[SystemSafety] multi-core validation for aerospace

[Steve Tockey]

Part of the issue here is that there is almost always a lot of other stuff going on in a typical operating system:
Network packets coming in
Clock interrupts
System even logging service writing stuff out to disk event logs
And so on

Some of the variation in timing could depend on the “randomness” of this kind of background computation happening in the OS.

If you are talking about an embedded OS like VxWorks, LynxOS, or similar then the level of background computational noise should be less than with Windows, MacOS, etc but there is almost certainly still a level of it going on.

What sort of percentage variation were you seeing?


— steve



From: systemsafety <systemsafety-bounces at lists.techfak.uni-bielefeld.de<mailto:systemsafety-bounces at lists.techfak.uni-bielefeld.de>> on behalf of Olwen Morgan <olwen at phaedsys.com<mailto:olwen at phaedsys.com>>
Date: Monday, November 19, 2018 at 7:05 AM
To: "systemsafety at lists.techfak.uni-bielefeld.de<mailto:systemsafety at lists.techfak.uni-bielefeld.de>" <systemsafety at lists.techfak.uni-bielefeld.de<mailto:systemsafety at lists.techfak.uni-bielefeld.de>>
Subject: Re: [SystemSafety] multi-core validation for aerospace



Interesting.

A few weeks back I tested a few C functions that performed, without using conditional constructs, some small tasks that usually are accomplished with conditionals. By way of comparative performance tests, I wrapped the functions in a loop that performed them 5,000,000,000 times and timed them using the facilities of the <time.h> library.

There were interesting variations in successive timings for the same function, which I attributed, possibly wrongly, to having aynschronous processes swapping among the cores of an Intel iCore7 processor. This made me wonder about the extent to which asynchronous processing in multi-core CPUs may introduce unhelpful temporal non-determinism into otherwise deterministic processes.

Apart from wondering, I lack the specialist expertise to prognosticate further.


Olwen


On 13/11/2018 16:10, Chris Hills wrote:
Hi All

I have a customer who is looking at a project that will be multi-core processors for aviation use.  They are going to have to validate the system and its software. They are trying to work out how they would gather sufficient evidence for this.

They are looking at different cores on the same device.  Different cores could potentially run different functions each, or could use something across all the cores. Design is up in the air just now, so they are looking for a general certification approach and what they  would need to achieve this and  any ways they  could do it using more automated methods where possible.

Does anyone have any pointers for them to get started?
This is a UK based project.

Regards
  Chris
Phaedrus Systems Ltd
96 Brambling B77 5PG
FREEphone 0808 1800 358    International +44 1827 259 546
Vat GB860621831  Co Reg #04120771
Http://www.phaedsys.com<http://www.phaedsys.com/>  chills at phaedsys.com<mailto:chills at phaedsys.com>




_______________________________________________
The System Safety Mailing List
systemsafety at TechFak.Uni-Bielefeld.DE<mailto:systemsafety at TechFak.Uni-Bielefeld.DE>
Manage your subscription: https://lists.techfak.uni-bielefeld.de/mailman/listinfo/systemsafety
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.techfak.uni-bielefeld.de/mailman/private/systemsafety/attachments/20181119/4158a68b/attachment.html>