Hacker News Reader

Comment by kardianos | Hacker News Reader

FYI, as someone who worked with the Edison I can't say I'm surprised. Flashing the Edison was nearly impossible and a big pain. The hardware routinely crashed. Much of the low power "specs" came from (overly) aggressive power management which introduced momentary delays and pauses. I'm happy to see them go and never want to work on such a platform again. The GPIO stopped working reliably at higher speeds despite their spec claims.

By contrast, the raspberry pis and even the Ci20 are significantly more stable and easier to work with. Their specs far more truthful.

View on HN

I was tangentially involved in a project that invested quite heavily in developing a device based on Edison. I don't recall hearing about any hardware problems you describe. In fact, I remember being surprised at how easy it was to setup the flashing tools on my computer (those might have been some in-house developed scripts though). What was a significant hurdle in development was Intel's lack of support and the outdated kernel in their distro.

I think when it came out Edison was quite nice. You got wireless, flash and a decent CPU in a very small package. The only really bad thing from the hobbyist perspective was the fine-pitch connector that was impossible to solder by hand. It made any DIY project completely dependent on Intel's expensive break-out boards. Yocto Linux also seemed more oriented towards serious products than random hacking (especially the "build a firmware image" approach vs. Raspberry Pi's "ssh in and apt-get stuff")

I won one from Sparkfun and just recently started using it. I have some experience developing on embedded platforms.

The unit I have was a pain to flash the first time. I guess something got corrupted at some point and I had to recover with a very unreliable process from a Debian box. That said after the first flash, everything install wise has been wonderful.

The development environment is great for my purposes, however setup was non-trivial. Had I not been comfortable in eclipse I doubt I could have gotten the ld flags set correctly, or changed the c++ std for the compiler.

I love being able to upload code over wifi.

Their price points were always too high. This is a perpetual problem with Intel--they seem terrified of undercutting themselves and can't compete at the low end. This is why ARM is winning.

I tried using the Galileo when it came out and was similarly disappointed. It claimed Arduino compatibility but had IO so slow that it couldn't interface with a DHT11 temperature/humidity sensor. In the end I got the feeling that Intel hadn't really thought things through :/

I had exactly the same problem ... I was disappointed especially since the DHT11 is something like a super common standard sensor that is used in nearly every maker 101 course.

You understate the issue-- the GPIO pins had a default throughput of 230 Hz [https://communities.intel.com/message/207904#207904]

Yeah, we tried them for some projects at our school, and had to daisy chain a real Arduino to do the io. Waste of time. This was Intel's attempt at bootstrapping a community by giving heaps of free kit to schools; ended up making us wary and distrustful of Intel.

You think they'd try to .. fix their platform, instead of ditching it. Or maybe it got so much bad press they'd have to relaunch it before people would take it seriously?

I heard that they've closed their IOT unit.

These chips were too expensive to supplant Raspberry Pis, Arduinos, and ESP8266s.

I wish Intel's IoT story had revolved around the Intel Compute Stick -- targeting people who know how to write Windows native applications and are less familiar with Linux/embedded development. Plus, Intel chips can be used in appliances (e.g. ATMs and kiosks based on Windows or Chrome OS).

The about face is a bit crazy here - I went to their SXSW event a couple of years ago and they were huge on pushing the Edison. So it's not like they weren't invested in these products' success.

Heck, they were the headline sponsor at Maker Faire last month, with a huge presence promoting Edison and Joule. Honestly, though, this was very predictable. There's no money in the "maker" market almost by definition; certainly not Fortune 100 money. What did Intel even think it was doing there in the first place?

I've got 5x edisons that sit at pretty high CPU util and give me no trouble, the Ci20 is the biggest piece of trash I've used; imagination deserve to go out of business (they'll beat intel to it).

Raspberry Pi, can't really knock it. I wish the Edison form factor took off.

I think Imagination made a decent attempt to bring a MIPS based board to market. The I/O specs seemed well thought out for makers, and the datasheets were comprehensive. It lacked any sizeable community, but there were enough skilled people around at the start to make it useful.

It was certainly one of the most enjoyable boards I've used - the hardware was very accessible, the flash was easy to program, and you pretty much owned it from the first block read from the NAND.

It's a bit of a shame the board didn't get much traction.

You must've been lucky, if you check the forums there's a trail of people who had hardware issues, there was a known issue. And they simply stopped responding to anyone. I spent $80 and got thoroughly ripped off; it was a half assed effort and they deserve to go under.

I was as unlucky as you - my board was from the initial release batch that crashed intermittently. Ultimately I took up the refund they were offering to anyone with a faulty board.

It was an unfortunately evasive and insidious bug - the board could run perfectly, maxed out, for hours or even days before freezing up.

From my own experience, I though Imagination handled things about as well as they could, though I understand this may not have been true for everyone.

Reply on news.ycombinator.com