> Those manual use/optimization of SIMD are done in C anyway.

C seems to be used only for SSE intrinsics support. ~16 LoC of "C".

This is all C in the whole project, if you really think you can call it as such:

https://github.com/google/font-rs/blob/master/src/accumulate...

  void accumulate_sse(const float *in, uint8_t *out, uint32_t n) {
    __m128 offset = _mm_setzero_ps();
    __m128i mask = _mm_set1_epi32(0x0c080400);
    __m128 sign_mask = _mm_set1_ps(-0.f);
    for (int i = 0; i < n; i += 4) {
      __m128 x = _mm_load_ps(&in[i]);
      x = _mm_add_ps(x, _mm_castsi128_ps(_mm_slli_si128(_mm_castps_si128(x), 4)));
      x = _mm_add_ps(x, _mm_shuffle_ps(_mm_setzero_ps(), x, 0x40));
      x = _mm_add_ps(x, offset);
      __m128 y = _mm_andnot_ps(sign_mask, x);  // fabs(x)
      y = _mm_min_ps(y, _mm_set1_ps(1.0f));
      y = _mm_mul_ps(y, _mm_set1_ps(255.0f));
      __m128i z = _mm_cvtps_epi32(y);
      z = _mm_shuffle_epi8(z, mask);
      _mm_store_ss((float *)&out[i], (__m128)z);
      offset = _mm_shuffle_ps(x, x, _MM_SHUFFLE(3, 3, 3, 3));
    }
  }

Also consider top level: https://github.com/google/font-rs/tree/master/src

> What is the main benefit of using Rust?

Safety and performance.

From the article:

"With the SIMD speedup, font-rs is approximately 7.6x faster than FreeType in larger sizes (keep in mind that 42 pixels/em is the default for xxhdpi Android devices)."