128 bit floating point arithmetic

[Jiri Hladky]

Hello,

I would like to use 128-bits floating point arithmetic on my
Pentium 4 Linux machine. Does gcc support it? I have tried to 
use the flags 

 -march=pentium4 -mcpu=pentium4 -mfpmath=sse 
 -m128bit-long-double 

but the precision of floating point numbers of type "long 
double" is just 80 bits. Could anybody kindly point me to C 
source code together with gcc compile flags to do basic 
computation (addition and multiplication)  with 128 bits 
precision floating point numbers? 

Also, are there any plans to implement 128 bits transcendent 
functions like sin and exp? 

I would highly appreciate any help on this.

Thank you very much!
Jirka