| >> 1) Does DEC Fortran permit access to a 64-bit unsigned integer data
>> type? Also (or perhaps this is the same question in disguise), does DEC
>> Fortran have an intrinsic function to access the UMULH instruction
>> (upper half of 64 x 64-bit unsigned integer multiplication)?
DF's INTEGER*8 data type is signed 64-bit integer. DF has no data type that
maps into unsigned integer of any size.
At present there is no Fortran construct that generates a UMULH instruction.
In a future release {By the end of the calendar year}, there will be an
intrinsic called MULT_HIGH that generats a UMULH inline.
/Stan
|
| >> 2) Ever since I upgraded to OSF V3.2, My Fortran programs which read from
>> and write to file don't do this properly unless I include the FULL pathname
>> (they used to write to the directory from which the program was started
>> by default). Also, the Unix "./{file}" local file specifier won't work
>> - I have to open a file, e.g. using
>> open(1,file='/usr/users/joe_blow/myfile.dat',status='unknown')
I have no record of this being reported... kinda hard to believe it doesn't
work. The simplest test works for me: this is x.f
character a*40
open (unit=1, file='./x.f')
10 read (1,1000,end=999) a
1000 format (A40)
write (6,*) a
go to 10
999 end
This is with the V4.1 compiler and DFARTL374 on DUNIX V3.2 214.
We need more info... like what versions of everything and a sample program and
files that fail.
/Stan
|