Re: [Emc-users] Pc choice ... help
thanks ... this is the same that i know but my last study about motherboard was wiyh Z80 so my knowelge about is not updated. regards Il giorno dom 31 mar 2019 alle ore 18:43 theman whosoldtheworld < bleachk...@gmail.com> ha scritto: > > > Il giorno sab 30 mar 2019 alle ore 04:26 Przemek Klosowski < > przemek.klosow...@gmail.com> ha scritto: > >> On Thu, Mar 28, 2019 at 2:13 PM theman whosoldtheworld < >> bleachk...@gmail.com> >> wrote: >> >> > >> > PCW in the forum normally suggest desktop motherboard ... Is it just a >> > matter of Platform Controller Hub that prevented the rt-preemt kernels >> > from having good performances? >> > >> I always thought the realtime latencies are mostly caused by the black >> magic of the SMB---the system management firmware which is allowed to take >> over completely from the main CPU. Somehow this is connected to the video >> hardware: even though video normally does not have its own non-preemptible >> firmware. The way I understand it, the SMB code ends up calling the video >> callbacks which are poorly designed and take excessively long. >> >> Since SMB is a black box subsystem, there's no way to predict its >> latencies. My understanding is that there is only one sure way to find >> out: run latency tests, like LinuxCNC is doing. >> >> ___ >> Emc-users mailing list >> Emc-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/emc-users >> > ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Pc choice ... help
Il giorno sab 30 mar 2019 alle ore 04:26 Przemek Klosowski < przemek.klosow...@gmail.com> ha scritto: > On Thu, Mar 28, 2019 at 2:13 PM theman whosoldtheworld < > bleachk...@gmail.com> > wrote: > > > > > PCW in the forum normally suggest desktop motherboard ... Is it just a > > matter of Platform Controller Hub that prevented the rt-preemt kernels > > from having good performances? > > > I always thought the realtime latencies are mostly caused by the black > magic of the SMB---the system management firmware which is allowed to take > over completely from the main CPU. Somehow this is connected to the video > hardware: even though video normally does not have its own non-preemptible > firmware. The way I understand it, the SMB code ends up calling the video > callbacks which are poorly designed and take excessively long. > > Since SMB is a black box subsystem, there's no way to predict its > latencies. My understanding is that there is only one sure way to find > out: run latency tests, like LinuxCNC is doing. > > ___ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Pc choice ... help
On Thu, Mar 28, 2019 at 2:13 PM theman whosoldtheworld wrote: > > PCW in the forum normally suggest desktop motherboard ... Is it just a > matter of Platform Controller Hub that prevented the rt-preemt kernels > from having good performances? > I always thought the realtime latencies are mostly caused by the black magic of the SMB---the system management firmware which is allowed to take over completely from the main CPU. Somehow this is connected to the video hardware: even though video normally does not have its own non-preemptible firmware. The way I understand it, the SMB code ends up calling the video callbacks which are poorly designed and take excessively long. Since SMB is a black box subsystem, there's no way to predict its latencies. My understanding is that there is only one sure way to find out: run latency tests, like LinuxCNC is doing. ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Pc choice ... help
a salesman insists with to sell me an intel I3 SOC laptop motherboard and claims that since he has no PCH control he will still have excellent performances regarding servo period etc etc ... Since I never understood what actually in modern motherboards worsened the servant period ... now it turns out that it is the PCH is this just what makes the difference between a laptop motherboard and a desktop? I believe for example that an I3 6200 processor in any case better than an I3 6200U removing the PCH from the motherboard can compensate for these differences? PCW in the forum normally suggest desktop motherboard ... Is it just a matter of Platform Controller Hub that prevented the rt-preemt kernels from having good performances? System On Chip (with ultralow power intel series) is better than traditional motherboard for LinuxCnc with intel normal I3 6200? Thank in advances for any help about these questions. bkt ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
