vtcf@NCSC.ARPA.UUCP (02/10/87)
Is there a way to create a RAM disk on a VAX running VMS? The Feb 9 issue of Digital Review, Page 8, has an article mentioning a "pseudo" backup disk in RAM. Any help would be greatly appreciated. Thanks Tom Williams
DHASKIN@CLARKU.BITNET.UUCP (02/11/87)
Tom Williams <VTCF@NCSC.ARPA> asks: > Is there a way to create a RAM disk on a VAX running VMS? The Feb 9 issue of > Digital Review, Page 8, has an article mentioning a "pseudo" backup disk > in RAM. Any help would be greatly appreciated. Interestingly, a colleague at Digital Review with whom I used to work forwarded two routines to establish a RAM disk as mentioned. They include a command procedure and an accompanying Macro-32 program. Please note that I have *not* tested it (in fact, I haven't even had time to look at it) but I would be willing to forward to anyone interested. Let me know... Denis W. Haskin ------------------------------------------------------------------------ DHASKIN@CLARKU.BITNET Office of Information Systems (617)793-7193 Clark University 950 Main Street Worcester MA 01610 "Anyone who _moves_ before Most Holy comes back out will spend the rest of eternity sipping lava through an iron straw." - Cerebus
carl@CITHEX.CALTECH.EDU.UUCP (02/12/87)
In order to connect the pseudo-disk to your system:
--------------------------------------------------------------------------------
$ run sys$system:sysgen
CONNECT device-name/NOADAPTER/DRIVER=PDDRIVER
--------------------------------------------------------------------------------
wheer you use something like PDA0 for device-name.
The following source (in PASCAL) formats a pseudo-disk.
--------------------------------------------------------------------------------
[inherit ('SYS$LIBRARY:STARLET')]
program Format_pd (input, output);
type
unsigned_word = [word] 0..65535;
status_block = [volatile] record
status, count : unsigned_word;
dev : unsigned;
end;
var
chan : unsigned_word;
iosb : status_block;
device : varying [80] of char;
block_size : integer;
stat : integer;
procedure LIB$SIGNAL (%IMMED stat : integer); extern;
begin
write ('Device name: ');
readln (device);
write ('Disk blocks: ');
readln (block_size);
stat := $ASSIGN (DEVNAM := device, CHAN := chan);
if not odd (stat) then LIB$SIGNAL (stat);
stat := $QIOW (CHAN := chan,
FUNC := IO$_FORMAT,
IOSB := iosb,
P1 := %IMMED (block_size));
if not odd (stat) then LIB$SIGNAL (stat);
if not odd (iosb.status) then LIB$SIGNAL (iosb.status);
end.
--------------------------------------------------------------------------------
As does the following C program:
--------------------------------------------------------------------------------
#include <descrip.h>
#include <iodef.h>
main(nargs, args)
int nargs;
char **args;
{ static char buffer[64];
$DESCRIPTOR(dev, buffer);
long chan, maxblock, iosb[2], stat;
if (nargs < 2) {
printf("Device to format: ");
gets(buffer);
} else
strcpy(buffer, *++args);
dev.dsc$w_length = strlen(buffer);
if (((stat = SYS$ASSIGN(&dev, &chan, 0, 0)) & 7) != 1)
exit(stat);
if (nargs < 3) {
printf("Number of blocks: ");
scanf("%x", &maxblock);
} else
maxblock = atol(*++args);
if (((stat = SYS$QIOW(0, chan, IO$_FORMAT, iosb, 0, 0, maxblock,
0, 0, 0, 0, 0)) & 7) == 1)
stat = *iosb;
exit(stat);
}
--------------------------------------------------------------------------------bob@seismo.CSS.GOV@cec2.UUCP (02/12/87)
In article <8702101849.AA00835@ncsc.ARPA> vtcf@NCSC.ARPA.UUCP writes: >Is there a way to create a RAM disk on a VAX running VMS? The Feb 9 issue of >Digital Review, Page 8, has an article mentioning a "pseudo" backup disk >in RAM. Any help would be greatly appreciated. > >Thanks > >Tom Williams I think such a program would be simple to write, although, under VMS, your RAM disk would constantly be paged back and forth to the physical system disk. A somewhat less than optimal result. :-) Try starting with the VMS Guide to Device Drivers. Bob
vtcf@NCSC.ARPA.UUCP (02/14/87)
To answer the question of why I wanted a RAM disk, it was actually for another collegue who has a Microvax, with a FORTRAN program that was reading some huge file over and over again. As it turns out, the RAM disk works fine, but it didn't really speed up his program any, due to the manner in which he was doing the I/O. The system manager of the Micovax was trying to find a way to speed up someone else's program, so I was getting third-hand knowledge. I think I'll go have a look at the person's code, because I know he isn't very Knowledgeable about FORTRAN. Tom Williams
sasaki@HARVARD.HARVARD.EDU.UUCP (02/16/87)
One warning, the PDDRIVER is fairly slow. The transfer from the "disk" to the user's buffer is done a byte at a time, rather than a movc3. Remember too that this slow transfer is done a at elevated IPL. We were thinking of setting aside a few megabytes on our 8650's to put commonly used images and small databases. Rather than improve performance, overall performace actually went down. This might be due to the particular work load. ---------------- Marty Sasaki uucp: harvard!sasaki Ziff Davis Technical Information Co. arpa: sasaki@harvard.harvard.edu 80 Blanchard Road bitnet: sasaki@harvunxh Burlington, MA 01803 phone: 617-273-5500