QNX Operating System/Discussion/OSTech/_KER_NOP Enter/_KER_NOP_Exit /List of Posts
 
Forum Topic - _KER_NOP Enter/_KER_NOP_Exit :
   
Update
Dennis Kellly(deleted)
10/08/2012 8:13 AM
post96104
_KER_NOP Enter/_KER_NOP_Exit   
Kernel trace for fftgen (mcdemo) i.mx6 (SP1), 4-cores, 8 compute-bound threads with round-robin schedulling,

Whenever the rr timer expires, the kernel trace shows a _KER_NOP Enter/_KER_NOP_Exit pair.  This seems to correspond to 
the switch of one thread to another thread ON THE SAME CORE.  It shows 1-2us between enter/exit.

Is this time lost?  What is really going on?
Colin Burgess(deleted)
10/09/2012 9:11 AM
post96119
Re: _KER_NOP Enter/_KER_NOP_Exit   
The NOP kercall is used to force a thread into the kernel so scheduling 
can take place.  So that time is spent selecting and switching the 
running thread.

On 12-10-08 8:13 AM, Dennis Kellly wrote:
> Kernel trace for fftgen (mcdemo) i.mx6 (SP1), 4-cores, 8 compute-bound threads with round-robin schedulling,
>
> Whenever the rr timer expires, the kernel trace shows a _KER_NOP Enter/_KER_NOP_Exit pair.  This seems to correspond 
to the switch of one thread to another thread ON THE SAME CORE.  It shows 1-2us between enter/exit.
>
> Is this time lost?  What is really going on?
>
>
>
>
> _______________________________________________
>
> OSTech
> http://community.qnx.com/sf/go/post96104
> To cancel your subscription to this discussion, please e-mail ostech-core_os-unsubscribe@community.qnx.com

-- 
cburgess@qnx.com
Dennis Kellly(deleted)
10/09/2012 9:34 AM
post96121
Re: _KER_NOP Enter/_KER_NOP_Exit   
Thanks.  Seems to roughly correspond to a cross-core interrupt except on same core.


----- Original Message -----
From: Colin Burgess [mailto:community-noreply@qnx.com]
Sent: Tuesday, October 09, 2012 09:07 AM
To: ostech-core_os@community.qnx.com <ostech-core_os@community.qnx.com>
Subject: Re: _KER_NOP Enter/_KER_NOP_Exit

The NOP kercall is used to force a thread into the kernel so scheduling 
can take place.  So that time is spent selecting and switching the 
running thread.

On 12-10-08 8:13 AM, Dennis Kellly wrote:
> Kernel trace for fftgen (mcdemo) i.mx6 (SP1), 4-cores, 8 compute-bound threads with round-robin schedulling,
>
> Whenever the rr timer expires, the kernel trace shows a _KER_NOP Enter/_KER_NOP_Exit pair.  This seems to correspond 
to the switch of one thread to another thread ON THE SAME CORE.  It shows 1-2us between enter/exit.
>
> Is this time lost?  What is really going on?
>
>
>
>
> _______________________________________________
>
> OSTech
> http://community.qnx.com/sf/go/post96104
> To cancel your subscription to this discussion, please e-mail ostech-core_os-unsubscribe@community.qnx.com

-- 
cburgess@qnx.com





_______________________________________________

OSTech
http://community.qnx.com/sf/go/post96119
To cancel your subscription to this discussion, please e-mail ostech-core_os-unsubscribe@community.qnx.com

Return