On 06/04/2015 07:13 AM, Ron Bean wrote: >> If you need one computer to see the GUI and one for realtime >> effects, why not just start out with a real computer and load Linux and >> LinuxCNC on it?
This "little box" with supporting software on BBB and PC side is very good idea IMO. It is likely the only way for LinuxCNC to prosper in the future. > > The problem I see is that, going forward, "real" computers that are > graphics-capable are becoming less and less "real time capable". We can Strongly agree. Ongoing search for "ideal motherboard" for LinuxCNC is very costly and making it very hard for anybody to provide consistent long term support when motherboard support lifetime is 2 years or less. This is a challenge that every industrial or military equipment manufacturer is facing with fast changing technologies. It is also very silly to have a 4 core CPU and only use one for real time OS while wasting other cores are doing nothing. It would make sense to have 1 CPU core run CNC while other cores would take care of networking, user interface, etc. As far as I know that's not possible now. It made sense for LinuxCNC to use whole PC for CNC functionality at the beginning but the technology is advancing and there might be need to change it's architecture at some point. > work around it for now by selecting motherboards that still work well > with LinuxCNC, but they could become scarce in the future (see various > threads here about selecting motherboards). Single-board computers like > the BBB, which are specifically built for real-time and don't care about > graphics, are the solution to that. > > I don't see any particular reason to run a GUI on the same machine (or Exactly. Not all CNC applications need GUI, 3D printers for example run stand alone for the most part. LinuxCNC with GUI is actually an odd anomaly that runs on a dedicated PC. > at least, on the same CPU) as the motion controller. Although if you do, > one solution might be something along the lines of the BBB, but two CPUs RaspberryPi, BBB and Arduino have a common problem, sandwich design with cables sticking out on all 4 sides. Boards stacked one on top of the other provide very limited amount of options, similar to PC104. While processors are ever more powerful, computer design is running into the same issues mainframe and industrial computers did decades ago. Industrial computers just happen to be larger in those days. Designers figured out that a universal or standard backplane provides most flexibility in terms of electronic and mechanical design. Unfortunately, BBB and Arduino, RaspberryPi designers did not take advantage of that kind of architecture and we ended up with very limited options. Too many times people think of a new idea but forget to visit a computer museum to see how others have solved the same or similar problems before. Good computer architecture includes a backplane, passive or active. Digital computers were among the most popular low cost industrial computers many years ago. Some used 4 slot backplanes, others more with possibility to use expanders for additional interfaces. Some interfaces used only part of the bus to save space. DEC computers had one thing in common, a well documented universal bus for other manufacturers to build different interfaces and controllers. To troubleshoot the interface you simply plugged it in an extender and gained access to both sides of the PCB. Try that on Arduino! In my HW support experience I came across PDP-11 systems running in steel mills, nuclear and hydro power plants, factories, etc. with little or no graphics. Most used VT100, some used more advanced color terminals. Systems with 32kW(ord) or 64kW RAM controlled huge machinery with RTOS on much slower CPU than we have today. > and a PCI slot for a GPU. Another solution might be something like a BBB > that plugs into a PCI slot in a generic PC. Either one eliminates the > USB connection, which is the real problem. Good idea assuming there would be a volume to keep the costs down. IMO it would be better if ARM architecture based universal bus would emerge for use in small embedded systems under $100 so that vendors would be encouraged to build controllers with "Mesa card" like functionality and other interfaces to handle digital and analog IO connections. This thread brought up interesting ideas and comments; good starting point for a "kickstart" project ;-) -- Rafael ------------------------------------------------------------------------------ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
