Bug 15154 – Wrong overload resolution with implicit conversion of other vectorized parameter __vector(void[16])

Overload resolution seems to see float and double as matching types when another function parameter is implicitly converted (float4 to void16). float dot_simd1(float4 a, float4 b) { float4 result = __simd(XMM.DPPS, a, b, 0xFF); float value; __simd_sto(XMM.STOSS, value, result); return value; } The following error is produced by the code above: simd1.d(34): Error: core.simd.__simd_sto called with argument types (XMM, float, __vector(float[4])) matches both: /usr/include/dlang/dmd/core/simd.d(434): core.simd.__simd_sto(XMM opcode, double op1, __vector(void[16]) op2) and: /usr/include/dlang/dmd/core/simd.d(435): core.simd.__simd_sto(XMM opcode, float op1, __vector(void[16]) op2) When casting to the correct type (void16) the code compiles without errors: float dot_simd1(float4 a, float4 b) { float4 result = __simd(XMM.DPPS, a, b, 0xFF); float value; __simd_sto(XMM.STOSS, value, cast(void16)result); return value; }

Not much of a bug here, in the first snippet you're calling __simd_sto with [XMM, float, float4] which is a MATCHconvert for both the overloads of the __simd_sto. On the other hand when you explicitly cast the vector to void16 we get a MATCHconvert for the [XMM, double, void16] overload (note the float -> double conversion) and a MATCHperfect for [XMM, float, void16] so that's what is picked.