[Bug 1647] Modular SuperCore Scheduler

[bugzilla-daemon at rtems.org]

https://www.rtems.org/bugzilla/show_bug.cgi?id=1647

--- Comment #60 from Gedare <giddyup44 at yahoo.com> 2010-11-15 17:28:36 CST ---
(In reply to comment #59)
> (In reply to comment #58)
> > Created an attachment (id=1076)
 --> (https://www.rtems.org/bugzilla/attachment.cgi?id=1076) [details]
[details]
> > New files.
> > 
> > Fix to inline more of the priority scheduling functions.  Addresses the concern
> > with overhead of Scheduler_priority_Block, which calls
> > Scheduler_priority_Schedule -- now it calls Scheduler_priority_Schedule_body,
> > thus inlining the scheduling check.
> 
> That appears to have knocked 14 instructions off, leaving 21.  :-D
> 
> Do you see anything else possible?  Or are we at a wall?

It seems that 14-15 instructions is the amount of 1 function call on the psim.
The overhead of Thread_Set_state is now 21 insns, so 1 function call and a
couple extra load/store insns I would guess.  

15-20 instruction overhead on each Scheduler call is as good as we can get with
the current design -- and I think we could only get better by having
conditional compilation #ifdef blocks, so I think we found the wall.

-- 
Configure bugmail: https://www.rtems.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.