoard and others. Although Pi pseudo clones mimic the Raspberry Pi expansion interface and some key features, they aren’t true clones since the Pi’s Broadcom SoC is unavailable to other vendors. In this respect, at least, the IBM PC was more open than the Raspberry Pi, which makes it difficult for software written for the Pi to run without modification on boards with other processors.
Thanks to Linux and other common attributes among today’s hacker boards, however, software usually can be easily ported between different ARM Linux SBCs. Also, software compatibility is not as important in the diverse, often IoT-focused, world of often purpose-specific hacker board projects than with multi-purpose PCs.
The one common theme in the success of both the PC and the Pi is that a generally open hardware platform wedded to a common OS – Windows then instead of Linux/Android now – led to domination over more closed platforms. IBM benefited greatly from this approach for decades even if it had to share the loot with dozens of other companies.
For more analysis, charts, and other details about the 2017 hacker board survey, please see the additional coverage at LinuxGizmos.