The two numbers being reported are Virtual Memory (the amount of address
space allocated to the process) and Resident Set Size (the amount of the
process address space that is occupying physical RAM). When the data
space's VM is bigger than its RSS, the difference is paged out to the
swap devices, and paged into physical memory as needed.

Essentially, RSS reflects how much of the program's address space has
been touched recently, as well as the kernel's decision about how to
allocate physical memory among processes that need it. If a program
needs to regularly touch more memory than it can get RSS from the kernel,
it tends to hammer the swap devices.

Nathan

On Wed, Nov 24, 1999 at 03:09:24AM -0800, Bradley Rosser wrote:
> Hi.  I'm running the Blackdown 1.2preV2 java on Linux 2.2.12.
> 
> I've found that I have to use the -mx option to run with 128Mb or
> so in my application.  I'm always curious as to how much memory
> it's "really" using, and I've found that 'top' will report this
> pretty well - at least I think so.  Memory increases slowly; I
> get the application to run gc() and it drops, etc.  Top's report
> seems faithful to what I know the application is doing.
> 
> However a 'ps -l' listing always shows the same amount of memory
> in use, typically the 128Mb plus a bit more (for code, bss and
> stuff, I guess).  So I'm assuming that java grabs all 128Mb worth
> at the start, and ps shows that; so what does 'top' report in
> what seems to be a more realistic portrayl of what Java is
> actually 'using'?
> 
> Just curious,
> 
> 
> 
> 
> Brad Rosser
> [EMAIL PROTECTED]
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Thousands of Stores.  Millions of Products.  All in one place.
> Yahoo! Shopping: http://shopping.yahoo.com
> 
> 
> ----------------------------------------------------------------------
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]


----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to