feat(host): vendor PyroWave + minimal Granite subset as crates/pyrowave-sys

Phase 0 of design/pyrowave-codec-plan.md — the opt-in wired-LAN ultra-low-
latency codec. Vendored at upstream 509e4f88 (API 0.4.0, Granite 44362775,
volk + vulkan-headers pins in PUNKTFUNK-VENDOR.txt), pruned to the 6.6 MB
the standalone no-renderer build needs; scripts/vendor-pyrowave.sh
reproduces the tree (a pin bump is protocol-affecting, plan §4.2).

build.rs drives the wrapper CMakeLists (static archives incl. a static
C-API lib upstream only ships shared) + bindgen over pyrowave.h; Linux and
Windows only, empty stub elsewhere (Apple gets a native Metal port, §4.7).
Offline-safe by construction: no network, no system lib, vendored Vulkan
headers — same model as the opus dep (flatpak builder has no network).

Phase-0 validation on .21 (RTX 5070 Ti, driver 610.43.03):
- upstream pyrowave-c-test + interop test (incl. dmabuf/DRM-modifier
  Vulkan<->Vulkan) pass, from the pristine AND the pruned tree
- GPU kernel times at ~1.6 bpp noise: encode/decode 0.090/0.042 ms @800p,
  0.146/0.067 @1080p, 0.226/0.103 @1440p, 0.477/0.201 @4K — order of
  magnitude under NVENC's 1-2 ms retrieve, CBR lands within ~100 B of
  target
- cargo test -p pyrowave-sys green (static link + API-version pin check)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
2026-07-15 00:35:10 +02:00
parent 1b73361372
commit 4c3b11445c
396 changed files with 140058 additions and 0 deletions
@@ -0,0 +1,460 @@
/* Copyright (c) 2017-2026 Hans-Kristian Arntzen
*
* Permission is hereby granted, free of charge, to any person obtaining
* a copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sublicense, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice shall be
* included in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
#include "matrix_helper.hpp"
#include "muglm_impl.hpp"
#include "simd_headers.hpp"
namespace muglm
{
mat3 mat3_cast(const quat &q_)
{
auto &q = q_.as_vec4();
mat3 res(1.0f);
float qxx = q.x * q.x;
float qyy = q.y * q.y;
float qzz = q.z * q.z;
float qxz = q.x * q.z;
float qxy = q.x * q.y;
float qyz = q.y * q.z;
float qwx = q.w * q.x;
float qwy = q.w * q.y;
float qwz = q.w * q.z;
res[0][0] = 1.0f - 2.0f * (qyy + qzz);
res[0][1] = 2.0f * (qxy + qwz);
res[0][2] = 2.0f * (qxz - qwy);
res[1][0] = 2.0f * (qxy - qwz);
res[1][1] = 1.0f - 2.0f * (qxx + qzz);
res[1][2] = 2.0f * (qyz + qwx);
res[2][0] = 2.0f * (qxz + qwy);
res[2][1] = 2.0f * (qyz - qwx);
res[2][2] = 1.0f - 2.0f * (qxx + qyy);
return res;
}
mat4 mat4_cast(const quat &q)
{
return mat4(mat3_cast(q));
}
mat_affine mat_affine_cast(const quat &q)
{
return mat_affine(mat3_cast(q));
}
mat4 translate(const vec3 &v)
{
return mat4(
vec4(1.0f, 0.0f, 0.0f, 0.0f),
vec4(0.0f, 1.0f, 0.0f, 0.0f),
vec4(0.0f, 0.0f, 1.0f, 0.0f),
vec4(v, 1.0f));
}
mat4 scale(const vec3 &v)
{
return mat4(
vec4(v.x, 0.0f, 0.0f, 0.0f),
vec4(0.0f, v.y, 0.0f, 0.0f),
vec4(0.0f, 0.0f, v.z, 0.0f),
vec4(0.0f, 0.0f, 0.0f, 1.0f));
}
mat_affine translate_affine(const vec3 &v)
{
return mat_affine(
vec4(1.0f, 0.0f, 0.0f, v.x),
vec4(0.0f, 1.0f, 0.0f, v.y),
vec4(0.0f, 0.0f, 1.0f, v.z));
}
mat_affine scale_affine(const vec3 &v)
{
return mat_affine(
vec4(v.x, 0.0f, 0.0f, 0.0f),
vec4(0.0f, v.y, 0.0f, 0.0f),
vec4(0.0f, 0.0f, v.z, 0.0f));
}
float determinant(const mat2 &m)
{
return m[0][0] * m[1][1] - m[1][0] * m[0][1];
}
mat2 inverse(const mat2 &m)
{
float OneOverDeterminant = 1.0f / determinant(m);
mat2 Inverse(
vec2(m[1][1] * OneOverDeterminant,
-m[0][1] * OneOverDeterminant),
vec2(-m[1][0] * OneOverDeterminant,
m[0][0] * OneOverDeterminant));
return Inverse;
}
float determinant(const mat3 &m)
{
return m[0][0] * (m[1][1] * m[2][2] - m[2][1] * m[1][2])
- m[1][0] * (m[0][1] * m[2][2] - m[2][1] * m[0][2])
+ m[2][0] * (m[0][1] * m[1][2] - m[1][1] * m[0][2]);
}
mat3 inverse(const mat3 &m)
{
float OneOverDeterminant = 1.0f / determinant(m);
mat3 Inverse;
Inverse[0][0] = +(m[1][1] * m[2][2] - m[2][1] * m[1][2]) * OneOverDeterminant;
Inverse[1][0] = -(m[1][0] * m[2][2] - m[2][0] * m[1][2]) * OneOverDeterminant;
Inverse[2][0] = +(m[1][0] * m[2][1] - m[2][0] * m[1][1]) * OneOverDeterminant;
Inverse[0][1] = -(m[0][1] * m[2][2] - m[2][1] * m[0][2]) * OneOverDeterminant;
Inverse[1][1] = +(m[0][0] * m[2][2] - m[2][0] * m[0][2]) * OneOverDeterminant;
Inverse[2][1] = -(m[0][0] * m[2][1] - m[2][0] * m[0][1]) * OneOverDeterminant;
Inverse[0][2] = +(m[0][1] * m[1][2] - m[1][1] * m[0][2]) * OneOverDeterminant;
Inverse[1][2] = -(m[0][0] * m[1][2] - m[1][0] * m[0][2]) * OneOverDeterminant;
Inverse[2][2] = +(m[0][0] * m[1][1] - m[1][0] * m[0][1]) * OneOverDeterminant;
return Inverse;
}
mat4 inverse(const mat4 &m)
{
float Coef00 = m[2][2] * m[3][3] - m[3][2] * m[2][3];
float Coef02 = m[1][2] * m[3][3] - m[3][2] * m[1][3];
float Coef03 = m[1][2] * m[2][3] - m[2][2] * m[1][3];
float Coef04 = m[2][1] * m[3][3] - m[3][1] * m[2][3];
float Coef06 = m[1][1] * m[3][3] - m[3][1] * m[1][3];
float Coef07 = m[1][1] * m[2][3] - m[2][1] * m[1][3];
float Coef08 = m[2][1] * m[3][2] - m[3][1] * m[2][2];
float Coef10 = m[1][1] * m[3][2] - m[3][1] * m[1][2];
float Coef11 = m[1][1] * m[2][2] - m[2][1] * m[1][2];
float Coef12 = m[2][0] * m[3][3] - m[3][0] * m[2][3];
float Coef14 = m[1][0] * m[3][3] - m[3][0] * m[1][3];
float Coef15 = m[1][0] * m[2][3] - m[2][0] * m[1][3];
float Coef16 = m[2][0] * m[3][2] - m[3][0] * m[2][2];
float Coef18 = m[1][0] * m[3][2] - m[3][0] * m[1][2];
float Coef19 = m[1][0] * m[2][2] - m[2][0] * m[1][2];
float Coef20 = m[2][0] * m[3][1] - m[3][0] * m[2][1];
float Coef22 = m[1][0] * m[3][1] - m[3][0] * m[1][1];
float Coef23 = m[1][0] * m[2][1] - m[2][0] * m[1][1];
vec4 Fac0(Coef00, Coef00, Coef02, Coef03);
vec4 Fac1(Coef04, Coef04, Coef06, Coef07);
vec4 Fac2(Coef08, Coef08, Coef10, Coef11);
vec4 Fac3(Coef12, Coef12, Coef14, Coef15);
vec4 Fac4(Coef16, Coef16, Coef18, Coef19);
vec4 Fac5(Coef20, Coef20, Coef22, Coef23);
vec4 Vec0(m[1][0], m[0][0], m[0][0], m[0][0]);
