FC3 sucks. It takes up too much memory!

[Gain Paolo Mureddu]

Jeff Vian wrote:

>On Mon, 2005-01-31 at 08:29 -0700, Robin Laing wrote:
>  
>
>>Gain Paolo Mureddu wrote:
>>    
>>
>>>James Mckenzie wrote:
>>>      
>>>
>>>>James McKenzie
>>>>A Proud User of Linux!
>>>>        
>>>>
>>>What will it take for Sun to really look into their Linux (and Solaris 
>>>for that matter) JavaVM. Currently the best performing JavaVM is the one 
>>>for Microsoft systems and it is not even written by Sun!! (the Microsft 
>>>JavaVM)
>>>      
>>>
>
>What do you men "best performing"?.  speed?, quality?, features?, or
>exactly what are you defining this as?
>
>I don't use anything but the Sun VM so I can't compare. However, I am
>very satisfied with what I get.
>  
>
First of all, yes, not only speed wise, actually the speed difference 
between the two is not THAT big, however MS VM uses for some reason hogs 
much less the system, even if the program would run at about the same 
speed in both VMs, the rest of the system would not slow down to a 
crawl, and the memory usage... I mean, have you seen how much memory 
will launching a Java application consume?? have you seen the amount of 
shared memory all the Java processes have?? Those are HUGE amounts of 
memory... I mean probably 80-100Mb are ok, but 400Mb?? That is way too 
much in my book... And yes, my computer is Linux-only, but eventually I 
have to use a WinXP machine, in which embedded applets while about as 
fast as in Linux, the rest of the computer feels about the same as 
always, plus the memory foot print is MUCH less (when loaded, the plguin 
would increase at most 20Mb memory usage). Both systems with very 
similar specs (actually my PC is slightly faster [AMD XP 2100 Vs 2200])

>  
>
>>But it isn't to the Java Standard.  MS lost the court battle over this 
>>one.
>>    
>>
>
>IIRC, not only did M$ lose the battle about code rights, they have
>deviated greatly from the java standards.  Thus any code written to use
>their vm and it's features is likely to fail to function correctly on a
>vm that complies with the standards.
>
>The best bet is to use the JRE package from Sun and be certain that all
>the code it runs is compliant with standards.  Any java programs that do
>not work with the Sun vm are likely not standards compliant.
>
>
>  
>
They may have lost the battle in court, and I'm happy for that. Really, 
is just like I said, Sun should pay more attention to their VM 
performance in Linux.