what is the counterpart of "dble" to convert to real*16?

[bsmile_2nd]

need to know this quickly, thanks!

[TimP]

How about real(x, selected_real_kind(20)) ?

[bsmile_2nd]

not homework, need to modify a lot of lapack subroutines to make them work under real*16, thus it would be most convenient to find out a replacement for dble, otherwise the modification would be terrible ...

[Steven_L_Intel1]

real(x,16)

There is not a specific name similar to dble for this.

[JVanB]

QEXT

[TimP]

QEXT does go way back, to DEC platforms even prior to VMS, and to HP3000. Unlike dble() and real(x, KIND=...), it was neverdefined ina standard, so is not included in many compilers with equivalent functionality.

[Steven_L_Intel1]

Ah, yes, missed QEXT. Sorry about that. As Tim notes, it is non-standard.

[bsmile_2nd]

Thank allso so so much! It works! But I cannot find it in intel fortran manual, maybe due to the fact that it is nonstandard, however, intel supports it, that's enough!

[Steven_L_Intel1]

It's in the Language Reference, A-Z section.