vec4 Vec1(m[1][1], m[0][1], m[0][1], m[0][1]);
vec4 Vec2(m[1][2], m[0][2], m[0][2], m[0][2]);
vec4 Vec3(m[1][3], m[0][3], m[0][3], m[0][3]);
vec4 Inv0(Vec1 * Fac0 - Vec2 * Fac1 + Vec3 * Fac2);
vec4 Inv1(Vec0 * Fac0 - Vec2 * Fac3 + Vec3 * Fac4);
vec4 Inv2(Vec0 * Fac1 - Vec1 * Fac3 + Vec3 * Fac5);
vec4 Inv3(Vec0 * Fac2 - Vec1 * Fac4 + Vec2 * Fac5);
vec4 SignA(+1, -1, +1, -1);
vec4 SignB(-1, +1, -1, +1);
mat4 Inverse(Inv0 * SignA, Inv1 * SignB, Inv2 * SignA, Inv3 * SignB);
vec4 Row0(Inverse[0][0], Inverse[1][0], Inverse[2][0], Inverse[3][0]);
vec4 Dot0(m[0] * Row0);
float Dot1 = (Dot0.x + Dot0.y) + (Dot0.z + Dot0.w);
float OneOverDeterminant = 1.0f / Dot1;
return Inverse * OneOverDeterminant;
}
void decompose(const mat4 &m, vec3 &scale, quat &rotation, vec3 &trans)
{
vec4 rot;
// Make a lot of assumptions.
// We don't need skew, nor perspective.
// Isolate translation.
trans = m[3].xyz();
vec3 cols[3];
cols[0] = m[0].xyz();
cols[1] = m[1].xyz();
cols[2] = m[2].xyz();
scale.x = length(cols[0]);
scale.y = length(cols[1]);
scale.z = length(cols[2]);
// Isolate scale.
cols[0] /= scale.x;
cols[1] /= scale.y;
cols[2] /= scale.z;
vec3 pdum3 = cross(cols[1], cols[2]);
if (dot(cols[0], pdum3) < 0.0f)
{
scale = -scale;
cols[0] = -cols[0];
cols[1] = -cols[1];
cols[2] = -cols[2];
}
int i, j, k = 0;
float root, trace = cols[0].x + cols[1].y + cols[2].z;
if (trace > 0.0f)
{
root = sqrt(trace + 1.0f);
rot.w = 0.5f * root;
root = 0.5f / root;
rot.x = root * (cols[1].z - cols[2].y);
rot.y = root * (cols[2].x - cols[0].z);
rot.z = root * (cols[0].y - cols[1].x);
}
else
{
static const int Next[3] = {1, 2, 0};
i = 0;
if (cols[1].y > cols[0].x) i = 1;
if (cols[2].z > cols[i][i]) i = 2;
j = Next[i];
k = Next[j];
root = sqrt(cols[i][i] - cols[j][j] - cols[k][k] + 1.0f);
rot[i] = 0.5f * root;
root = 0.5f / root;
rot[j] = root * (cols[i][j] + cols[j][i]);
rot[k] = root * (cols[i][k] + cols[k][i]);
rot.w = root * (cols[j][k] - cols[k][j]);
}
rotation = quat(rot);
}
mat4 ortho(float left, float right, float bottom, float top, float near, float far)
{
mat4 result(1.0f);
result[0][0] = 2.0f / (right - left);
result[1][1] = 2.0f / (top - bottom);
result[3][0] = -(right + left) / (right - left);
result[3][1] = -(top + bottom) / (top - bottom);
result[2][2] = 1.0f / (far - near);
result[3][2] = 1.0f + near / (far - near);
result[0].y *= -1.0f;
result[1].y *= -1.0f;
result[2].y *= -1.0f;
result[3].y *= -1.0f;
return result;
}
mat4 frustum(float left, float right, float bottom, float top, float near, float far)
{
mat4 result(0.0f);
result[0][0] = (2.0f * near) / (right - left);
result[1][1] = (2.0f * near) / (top - bottom);
result[2][0] = (right + left) / (right - left);
result[2][1] = (top + bottom) / (top - bottom);
// Inverse Z
if (far == InfiniteFarPlane)
{
result[3][2] = -near;
}
else
{
result[2][2] = -1.0f - far / (near - far);
result[3][2] = -(far * near) / (near - far);
}
result[2][3] = -1.0f;
// Y-flip so we don't have to bother with negative viewport heights.
