Need references on architecural comparison

Thread View: comp.arch

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by Ashish Mehra Tue, 26 Jan 1993 23:46

#12370

Author: Ashish Mehra
Date: Tue, 26 Jan 1993 23:46

28 lines
1510 bytes

I would appreciate if the knowledgable amongst you
can give me some pointers to articles on architectural
comparisons focusing on context switching overhead and
interrupt response time. Relevant processor architectures
are the R2000/3000, SPARC, 68030/40 and 486. Also,
is there any study done that compared the _relative_
performance gains of a synchronization instruction
like test-and-set over pure software mechanisms?

On a related note, the (context switching) numbers 
reported in the literature are directly related to 
the performance of the memory subsystem. Most comparisons 
report the absolute numbers without commenting on the 
speed/bandwidth of the memory subsytem on which the 
measurements were made. Am I missing something here? 
Shouldn't such factors be normalized out? 

I would appreciate if all responses are sent to
ashish@eecs.umich.edu . If there is sufficient 
interest, I will post a summary on this newsgroup.

Thanks,

Ashish Mehra
ashish@eecs.umich.edu

Re: Need references on architecural comparison

#12371

Author: Robert Hyatt
Date: Wed, 27 Jan 1993 01:59

19 lines
1272 bytes

In article <1993Jan26.234628.23217@zip.eecs.umich.edu> ashish@blueridge.eecs.umich.edu (Ashish Mehra) writes:
>
>I would appreciate if the knowledgable amongst you
>can give me some pointers to articles on architectural
>comparisons focusing on context switching overhead and
>interrupt response time. Relevant processor architectures
>are the R2000/3000, SPARC, 68030/40 and 486. Also,
>is there any study done that compared the _relative_
>performance gains of a synchronization instruction
>like test-and-set over pure software mechanisms?
>

What is a "pure software mechanism"???  Assuming you are implementing
a lock for multiprocessors, I haven't seen a pure software approach that
will work.....

-- 
!Robert Hyatt                    Computer and Information Sciences   !
!hyatt@cis.uab.edu               University of Alabama at Birmingham !

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads