sumanth <sumanth.gundapneni@xxxxxxxxxxxxxxxxxx> writes: > Size of double in architecture is 32bit wide which means float > and double are same. > The problem here is how can I tell a softfloat library in libgcc > to compile libraries for double of 32-bit wide. You shouldn't have to. gcc invokes its internal functions based only on the size of the floating point value, not the type. If you build the SFmode libraries, they will be used for both float and double. Ian
