Hi Paul, I meant inlining as in what the compiler / linker does instead of calling a function explicitly. I was speculating that some low-level math in motion / TP might be sped up if link-time-optimization can do more inlining of small functions.
For example, here's an article on how it can be used in the Linux Kernel to improve performance and reduce executable size: https://lwn.net/Articles/744507/. Also this: https://www.researchgate.net/publication/47374866_Optimizing_real_world_applications_with_GCC_Link_Time_Optimization Your issue is most likely caused by something in how the kinematics are configured (or maybe a bug in the kinematics themselves). Can you post a config? One of us may be able to try it out and see if anything looks funny. Best, Rob On Thu, Aug 15, 2019 at 4:09 PM Paul Kessler < paul.kess...@cadminservices.com> wrote: > Robert, > > What do you mean by "in-lining of things like posemath". I have been > having some trouble with the math in LinuxCNC. I took the work that was > done for maxkins (RTCP) and changed it for my machine as one kinematic > model I can use. I also have a second one that uses trivkins kinematic. > The python model of my machine works correctly when I simulate the trivkins > kinematic machine but the maxkins(RTCP > ) kinematic model does not work. It is just in the calculation of the new > position of the X axis erroring out with any B axis angle change. Funny > the Z axis works fine because I can comment out the X calculation portion > in the equation and the spindle keeps Z steady with all the B axis angle > changes taken out in X. (Tool tip moves in X axis with Z height remaining > the same for any B axis angle change.) > > Is this one of the problems? > > Paul Kessler > > -----Original Message----- > From: Robert Ellenberg <rwe...@gmail.com> > Sent: Thursday, August 15, 2019 11:39 AM > To: EMC developers <emc-developers@lists.sourceforge.net> > Subject: [Emc-developers] Link-time optimization in realtime objects > > Hi All, > > Has anyone tried to use link-time optimization in the realtime object > builds? I think the motion / TP build would be a good candidate for it, so > we could improve inlining of things like posemath. > > Best, > Rob > > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > > > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
