[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference hydra::axp-developer

Title:Alpha Developer Support
Notice:[email protected], 800-332-4786
Moderator:HYDRA::SYSTEM
Created:Mon Jun 06 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:3722
Total number of notes:11359

3685.0. "Aspen Technologies" by HYDRA::AXPDEVELOPER (Alpha Developer support) Fri May 30 1997 13:33

    Company Name :  Aspen Technologies
    Contact Name :  Peter Piela
    Phone        :  (617) 577-0100 x293
    Fax          :  
    Email        :  
    Date/Time in :  30-MAY-1997 12:30:07
    Entered by   :  Jeff Kenyon
    SPE center   :  MRO

    Category     :  vms
    OS Version   :  6.2
    System H/W   :  Server


    Brief Description of Problem:
    -----------------------------

	Customer is having problems with mallocing greater than 1.0GB of
	memory.  What does he need to change to be able to do this?

	email is [email protected]
T.RTitleUserPersonal
Name		DateLines
3685.1can't do that on VMS V6.2 (Chuck N consult).HYDRA::KENYONThe Foundation of Science...FictionFri May 30 1997 15:0528
Peter,

The unfortunate scoop is that until OpenVMS V7.0, you are stuck with using
the P0 and P1 spaces for heap and stack.  Unless you can do something like
allocate what you are after on the stack (it would be volitale to that
scope,
and yuor >1GB would not be contguous), then 1GB is it.

As of OpenVMS V7.0, there is an additional space, P2, which you can get at.
There is Postscript documentation on the ASAP Web site about these
capabilities -- "OpenVMS Alpha Guide to 64-bit Addressing".  You can find
it
at:

http://www.partner.digital.com/www-swdev/pages/Home/TECH/documents/vms-docs-v7.

If you need me to send this to you in emial, I can.  The relavent issues
for
generally alloacting memory on OpenVMS are WSMAX in the sysgen facility,
and
wsextent in the user authorization file (as well as having a reasonable
pagefile.sys to back it).

Regards,

Jeff Kenyon
Alpha Developer Support
3685.2response...HYDRA::KENYONThe Foundation of Science...FictionFri May 30 1997 18:2231
From:	SMTP%"[email protected]" 30-MAY-1997 16:27:07.82
To:	[email protected] ([email protected])
CC:	
Subj:	Re: OpenVMS and getting more than 1GB...

Thanks for your prompt reply, although I had wished for better news. I
hope you don't mind indulging me a bit further on finding a workable
solution.

>To: [email protected]
>Subject: Re: OpenVMS and getting more than 1GB...
>From: [email protected] ([email protected])
>Date: Fri, 30 May 1997 13:44:46 -0400
>
>Peter,
>
>The unfortunate scoop is that until OpenVMS V7.0, you are stuck with using
>the P0 and P1 spaces for heap and stack.  Unless you can do something like
>allocate what you are after on the stack (it would be volitale to that scope,

I could probably get memory in P1 by defining a large array in an
outer scope in my C/Fortran program, but this is not very nice. Can I
allocate memory in P1 using the lib$vm calls? If so, am I correct in
assuming this memory remains allocated until it is freed? Does the
memory manager know that the max stack size has been reduced?

>and your >1GB would not be contguous)

I can live without the >1Gb not being contiguous.

Thanks. Peter
3685.3reply from AspenTec...HYDRA::KENYONThe Foundation of Science...FictionFri Jun 06 1997 10:0926
From:	SMTP%"[email protected]"  2-JUN-1997 17:11:51.43
To:	[email protected] ([email protected])
CC:	
Subj:	Re: Allocating memory from the stack -- re: problem 1997-3685

>To: [email protected]
>Subject: Allocating memory from the stack -- re: problem 1997-3685
>From: [email protected] ([email protected])
>Date: Mon, 2 Jun 1997 12:41:58 -0400
>
>Peter --
>
>Here's one thing you might be able to do on pre-V7 systems:
>
>Start at hex 40000000.  This is the bottom of p1 space (where the 
>stack = lives)
>ProbeW this address to see if it can be written.

I don't know what probeW is, please explain.

>Move down by pages, probing once per page, until you get the amount of =
memory
>	you need.=20
>
Thanks for such a quick response. Peter