7521
Mach3 under Vista / Re: Minimum PC requirements vs real life (Newbie)
« on: July 09, 2015, 04:10:05 AM »
Hi Bruce,
you are obviously way more knowledgeable than I about dedicating cores and similar and that may prove to be an excellent solution.
My experience is not so much that the CPU gets overloaded but rather it gets a stream of high priority interrupts with certain processes
or apps. What reading I have done suggests interrupts are an important mechanism that Windows uses to communicate between threads.
Further those interrupts are assigned a priority by Windows with little or no control by the user.
Competing interrupts screw with the pulse stream generated by the Mach pulse engine. Admittedly I still use a PP and of course have made myself
vunerable to this problem. From what I have heard in the forum is that an external controller relieves the CPU of demanding interrupt timing and
much improves Machs ability to run smoothly.
Are you of the opinion that a dedicated core is less suspect to competing interrupts than a multiple cores running the same thread?
There are some questions I have about external controllers as well. I understand that Mach becomes a trajectory planner and sends via a buffer
by USB or Ethernet to the controller. What I want to know is wether the contents of the buffer have to be flushed thru to the controller before
the controller can signal back to Mach some external event like a limit switch or probe switch event. If that is the case then the latency of USB
could be significant if you were expecting a critically timed event signal. It may be for that reason that Ethernet controllers seem to enjoy a better
rep than USB amongst our most experienced users.
Craig
you are obviously way more knowledgeable than I about dedicating cores and similar and that may prove to be an excellent solution.
My experience is not so much that the CPU gets overloaded but rather it gets a stream of high priority interrupts with certain processes
or apps. What reading I have done suggests interrupts are an important mechanism that Windows uses to communicate between threads.
Further those interrupts are assigned a priority by Windows with little or no control by the user.
Competing interrupts screw with the pulse stream generated by the Mach pulse engine. Admittedly I still use a PP and of course have made myself
vunerable to this problem. From what I have heard in the forum is that an external controller relieves the CPU of demanding interrupt timing and
much improves Machs ability to run smoothly.
Are you of the opinion that a dedicated core is less suspect to competing interrupts than a multiple cores running the same thread?
There are some questions I have about external controllers as well. I understand that Mach becomes a trajectory planner and sends via a buffer
by USB or Ethernet to the controller. What I want to know is wether the contents of the buffer have to be flushed thru to the controller before
the controller can signal back to Mach some external event like a limit switch or probe switch event. If that is the case then the latency of USB
could be significant if you were expecting a critically timed event signal. It may be for that reason that Ethernet controllers seem to enjoy a better
rep than USB amongst our most experienced users.
Craig