recp/cglm v0.5.2

v0.5.2: SIMD Update + New Features

on GitHub

New Options:

• CGLM_SSE4_DOT : Enable SSE4 dot product
• CGLM_SSE3_DOT : Enable SSE3 dot product

New Functions:

• glm_persp_move_far() extend frustum's far distance
• void glm_vec4_cubic(float s, vec4 dest) fills vec4 as [s^3, s^2, s, 1.0]
• float glm_mat4_rmc(vec4 r, mat4 m, vec4 c) multiplies row vector, matrix and column vector and returns scalar. This is good helper to get SMC result easily for curves.
• float glm_smc(float s, mat4 m, vec4 c) calculates SMC multiplication by using glm_mat4_rmc()and glm_vec4_cubic()
• float glm_bezier() cubic bezier equation
• float glm_hermite() cubic hermite equation
• float glm_decasteljau() solve cubic bezier equation using decasteljau

New glmm (SIMD) functions:

• glmm_vhadds(v) horizontal add, returns register
• glmm_hadd(v) horizontal add, returns scalar
• glmm_vdots(a, b) dot product, single lane contain dot product to convert result to scalar
• glmm_vdot(a, b) dot product, all lanes contain dot product to use result with other vector operations

Improvements:

• glmm_ functions are moved to platform specific headers.
• Now some glmm_ functions supports NEON
• CGLM_SIMD_x86, CGLM_SIMD_ARM and CGLM_SIMD are defined if it is supported.