Re: JavaMemoryModel: More synchronization statistics

From: Sean McDirmid (mcdirmid@cs.utah.edu)
Date: Thu Nov 11 1999 - 00:07:12 EST


My opinion is that:

(a) For many programs, programmers depend on the system to worry about
synchronization, e.g. they expect or choose to use the synchronized APIs,
it is too much for the programmers

(b) For some programs, programmers aren't writing multithreading code and
just incidentally use synchronization b/c the APIs are synchronized (pre
1.2 legacy collection classes).

When/if people start using synchronization in some way that isn't blind,
lock performance will become important.

On Wed, 10 Nov 1999, Bill Pugh wrote:

> The results from the Volano benchmark.
>
> Should we take the following lesson from all of these statistics?
>
> * For most programs, including many multithreaded benchmarks,
> 99% of the synchronization cost involves useless locks
> (e.g., locks that are not being used to synchronize threads).
>
> Thus, the big issue is not how to use faster techniques when
> we need to communication between threads, but to avoid
> paying the cost for synchronization when we aren't.



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:22 EDT