| Author |
Message |
stevech
Joined: 23 Feb 2006
|
|
Posted: 02 November 2006, 5:14 AM Post subject: Can system.alloc be slow? |
|
|
can the execution time of system.alloc() be quite slow, like 100's of mSec?
can it cause a rescheduling before it returns, due to memory management work such as garbage collection/defrag? |
|
| Back to top |
|
 |
dkinzer
Site Admin
Joined: 03 Sep 2005
Location: Portland, OR
|
|
Posted: 02 November 2006, 6:10 AM Post subject: Re: Can system.alloc be slow? |
|
|
| stevech wrote: | | can the execution time of system.alloc() be quite slow, like 100's of mSec? |
That's a long time. I'd be suprised if an allocation request would take that long. A highly fragmented heap will cause longer average allocation times, of course. I'd be interested in hearing more about what you're seeing and the circumstances that lead up to it.
| stevech wrote: | | can it cause a rescheduling before it returns, due to memory management work such as garbage collection/defrag? |
No. Allocation requests do nothing other than their specific function. |
|
| Back to top |
|
|
stevech
Joined: 23 Feb 2006
|
|
Posted: 02 November 2006, 6:26 AM Post subject: |
|
|
I'm just fishing for why my state machine program doesn't work well enough now after I changed from static memory stacks in the tasks to stacks obtained via system.alloc(). I can't blame that change. But that's about all that I changed. The problem is speed related. By design, state machine changes states quite slowly, on the scale of a second or several seconds.
I need to do more analysis/measurements. |
|
| Back to top |
|
|
|
