Compiler/TMS320C6678: Where can I find C/C++ SIMD intrinsic instructions (SSE)?

Tool/software: TI C/C++ Compiler

Hi All,

I'd like to improve some vector-matrix operations on a TMDSEVM6678LE  via intrinsic instructions (SSE). I have done it via Intel-best instructions for GCC. as you can see below; however, our main purpose is to apply SSE for a TMDSEVM6678LE.

Please let me know if you are aware of a document/tutorial.

I wasn't able to find any. I only could find the two below: 

http://www.ti.com/lit/an/spraa14/spraa14.pdf

http://www.ti.com/lit/an/spra666/spra666.pdf?ts=1591768687912

Regards

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>

#include <pmmintrin.h>

#ifndef __SSE3__
#error This example requires SSE3
#endif

#include "util.h"

/* Size of the matrices to multiply */
#define SIZE 256
#define ITERS 100

#define XMM_ALIGNMENT_BYTES 16

static float mat_a[SIZE][SIZE] __attribute__((aligned (XMM_ALIGNMENT_BYTES)));
static float mat_b[SIZE][SIZE] __attribute__((aligned (XMM_ALIGNMENT_BYTES)));
static float mat_c[SIZE][SIZE] __attribute__((aligned (XMM_ALIGNMENT_BYTES)));
static float mat_ref[SIZE][SIZE] __attribute__((aligned (XMM_ALIGNMENT_BYTES)));


/**
 * Matrix multiplication. This is the procedure you should try to
 * optimize.
 */
static void
matmul_sse()
{
    for (i = 0; i < SIZE; i++) {
        for (j = 0; j < SIZE; j++) {
            float c_sum_float[4] = {0, 0, 0, 0};
            __m128 c_sum = _mm_set_ps1(0.f);
            for (k = 0; k < SIZE; k+=4) {
                // load
                const __m128 a = _mm_load_ps(&(mat_a[i*SIZE+k]));
                const __m128 b = _mm_load_ps(&(mat_b[k*SIZE+j]));
                // mul
                const __m128 c = _mm_mul_ps(a, b);
                // sum & add
                c_sum = a + b;
            }
            _mm_storeu_ps(&c_sum_float[0], c_sum);
            mat_c[i][j] = c_sum_float[0] + c_sum_float[1] + c_sum_float[2] + c_sum_float[3];
        }
    }

}

Ple