On 12/5/16, Mike Leimon leimon@gmail.com wrote:
Since as I understand it, the people at t-firefly have been pretty helpful towards getting the RK3288 EOMA board put together, I figured I would point out that they have just launched a crowdfunding campaign to build a RK3399 development board.
yeah i saw that
I'm not sure if that rk3399 processor would be able to work for a future EOMA68 design however, if it did, it would bring some cool advantages. It looks like it supports USB 3.0, it is a 64-bit processor, and like the RK3288, it can support 4 GB of memory.
yes. here's the key advantage of the RK3288 over the RK3399 right now: it's been around longer. i'm just writing up an update at the moment, but basically the fact that the RK3288 has been around for nearly two years means that there's such a stupid amount of "metoo HOWTOs" out there it's actually decreasing the informational signal-to-noise ratio on setting up an RK3288-based system.
it's got mainline u-boot support, it's got mainline linux kernel support, it's got nearly a dozen different products out there, it's got PCB CAD files that have "accidentally" made their way onto the internet and can be picked up for a whopping $24 on taobao.com, it's got the Technical Reference Manual available online... all without an NDA.
now, whilst it would be *really nice* to do an RK3399 board, although USB3 would be a nice-to-have, the performance/watt figures for the arm64 cores are a whopping FIFTEEN PERCENT lower than for the 32-bit ones right now.
so my feeling is, it would be a good idea to wait a year, see how things progress with the RK3399, see if it becomes as popular, and in the meantime keep a lookout for both entirely libre processors as well as some that can handle 8GB of RAM.
l.