Apple LLVM 3.0 compiler automatically floating to double

Question

There is a piece of code:

*det_ptr++ = (float)(dx*dy - 0.81*dxy*dxy);

where dx, dy and dxy are float.

The Apple LLVM 3.0 compiler makes the following assembly for it:

 +0x250 vldr.32 s0, [r13, #+140] +0x254 vldr.32 s1, [r13, #+136] +0x258 vmul.f32 s0, s0, s1 +0x25c vcvt.f64.f32 d16, s0 <-------------- cast from float to double +0x260 vldr.32 s0, [r13, #+132] +0x264 vcvt.f64.f32 d17, s0 <-------------- cast from float to double +0x268 vldr.64 d18, [r13, #+16] +0x26c vmul.f64 d17, d18, d17 +0x270 vldr.32 s0, [r13, #+132] +0x274 vcvt.f64.f32 d19, s0 <-------------- cast from float to double +0x278 vmul.f64 d17, d17, d19 +0x27c vsub.f64 d16, d16, d17 +0x280 vcvt.f32.f64 s0, d16 +0x284 ldr r0, [sp, #+104] +0x286 adds r1, r0, #4 ; 0x4 +0x288 str r1, [sp, #+104] +0x28a vstr.32 s0, [r0] 

Is there a way to ban these ghosts?

Answer 1

The way you wrote your program requires these casts. 0.81 is a double-precision literal, so dxy must be doubled before multiplication occurs, and dx*dy must be raised before subtraction. The fact that you pass the final result back to the float does not matter - the C standard is very clear that these terms are evaluated with double precision independently.

To prevent the commercial from doubling, use a literal with only precision instead (adding the suffix f ):

 *det_ptr++ = dx*dy - 0.81f*dxy*dxy;