Intel kind of just dropped this one on us, it’s called Edison - it’s a SD card sized SoC (System on a Chip), 22nm, 2x 400 MHz cores, integrated WiFi & Bluetooth, based on Intel’s Quark architecture.
I’m very excited. It’s basically the desktop computer I had as a twelve year old, but it’s an SD card. That’s freaking cool.
Can you imagine the processing power that sensors could now have for your robots. You could even build these into motors for better control. Interesting…
Assuming the interface is 4X the size of the card (I could not find what it truly is, so it is just a guess), if you had a cubic foot of these you would have a 7030 core unit that would clock at 2812 ghz.
I agree, the intercommunication would make for incredible power, even though it would not be fully utilized because the communication methods (wifi or messes of wires) and speeds could never be as good as in integrated 7030 core processor (not reasonably possible by any standards other than supercomputing)
I could see a touch screen monitor come out that has a SD card slot (or two) for these Edison SoC’s. with other added posibilities for connections (like usb, vga/hdmi, and audio) I could also see in the future a concept like a Chromebook that, instead of an open up screen like a laptop, has a flat screen like a tablet, but could roll up and fit in a pocket for ultra-ultra portability.
But in reality I think Intel sees this as a computers everywhere opportunity. Connectivity and Internet of Things kind of enabler.
I’m sick of the Cortex’s limitations and cannot bring myself to really work with it given the other projects I have going on, I thought the Pi could bridge the gap but even that became a chore.
I’m excited to see if this cannot finally bring x86 to a robot in a reasonable way.
As for the interface, I believe a PCI-E x1 connector is exposed somehow which you then connect to whatever you want. A breakout board of sorts would accompany this well.
James, as usual that’s a really good question - one that’s hard to answer quickly and completely, so I’ll do my best.
I’m not happy with the Cortex for a couple key reasons, no one of these would be a problem but together, it’s not good so let’s run through them as constructively as we can…
It’s $249.99
I (being an electronics novice) can build my own Cortex with the following:
So how’d I do? Humm well it’ll cost $134, so even if I left a few things out I have another $116 chunk change holding down my wallet.
I have 12 motor ports (plus 7 from the Arduino for a total of 19), not 10.
I have 20 digital I/O (not counting the ones on the Pi), not 12.
I have 12 analog ports, not 8.
I have 512MB of RAM, not 64KB.
I have 8GB flash, not 384KB.
I have a video card.
I have a 700 MHz ARM11 benched at 847 MIPS, not 90 MIPS.
So yeah, if I can put together at consumer pricing a “Cortex” that’s over $100 cheaper, has a CPU 10 times more powerful, has over 8000 times the memory and over 21 times the ROM and in every way matches (and beats) the current Cortex in I/O, I have reason to be unhappy with it.
Then Edison comes around, so do the same thing but imagine the Pi is the SD card and the Arduino is basically the breakout board. It’s freaking amazing, this is the current day. VEX is like 3 years behind the curve at least.
2. I need OOP, and I don’t care how it’s done
Yes I know PROS and ConVEX can technically do it with C++ but even you guys don’t support it, and those projects are not supported by VEX - at all. So it’s a crapshoot.
I now confidently know C, C++, Java, JS and PHP. By confidently I mean I’ve done significant, multi-month projects in all of the above and feel like I “get” them enough to write good code.
I’m telling you I need method overloading, classes, objects and at least a reasonable amount of memory for heap stuff.
That and I really want to see VEX support both C++ and Java. While Java isn’t as efficient (arguably), I can teach it a LOT faster than I can teach C++.
I had a very fancy sensor and motor library on paper that pretty much needed classes to work so we couldn’t do it, yet I could have banged it out in Eclipse in like 2 hours on my desktop. It’s a shame.
3. It’s physically large
And yes I know my fake Cortex would be larger but IFI has the power to do better, much better. I would like a much flatter Cortex (1/2" tall, with ports on the sides, not coming out of the top!). I considered rendering a concept but I doubted it would matter, they tend to do what they want at the end of the day.
**4. EasyC / RobotC
Just no, no more charging for IDE’s. Embrace the community and give us enough documentation to let projects like ConVEX and PROS do what they need to do. Let RobotC and EasyC “adjust” to the new open (ish) way of things by lowering their price and they will still do fine, more novice users will still pay -something- for an easier IDE while the non-novice will happily use their IDE of choice, like we already do for pretty much all other kinds of programming.
The Cortex is a dev kit, sell me a dev kit or a programming IDE - NOT BOTH.
Yeah it does
Pi works great, we just lack the money/time/equipment to really do what we wanted. We’re broke college kids who dream big. I refuse to stop dreaming big, so I guess we’re going to have to make some monies, which is kind of what I’m trying to work on now (instead of robots and Pi’s and stuff).
Anyway I’m not feeling well at the moment, so I’ll stop here. -Cody**
Yeah Wallaby has it, if this was a personal project I would have such an option. But to take it even farther, who said this was about me? I use robotics to teach people engineering and CS. I want a better teaching tool and I want it at a reasonable cost too.
I think that was an issue of supply and demand, though. It was originally much more expensive (can’t remember the price) but nobody was buying it. It’s discontinued now and they’re probably just trying to get rid of it all.
The reason no one bought it was that it’s not competition legal. Let’s be honest here, the competition is the main draw to most of us on the forums, and not the classroom stuff. Were VEX to declare at Worlds “You can use the ARM9 VEX Pro Microcontroller next year,” their remaining supply would sell out in an instant. We’d have ordered 4 of them from the stands before Karthik even sat down.
I’m with Cody on wanting a better system to use. We’re currently being stopped in a project we were working on because of the Cortex’s tiny flash memory. 384 KB is nothing. If the next iteration had an SD card as its memory (making downloading code faster, too) that alone would be a huge step up.
The classroom stuff, though, doesn’t need anything more. Have you seen the curriculum? It’s incredibly basic. The Cortex is more than sufficient for their needs.
Do I think we need an Intel Edison for competition, though? No. That would be hugely overkill. Bluetooth paired motors and sensors eliminating the need for wires would be nice, but (from what I understand) impractical due to the number of teams that compete. Built in wifi would be helpful for debugging and downloading, but the ARM9 even has that. Jumping from the Cortex to an Edison would be like trading a tricycle in for a Lamborghini. Most people would have no idea what to do with the extra processing power, anyway. I know we would never get anywhere near using a full 2 GhZ processor for VEX. Maybe VEXU needs it. But I know we won’t.
…I’m sure many many more, all capable of being a lost cost “teaching tool”
Don’t blame C for not being used to a certain kind of programming. If you really wanted to, a quick google search found this:
[How do you implement a class in C? - Stack Overflow
(Also, never forget KISS ;))
Anyways, I always felt the cortex was designed to be a simple robotics platform, not much more. If your teaching robotics to students, I don’t see any reason you need anything more (and I think C is much easier to teach then Java as you don’t have to worry about oop concepts, but maybe that’s just me?).
The Edison is indeed very, very cool (I’m sure pricey though!), however does it have a position in robotics? I personally don’t see it in vex, or at least to me it seems largely unnecessary.](How do you implement a class in C? - Stack Overflow)
You misunderstand. There is no way cheaper and 20 times as powerful is a bad thing. It also helps students learning to code actually have a controller that can handle their inefficiency. They then can see how better coding speeds up the processor rather than just not be able to do something.
If it was more expensive for a better processor I would suggest Vex allow 2 processors at once like cortex vs pic. BUT its cheaper.
FYI, I, one of 46 other people in the world ever, have personally upvoted that question in the past (on July 8th, 2013). You can [verify for yourself, the record is towards the bottom.
I honestly lack the time or energy to fully defend my position. It would end up in a slam post anyway.
but apart from ssh and maybe vnc I don’t know how a monitor would work with it
