Tai Chi Minh Ralph Eastwood
eddaf464ff
style: fix missing whitespace for alignment
...
Co-authored-by: Michael <michael.guyver@gmail.com >
2021-05-14 15:25:29 +02:00
Tai Chi Minh Ralph Eastwood
7065011bf3
clipspace: add initial implementations LH, RH and NO, ZO
...
Add the initial implementations of the left-handed and right-handed
coordinate systems as well as clipspace depth values of [-1, 0] and
[0, 1].
2021-05-13 23:22:58 +02:00
michaelg
c013bd462c
Add LH & RH_NO perspective functions
...
This commit adds functions `glm_perspective_lh_no` and
`glm_perspective_rh_no` to the code. Unit tests are added and this
commit follows the new pattern of adding the a new file per
coordinate-system and clip-space tuple.
. Makefile.am updated
. removed test/glm_cmp project stub
. unit tests include naive implementations to as well as magic number
ref-data generated by the corresponding GLM functions.
No tests run yet on Windows or Mac.
2021-05-13 23:18:05 +02:00
michaelg
b3a18b8a15
Add glm_perspective_rh_zo function + tests
...
This commit adds the RH/ZO perspective function. It does so in the new
file `cam_rh_zo.h` and further refactors the LH variant into new file
`cam_lh_zo.h`. This creates some churn in the tests and configuration
files as new test files were added as well, and all these changes found
their way into the build files.
Tests passing on Linux.
2021-05-13 23:18:05 +02:00
michaelg
1bce62c371
Add function glm_perspective_lh_zo
...
This commit adds the function `glm_perspective_lh_zo`, modelled on the
implementation of glm_perspective, but amended to provide a left-hand
coordinate system expected by DirectX, Metal and Vulkan (per the GLM
project's `glm/detail/setup.hpp`). It uses a clip-space of zero-to-one.
The function is tested against a longhand version of the algorithm it
seeks to implement as well as against the output of the GLM project's
`glm::perspectiveLH_ZO` function. This commit adds a new subdirectory
`test/glm_cmp` which contains a basic CMake file and `main.cpp`. An
interested user should link or copy or clone the GLM project into this
directory. The `main` function can be used to print the reference data
used so others can verify behaviour in the future, or add new literal
reference values.
2021-05-13 23:18:05 +02:00
Recep Aslantas
a242d83805
style: rename nearVal, farVal to nearZ and farZ
2021-05-08 22:42:44 +03:00
Recep Aslantas
607182982d
Merge branch 'master' into simd-3
2021-05-07 14:08:56 +03:00
Sundaram Ramaswamy
9665be3138
Add struct API wrapper
2021-05-07 13:49:44 +05:30
Recep Aslantas
5c22ca3abb
arrm, neon: use negate instruction instead of xor in glm_inv_tr_neon()
2021-05-07 01:52:12 +03:00
Recep Aslantas
7f9585ca72
arrm, neon: impove hadd performance
2021-05-07 01:46:24 +03:00
Recep Aslantas
d0ab3aaa2e
arm, neon: util macros
2021-05-07 01:46:03 +03:00
Sundaram Ramaswamy
8ce45b4303
Add call and struct interfaces
2021-05-07 00:35:20 +05:30
Sundaram Ramaswamy
f19ff5d064
Use scale for both cases
2021-05-06 23:47:03 +05:30
Sundaram Ramaswamy
d6b93f052e
Add nlerp for quaternions
...
Normalized linear interpolation for quaterions are a cheaper
alternative to slerp. This PR adds nlerp operaiton for quaternions.
2021-05-06 23:38:26 +05:30
Recep Aslantas
28705be5a3
simd, sse: reduce some computation at glm_mul_rot_sse2()
2021-05-01 23:16:03 +03:00
Recep Aslantas
e1b142bce7
add todo to quat.h
2021-05-01 23:03:41 +03:00
Recep Aslantas
0f96eaad20
sse2: optimize glm_mat3_mul_sse2() with sse2
...
* reduce memory access for dest[2][2]
* the speed is increased ;)
2021-05-01 22:55:19 +03:00
winter091
13269f4af8
fix typo: vec3 -> vec4
2021-05-01 16:48:31 +03:00
Recep Aslantas
faf6186c29
sse: optimize glm_mat2_mul_sse2 with sse
2021-05-01 03:44:04 +03:00
Recep Aslantas
2be6ac949b
sse: optimize glm_quat_mul with sse
2021-05-01 03:18:26 +03:00
Recep Aslantas
5b7bc522ac
sse: optimize affine with sse
...
