c++ - Is there an equivalent of _mm_slli_si128(__m128i a, int num) for floats? -

let's have vector of 4 floats:

__m128 vector = |f0|f1|f2|f3| (pseudocode) 

my intention transform variable this:

|0.0|f0|f1|f2| 

doing shift right appear simplest choice, haven't been able find such intrinsic available floats.

what fastest way accomplish this?

here's solution:

__m128 const mask = _mm_castsi128_ps(_mm_set_epi32(0, -1, -1, -1));  vector = _mm_shuffle_ps(vector, vector, _mm_shuffle(0,3,2,1)) vector = _mm_and_ps(vector, mask);