Re: % Unable to allocate memory: to make array. [message #8500 is a reply to message #1014] |
Fri, 21 March 1997 00:00   |
steinhh
Messages: 260 Registered: June 1994
|
Senior Member |
|
|
In article <5grkoh$etm@post.gsfc.nasa.gov>, thompson@orpheus.nascom.nasa.gov (William Thompson) writes:
|> davidf@dfanning.com (David Fanning) writes:
|>
|> >buzzcut <buzzcut@highland.com> wrote:
|>
|> >> Are there anyways around this error besides .size 65000 65000?
|> >> Any useful tips?
|> >>
|> >> % Unable to allocate memory: to make array.
|> >> not enough core
|>
|> >Allocate more virtual memory to your IDL process.
|>
|> I don't believe that the .size command is relevant to the "not enough core"
|> error message. As I understand it, this cannot be solved at the user level,
|> but must be addressed at the system manager level. Either one needs to create
|> more swap space, or increase the user's quota if the machine uses process
|> quotas. I know that the latter is true on VMS systems, but I don't think it's
|> generally true under Unix. I don't know about Windows or MacOS.
|>
Under e.g., Digital UNIX, this problem may also be artificially imposed on
a process by software limits on memory use, maintaned by the user's shell - e.g.,
unix> limit
cputime unlimited
filesize unlimited
datasize 131072 kbytes
stacksize 2048 kbytes
coredumpsize 0 kbytes
memoryuse 90200 kbytes
vmemoryuse 1048576 kbytes
descriptors 4096
The limits may be changed by e.g.,
unix> limit datasize 1024m
which gives the child processes an allowed data size of 1 Gbyte.
I'd also like to point out the use of the coredumpsize, which disallows
any core dumps when set to zero - how often do you actually *use* those
humongous files, anyway....
Stein Vidar Haugan
|
|
|