Function _mm256_insertf128_ps

pub unsafe fn _mm256_insertf128_ps<const IMM1: i32>(
    a: __m256,
    b: __m128,
) -> __m256

Available on x86 only.

Copies a to result, then inserts 128 bits (composed of 4 packed single-precision (32-bit) floating-point elements) from b into result at the location specified by imm8.

Intel’s documentation