* re-oder instructions for ILP
2021-05-01 02:58:14 +03:00
Recep Aslantas
376cf31ee7
armi neon: optimize affine with neon
2021-05-01 02:46:14 +03:00
Recep Aslantas
d28b381dd6
armi neon: optimize mat4 mul with neon
2021-05-01 02:45:15 +03:00
Recep Aslantas
3673622cc3
simd, sse: optimize mat4 mul-v with sse
...
* re-oder instructions for ILP
2021-05-01 02:17:34 +03:00
Recep Aslantas
a90f706e12
simd, sse: optimize mat4 mul with sse
...
* re-oder instructions for ILP
2021-05-01 02:17:08 +03:00
Recep Aslantas
c065d71a2f
simd, sse: optimize mat4 inv with sse
...
* reduce a few shufflees
* re-oder instructions for ILP
2021-04-30 21:12:17 +03:00
Recep Aslantas
ba634d6c83
simd: optimize glm_mat4_zero() with simd
2021-04-30 04:04:27 +03:00
Recep Aslantas
f35badd436
now working on v0.8.3
2021-04-30 01:00:42 +03:00
Recep Aslantas
04eaf9c535
arm, neon: neon/fma support for glm_quat_mul()
2021-04-29 01:12:00 +03:00
Recep Aslantas
4e4bff418d
arm, neon: neon/fma support for glm_mat2_mul()
2021-04-28 22:06:46 +03:00
Recep Aslantas
55ebbdbe40
arm, neon: neon/fma support for glm_inv_tr()
2021-04-28 14:46:14 +03:00
Recep Aslantas
155eb109a8
arm, neon: neon/fma support for glm_mul_rot()
2021-04-25 03:49:35 +03:00
Recep Aslantas
2903813765
arm, neon: neon/fma support for glm_mul()
2021-04-25 03:45:00 +03:00
Recep Aslantas
0ab50f7208
arm, neon: update mat4_mul to use FMA
2021-04-25 03:41:39 +03:00
Recep Aslantas
701e015bfd
avoid loading vec3 by glmm_load()
...
* use glmm_set1() for each for now
2021-04-25 02:36:06 +03:00
Recep Aslantas
1fb941a41b
drop swizzling helpers fro now for simplicity
2021-04-25 02:35:55 +03:00
Recep Aslantas
92151c6328
arm, neon: use div instead of mul by 1 / det for mat4_inv
2021-04-24 18:02:47 +03:00
Recep Aslantas
afac887850
arm, neon: implement mat4 inv with neon
2021-04-24 17:54:01 +03:00
Recep Aslantas
a111693b6b
arm, neon: implement mat4 determinant with neon
2021-04-24 15:45:36 +03:00
Recep Aslantas
ce9e5f5575
arm: update glmm swizzling func names
2021-04-24 01:38:04 +03:00
Recep Aslantas
d13842e7de
arm: optimize vec4 div with NEON
2021-04-24 00:51:09 +03:00
Recep Aslantas
059e5010e6
arm: define CGLM_ARM64 for identify arm64
2021-04-24 00:44:07 +03:00
Recep Aslantas
65292a94a6
swizzling functions for NEON
2021-04-24 00:00:00 +03:00
Recep Aslantas
f303984aad
use unified glmm api for vec4
2021-04-23 23:34:36 +03:00
Recep Aslantas
7d5c4da7cf
optimize translate functions
2021-04-23 23:32:48 +03:00
Recep Aslantas
63988a515c
glmm: new defines for splat
2021-04-23 23:32:21 +03:00
Recep Aslantas
9725b60d46
rename glmm_shuff1x() to glmm_splat()
...
* mark glmm_shuff1x() as DEPRECATED
2021-04-23 22:12:57 +03:00
Recep Aslantas
f14ca0c3f2
now working on v0.8.2
2021-04-23 17:03:39 +03:00
Recep Aslantas
2de1133012
fix enabling FMA macro for MSVC
2021-04-19 06:04:22 +03:00
Recep Aslantas
ebba4eea8e
win, msvc: enable FMA macro for MSVC
2021-04-19 04:14:14 +03:00
