On Jul 24, 2007, at 22:09, Andreas L Delmelle wrote:
For those interested, I (think I) managed to verify the below
statement, while toying around with improperly synchronized code.
Although nothing was explicitly multi-threaded, I did use Object.wait
() and Object.notify(), and those map to native methods that crashed
with errors like "current thread not owner", because the Java code
was not in synch.
Apparently, the JVM was taking advantage of native OS multithreading
for its implementation of monitors, and those monitors did not like
what I was doing...
Cheers
Andreas
<snip />
Just thought some more about this, and it occurred to me that, in
principle, provided that:
a) your machine has multiple CPUs
b) an OS with a robust threading model and
c) a JVM that knows how to use that model to its advantage
then it could even be that a /single/ run of FOP is processed by
more than one processor... I'm not 100% certain, but I'd say that
sets of bytecode instructions are not necessarily performed
sequentially. Like a single CPU ultimately can create the illusion
of parallelism, multi-CPU systems can generate the illusion of
sequential processing, by synchronizing behind the scenes on very
low level.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
