External/cglm
1
0
Fork 0
mirror of https://github.com/recp/cglm.git synced 2025-10-04 17:09:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

arm, neon: use div instead of mul by 1 / det for mat4_inv

Browse Source
This commit is contained in:
Recep Aslantas
2021-04-24 18:02:47 +03:00
parent afac887850
commit 92151c6328
1 changed files with 9 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
include/cglm/simd/neon/mat4.h
View File

@@ -295,12 +295,21 @@ glm_mat4_inv_neon(mat4 mat, mat4 dest) {
x0 = vcombine_f32(vget_low_f32(vzipq_f32(v0, v1).val[0]), x0 = vcombine_f32(vget_low_f32(vzipq_f32(v0, v1).val[0]),
vget_low_f32(vzipq_f32(v2, v3).val[0])); vget_low_f32(vzipq_f32(v2, v3).val[0]));
/*
x0 = glmm_div(glmm_set1(1.0f), glmm_vhadd(vmulq_f32(x0, r0))); x0 = glmm_div(glmm_set1(1.0f), glmm_vhadd(vmulq_f32(x0, r0)));
glmm_store(dest[0], vmulq_f32(v0, x0)); glmm_store(dest[0], vmulq_f32(v0, x0));
glmm_store(dest[1], vmulq_f32(v1, x0)); glmm_store(dest[1], vmulq_f32(v1, x0));
glmm_store(dest[2], vmulq_f32(v2, x0)); glmm_store(dest[2], vmulq_f32(v2, x0));
glmm_store(dest[3], vmulq_f32(v3, x0)); glmm_store(dest[3], vmulq_f32(v3, x0));
*/
x0 = glmm_vhadd(vmulq_f32(x0, r0));
glmm_store(dest[0], glmm_div(v0, x0));
glmm_store(dest[1], glmm_div(v1, x0));
glmm_store(dest[2], glmm_div(v2, x0));
glmm_store(dest[3], glmm_div(v3, x0));
} }
#endif #endif