Re: Dumb question on embedded Linux - why?

From: Grant Edwards (grante_at_visi.com)
Date: 10/15/05 

Date: Sat, 15 Oct 2005 20:25:40 -0000

On 2005-10-15, Drily Lit Raga <midicad2001@yahoo.com> wrote:

>> >> 2) It's not real-time?
>>
>> Linux doesn't guarantee latencies between events and the
>> handling of those events. The latencies caused by the kernel
>> have been greatly improved recently, but it's still not
>> suitable for hard, low-latency, real-time projects.
>
> But it still gets used in "real-time" applications,

It's good enough for some applications, not for others. A
buddy of mine recently worked on a project where they had to
ditch Linux because the latencies were just too large. IIRC
they switched to LynxOS.Since it has a Linux-compatible ABI,
they could just re-compile the applications and go.

> being "good enough" because a modern-day CPU has plenty of raw
> speed.

That depends on the application and on the CPU.

> Of course if you want that on a PIC I don't know....

No, I wasn't expecting to run Linux on a PIC. 

-- 
Grant Edwards                   grante             Yow!  ... A housewife
                                  at               is wearing a polypyrene
                               visi.com            jumpsuit!!

Relevant Pages

  • Re: RT patch acceptance
    ... cowardly luring those embedded developers into using the stock Linux ... we indirectly improve latencies of the desktop preemption model too. ... embedded world, the more of them use Linux, the better for all Linux ... needs Linux to scale to 500 CPUs (in fact only a small precentage needs ...
    (Linux-Kernel)
  • Re: realtime preemption performance difference
    ... I am not using montavista kernel, I am using standard linux kernel. ... Realtime is known by worst case latencies. ...
    (Linux-Kernel)
  • Re: [announce] [patch] Voluntary Kernel Preemption Patch
    ... While locking a RT process to a CPU to achieve even lower latencies ... the general userbase wants good latencies on simple UP, ... Of course Linux!= RTLinux and we never expect nor pretend Linux to ... (see the old issue with kernel 2.4, ...
    (Linux-Kernel)
  • Re: [PATCH 0/5] LSM hook updates
    ... These are just lmbench and kernel build numbers (certainly not the best ... vert.sous Linux 2.6.13- x86_64-linux-gnu-oldhoo 2997 ... Processes - times in microseconds - smaller is better ... File & VM system latencies in microseconds - smaller is better ...
    (Linux-Kernel)
  • Re: [PATCH] [request for inclusion] Realtime LSM
    ... applications that have the right to get RT scheduling, ... in a way that they cannot on Linux. ... OS X says "we know that running realtime applications matters ... with the kernel is, how LSM makes a solution possible, acknowledged ...
    (Linux-Kernel)