This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

sizeof(long double) ...


Compiled the following code on a number of platforms and got varying
output...

<BEGIN CODE>
#include <stdio.h>
#include <stream.h>

int main() {
  cout << "sizeof(long double) : " << sizeof(long double) << endl;

}
<END CODE>

Using g++ version 2.95.2:
On Linux 2.2.15 (i686, no options):           "sizeof(long double) : 12"
On IRIX 6.5.1m (R10000, option: -mabi=n32):   "sizeof(long double) : 8"
On IRIX 6.5.1m (R10000, option: -mabi=n64):   "sizeof(long double) : 8"
On Solaris 2.6 (UltraSPARC-II, no options):   "sizeof(long double) : 16"

Using native compilers:
On IRIX 6.5.1m MIPSpro Compilers: Version 7.2.1
(CC -o32):     "sizeof(long double) : 8"
(CC -n32):     "sizeof(long double) : 16"
(CC -64):      "sizeof(long double) : 16"
On Solaris 2.6 WorkShop Compilers 4.2 30 Oct 1996 C++ 4.2
"sizeof(long double) : 16"

I would have expected more consistent behaviour across the platforms
for gcc. If 16-byte variables give trouble on Intel I can understand
that,
but why not use 16-bytes on MIPS if the native compiler does so, and you
_do_ use them on SPARC ??

Kind regards,

Matthijs van der Molen
Shell E&P Technology Applications and Research
begin:vcard 
n:van der Molen;Matthijs
tel;cell:+31 - 6 - 24 47 38 07
tel;fax:+31 - 70 - 311 2393
tel;home:+31 - 15 - 213 7711
tel;work:+31 - 70 - 311 6114
x-mozilla-html:TRUE
url:http://www.geocities.com/ResearchTriangle/Lab/7301
org:Shell E&P Technology Applications and Research;Shared Earth Modelling (EPT-AEF)
version:2.1
email;internet:m.w.vandermolen@siep.shell.com
title:Research Physicist
adr;quoted-printable:;;Volmerlaan 8=0D=0AP.O. Box 60;Rijswijk;ZH;2280 AB;the Netherlands
x-mozilla-cpt:;0
fn:Matthijs van der Molen
end:vcard

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]