Did you add an extra > character for your response? It is messing with my email client, which interprets these as "reply" nesting...
> sorry , i will add an extra > character for my response on later reply. >i have know what you said ,thank you very much bin.w...@qkmtech.com From: Gedare Bloom Date: 2018-01-11 23:09 To: bin.w...@qkmtech.com CC: Chris Johns; Users Subject: Re: Re: smp support On Wed, Jan 10, 2018 at 8:48 PM, bin.w...@qkmtech.com <bin.w...@qkmtech.com> wrote: > >> another question: >> >> if i use the smp configuration macro define，am i still need to do the >> mutiprocessor configuration macro define for confdef.h ? >> > With SMP you have to enable SMP when you configure/build RTEMS, and > you need to provide the CPP configuration macros to set up the SMP > system and schedulers. > >>> what the CPP means? C Preprocessor. Did you add an extra > character for your response? It is messing with my email client, which interprets these as "reply" nesting... >>>in CONFIGURING A SYSTEM chapter in c user doc, i find the two >>> configurations items . one is 24.22 SMP Specific Configuration Parameters >>> ,another is 24.24 Multiprocessing Configuration >>>what i want to ask is if i want to use smp, do i need use the >>> Multiprocessing Configuration? or i just need use SMP Specific >>> Configuration Parameters and do not need care about >>the Multiprocessing >>> Configuration. You do not need to care about Multiprocessing Configuration. > > "SMP means all cores are treated equally by the operating system." > "AMP means cores are not treated the same by the OS, which can cover a lot > of heterogeneous behavior." >>>i know what you said >>>so the basic problem which confused me is the difference betweent >>> Multiprocessing Configuration and SMP Configuration, not the concept of amp >>> and smp >>>in my point of view ,no matter AMP or SMP is all called Multiprocessing >>> because they all use the muticore .is it right ? so if i want to use AMP >>> mode ,than i need >>just Multiprocessing Configuration? There is not support for AMP out of the box with RTEMS as far as I know. This "Multiprocessing Configuration" stuff is about how to run a distributed system of multiple RTEMS images, one on each processor, that share a global object namespace. It is not the same as how to run a single RTEMS image that manages multiple processors. >>>if i want to use SMP mode,than i need Multiprocessing Configuration and >>> SMP Configuration ? > > > > > > ________________________________ > bin.w...@qkmtech.com > > > From: Gedare Bloom > Date: 2018-01-11 00:43 > To: bin.w...@qkmtech.com > CC: Chris Johns; Users > Subject: Re: Re: smp support > On Fri, Dec 29, 2017 at 3:55 AM, bin.w...@qkmtech.com > <bin.w...@qkmtech.com> wrote: >> thanks >> >> but i still confused about the smp configuration and the mutiprocessor >> configuration . >> >> for example. for zynq ,they has two arm core. if i use the mutiprocessor >> configuration ， Is rtems use the two cores in amp mode ? is the rtems >> manage the two cores？ if rtems manages two cores，i think it is in smp mode >> because i think the concept of smp mode is the one system can manage >> muticores which are same cpu architecture。this is conflictive for my >> point >> of view >> > SMP means all cores are treated equally by the operating system. > > AMP means cores are not treated the same by the OS, which can cover a > lot of heterogeneous behavior. > > RTEMS provides SMP capability. > >> i think amp mode is rtems run in one core, another core is bare-metal or >> linux. but the mutiprocessor configuration let me confused, it said as >> follows >> >> "The RTEMS multiprocessor execution model is multiple instruction streams >> with multiple data >> streams (MIMD). This execution model has each of the processors executing >> code independent >> of the other processors. Because of this parallelism, the application >> designer can more easily >> guarantee deterministic behavior." >> > This statement is true regardless of SMP/AMP: in either case, program > code runs independent of other processors. > >> is it means one rtems system instance run in one core ,and another rtems >> system instance run in another core? and they can excute code >> independent? >> sorry, i think maybe i have some concept confused >> > One RTEMS instances executes across both cores, and both cores can > execute code independent. It is possible for both cores to be > executing OS functions simultaneously, or one core to execute OS code > while another runs application, or both cores to execute application > code in parallel. > >> if just one rtems run in two core , how to put code into different core to >> run ? how to realize the parrallelism? >> > If you select RTEMS with SMP and provide scheduler with two cores, the > scheduler will assign tasks/threads to the cores according to its > scheduler policy/algorithm. > >> another question: >> >> if i use the smp configuration macro define，am i still need to do the >> mutiprocessor configuration macro define for confdef.h ? >> > With SMP you have to enable SMP when you configure/build RTEMS, and > you need to provide the CPP configuration macros to set up the SMP > system and schedulers. > >> ________________________________ >> bin.w...@qkmtech.com >> >> >> From: Chris Johns >> Date: 2017-12-29 14:21 >> To: bin.w...@qkmtech.com; Users >> Subject: Re: smp support >> On 29/12/17 12:16 pm, bin.w...@qkmtech.com wrote: >>> thanks a lot . >>> >>> i have another question , if i use the smp, can i use the interrupt of >>> rtems? >>> which i want to interrupt period of hard real time of 125us. >>> >> >> Yes you can. >> >>> i do not know the difference of the bare-metal interrupt and the rtems >>> interrupt >>> . is the rtems interrupt have performance loss of the hard real-time? >> >> The interrupt overhead will have the same overhead each time and its >> deterministic behaviour will depend on the other interrupts in your system >> and >> their priority. The performance will depend on what you need to do in the >> interrupt. >> >>> >>> what the schedule method of smp use? is it can affect the interrupt >>> performance? >>> >> >> That depends on your application. >> >> Chris >> >> >> _______________________________________________ >> users mailing list >> users@rtems.org >> http://lists.rtems.org/mailman/listinfo/users >
_______________________________________________ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
