Richard is correct here. The listings in Top are inter-process threads, not processes. The memory trick is a way to tell; or the knowledge that Java is threaded tells you all you need to know.
On the question: I wonder if there is something going on with Volano's app. For instance, there are a number of changes in 1.4 that are not in 1.3 and require new code to take advantage of them. For instance, we have an app that we run that cannot run in 1.3 because we use the 1.4 SDK. It may be that there app is optimized for 1.3; there are some changes to the networking IO. But this is a stab in the dark. I do not know exactly why that is the case. Ben Ricker Wellinx.com On Fri, 2003-06-06 at 11:01, Richard Kilgore wrote: > On Fri, Jun 06, 2003 at 05:32:19PM +0200, Gwendolyn van der Linden wrote: > > > Has anyone ever heard why the JVMs out there always lose to their > > > Windows counterparts in performance figures? The latest study at > > > this site shows all the Linux JVMs failing the number of > > > concurrent connections test miserably, except for the Blackdown > > > 1.3.1 JVM which otherwise sucks in terms of performance. > > > > I haven't heard, but I guess that Windows JVMs take advantage of > > Windows native threads and the efficient use of CriticalSection (which > > is quite fast), while most Linux JVMs use multiple processes. Just do > > a 'ps aux' after you started some big app like Borland Enterpise > > Server... Linux is supposed to have much better threading in 2.6, and > > hopefully app's will start taking advantage of that. > > > > Gwendolyn. > > Actually, those are native threads, unless you have an old JVM. > Linux lists all threads in the output of ps aux. The only way I > know to tell when they are actually threads in the same process > is that a lot of the stats reported (memory size, resident > memory, etc.) are identical. I haven't looked carefully at the > ps man page in a while, though. > > - richard -- [EMAIL PROTECTED] mailing list
