From: Ray H. <re...@up...> - 2003-06-15 18:08:21
|
Hi Dave. My understanding of all of these issues is at best theoretical and murky. My thinking comes from reading some of the rt and rtai posts by Yodaikin and the other real-time folk. When I wrote that, I was thinking of some of the hard wired work arounds that have been done for buggy silicon. At worst these kinds of problems add jitter to our steper pulse train. The newest intel stuff can be difficult and I have not tried to compile for any of these. The 8?? chipsets have been causing problems with BDI because they were developed after most of the device drivers and x window versions available in those distributions. For my money I'd choose something like the $219.00, 1.1 GHz Lindows pc that TigerDirect is offering right now. (no connection with them) Yes it is old technology and the first thing I'd do is sacrefice the newer Lindows distro for BDI. It sounds like one giant leap backwards but at least I'd have a known, and very fast, system for making chips. I'm sure that the Linux community is working on threading issues. I've heard good things about SuSE and some of the pioneering hardware work they are about. I find it fascinating that their Linux release for the 64 bit AMD chip was the only OS available when the chip was released. I also know that Linus and the core developers have been working on several schemes and that some of this work is available in the 2.5++ versions of the kernel. I have no urge to get that far out on the bleeding edge. I am certain also that the EMC community will be caused to make changes to the way we do real time by these new technologies. We have had a recent, significant discussion with another software development group about user space vs kernel space real-time. While they have much to offer, our feeling was that we could not sacrefice hard real time at least as we do it now. None of this should be taken as saying that Ray Henry is into old and not enouraging new. I am excited to hear about people working to apply the latest and greatest stuff. If I can help, I will in any way that I can. EMC has a real advantage in that it can run on the cast off hardware from those other OS's but we should not be content to call that advantage enough. Hope this helps. Ray On Sunday 15 June 2003 12:30 pm, you wrote: > HI Ray; > > Important questions regarding the paragraph below. > > Are there specific issues with hardware based scheduling? I'm > assuming this is a reference to Intels newest processors and the Multi > Threading features they have. If so should one stay away from such > hardware? Since I'm shopping around for emc hardware, understanding > this issue in relation to the latest Intel hardware is important. > > Also while I do not like RedHats approach at all to the new threading > model, I do hope that everyone realizes that the Linux community will > have to change threading models at some time in the future. Of course > we could stick with older Kernels for a very long time if emc required. > > Thanks > dave > > > Ray Henry wrote: > > From: Ray Henry <re...@up...> > > >The RedHat based threading competes with RT and RTAI for the attention of > > the processor as do some of the hardware based scheduling that we are > > seeing from chip and board makers. Our dependence upon hard-real-time > > makes a good working system a bit more difficult the farther into these > > things those we move. > > > > > >Snipped<<<<<<<<<<<<<<<<<<<<<<, > > ------------------------------------------------------- > This SF.NET email is sponsored by: eBay > Great deals on office technology -- on eBay now! Click here: > http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5 > _______________________________________________ > Emc-users mailing list > Emc...@li... > https://lists.sourceforge.net/lists/listinfo/emc-users |