On Wed, 14 Nov 2012, Kent A. Reed wrote: > Date: Wed, 14 Nov 2012 18:03:11 -0500 > From: Kent A. Reed <[email protected]> > Reply-To: EMC developers <[email protected]> > To: EMC developers <[email protected]> > Subject: Re: [Emc-developers] Preview: LinuxCNC 2.5 single codebase - supports > RTAI, RT_PREEMPT, Xenomai-kernel, Xenomai-User, Simulator > > On 11/14/2012 5:18 PM, John Morris wrote: >> Just tested on the i686 in the shop, where the numbers looked pretty >> bad. The base numbers looked like 13us/10us, but spikes quickly brought >> those up to 100-200us. There's still tuning to be done, such as turning >> the SMI workaround kernel options on. >> >> As I keep saying, my goal is to lower the barrier to entry for novice >> users. I predict that a big challenge in making LinuxCNC more >> accessible will be inability to provide a single kernel image that works >> for everyone. The SMI workarounds are an example: these should not be >> turned on for all machines, but on ones with jitter spikes but no >> SMI-related risk of overheating, these should be turned on. Maybe this >> will change in the future if a run-time switch is implemented, but for >> now it's a problem. >> >> > > For grins, I jwent to test the xenomai-user configuration on an old P4 > machine, from which I've obtained sub-20us max jitter values with past > releases of LinuxCNC. Wouldn't you know it, the xenomai kernel borks my > Xserver on this machine. I hate it when that happens. I'm sure I'll get > it sorted but try explaining it to a Newbie. > > I'm hoping ARM-based boards and Michael's work with xenomai-user threads > will give us a more stable platform. > > I'm fearing, of course, that this may be a victory we have to fight for > over and over again, what with the constant flux in ARM offerings, > notably in the way IP cores are mixed and matched in physical devices. I > almost responded to Eric Johnson's message about the new, low-cost, > linux-capable ARM board that these things remind me of the old saying > about buses. If you don't like one, just wait. Another will come along > in a few minutes. The kicker is, if you do like one, you better hope it > doesn't pull away soon. Just look at the rapid evolution in the Intel Atom. > > Regards, > Kent >
I think that is an issue with ARM CPUs because unlike a PC, theres no 'platform' standard at all other than maybe a cell phone/tablet with a 6 month platform lifetime. A good thing about the Atom is that Intel seems to have learned their lesson with the PowerVR video fiasco so the next Atoms will have Intel video again. Peter Wallace Mesa Electronics ------------------------------------------------------------------------------ Monitor your physical, virtual and cloud infrastructure from a single web console. Get in-depth insight into apps, servers, databases, vmware, SAP, cloud infrastructure, etc. Download 30-day Free Trial. Pricing starts from $795 for 25 servers or applications! http://p.sf.net/sfu/zoho_dev2dev_nov _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
