Currently I’m using the X-Controller and grblweb running on a Raspberry Pi, and I’m looking to upgrade to the Gecko g540 controller running LinuxCNC.
I have an old motherboard / single core CPU (with a parallel port) that I’ve installed LinuxCNC on to test the “jitter” numbers . With a pretty heavy load (6 instances of glxgears, copying large files over the network, etc. I am maxing out at 30,000. Is this good enough?
Also, if I upgrade from the parallel port to a Mesa “anything I/O” card such as the 5i25, do I need to be concerned about the jitter numbers? What advantage does the Mesa card have over a standard parallel port?
Would love to hear from anyone who has experience with this, thanks …
Yes, the idea of squeezing some more life out of 15-year-old hardware certainly is appealing. Just trying to understand how to get the most “bang for the buck” (and also, what advantage does the Mesa card provide over a plain 'ole parallel port?)
- Are the 30,000 latency numbers I see now good enough?
- Will adding Mesa card ($89) with the current hardware will give me better latency numbers?,
- Or, do I need to go to a newer motherboard / CPU with a parallel port?
From what I’ve been reading about LinuxCNC, the real-time kernel can’t really take advantage of multiple cores - most of the tweaks I’ve seen regarding reducing latency involve pinning interrupts to a single core. So, maybe the newest, bleeding edge hardware isn’t the best solution for this application.
Going to continue to scour the LinuxCNC forums (and this one) and will eventually do some experimentation on my own; just wondering if anyone here had any experience / insight …
