[comp.sys.sequent] Difference in access time between local and shared vars?

amehta@euclid.rdrc.rpi.edu (10/24/89)

Hi,

	Can someone tell me if there is a difference in performance
between the following types of data (I am programming in C, on
a Balance 21000):

   Local data, allocated:
	a) at compile time.
	b) by function invokation
	   (ie, a variable declared inside a function).
	c) by a call to "malloc".

   Shared data, allocated:
	a) at compile time.
	b) by a call to "shmalloc".

   Shared data, that is, 
	a) read/updated by one processor.
	b) read by many; updated by one.
	c) read by many; updated by many.


I understand that each processor has 8K of local RAM, 8K of cache
RAM, and the rest is Shared Memory, which is accessed through a
common bus.

Does all local data get stored in a process's Local RAM?
Does the time for accessing/updating data depend on any of the 
above factors, and if so, by how much?

Inquiring minds want to know...

Thanks in advance for your time and help!!

Alok Mehta
mehtaa@cs.rpi.edu