christophe-lunarg
commented
8 months ago - Add simd aligned_vec3 (and sse aligned_dvec3 - 2 x xmm)
- Fast packed_vec3 <=> aligned_vec3 and packed_vec4 <=> aligned_vec4 conversion
- Fast aligned_vec3 <=> aligned_vec4 conversion
- Optimized aligned_mat x aligned_mat and aligned_mat x aligned_vec
- Inverse aligned_mat3 simd version (actually slower than ssid on my computer even it has 30% less instruction ?)