result[0].y *= -1.0f;
result[1].y *= -1.0f;
result[2].y *= -1.0f;
result[3].y *= -1.0f;
return result;
}
mat4 perspective(float fovy, float aspect, float near, float far)
{
float tanHalfFovy = tan(fovy / 2.0f);
mat4 result(0.0f);
result[0][0] = 1.0f / (aspect * tanHalfFovy);
result[1][1] = 1.0f / (tanHalfFovy);
// Inverse Z
if (far == InfiniteFarPlane)
{
result[3][2] = near;
}
else
{
result[2][2] = -1.0f - far / (near - far);
result[3][2] = -(far * near) / (near - far);
}
result[2][3] = -1.0f;
// Y-flip so we don't have to bother with negative viewport heights.
result[0].y *= -1.0f;
result[1].y *= -1.0f;
result[2].y *= -1.0f;
result[3].y *= -1.0f;
return result;
}
void transpose(mat4 &dst, const mat4 &src)
{
#if __SSE__
__m128 r0 = _mm_loadu_ps(src[0].data);
__m128 r1 = _mm_loadu_ps(src[1].data);
__m128 r2 = _mm_loadu_ps(src[2].data);
__m128 r3 = _mm_loadu_ps(src[3].data);
_MM_TRANSPOSE4_PS(r0, r1, r2, r3);
_mm_storeu_ps(dst[0].data, r0);
_mm_storeu_ps(dst[1].data, r1);
_mm_storeu_ps(dst[2].data, r2);
_mm_storeu_ps(dst[3].data, r3);
#elif defined(__ARM_NEON)
float32x4x4_t a = vld4q_f32(src[0].data);
vst1q_f32(dst[0].data, a.val[0]);
vst1q_f32(dst[1].data, a.val[1]);
vst1q_f32(dst[2].data, a.val[2]);
vst1q_f32(dst[3].data, a.val[3]);
#else
dst = transpose(src);
#endif
}
void transpose_to_affine(vec4 dst[3], const mat4 &src)
{
#if __SSE__
__m128 r0 = _mm_loadu_ps(src[0].data);
__m128 r1 = _mm_loadu_ps(src[1].data);
__m128 r2 = _mm_loadu_ps(src[2].data);
__m128 r3 = _mm_loadu_ps(src[3].data);
_MM_TRANSPOSE4_PS(r0, r1, r2, r3);
_mm_storeu_ps(dst[0].data, r0);
_mm_storeu_ps(dst[1].data, r1);
_mm_storeu_ps(dst[2].data, r2);
#elif defined(__ARM_NEON)
float32x4x4_t a = vld4q_f32(src[0].data);
vst1q_f32(dst[0].data, a.val[0]);
vst1q_f32(dst[1].data, a.val[1]);
vst1q_f32(dst[2].data, a.val[2]);
#else
mat4 m = transpose(src);
for (int i = 0; i < 3; i++)
dst[i] = m[i];
#endif
}
void transpose_from_affine(mat4 &dst, const vec4 src[3])
{
#if __SSE__
__m128 r0 = _mm_loadu_ps(src[0].data);
__m128 r1 = _mm_loadu_ps(src[1].data);
__m128 r2 = _mm_loadu_ps(src[2].data);
__m128 r3 = _mm_set_ps(1, 0, 0, 0);
_MM_TRANSPOSE4_PS(r0, r1, r2, r3);
_mm_storeu_ps(dst[0].data, r0);
_mm_storeu_ps(dst[1].data, r1);
_mm_storeu_ps(dst[2].data, r2);
_mm_storeu_ps(dst[3].data, r3);
#elif defined(__ARM_NEON)
alignas(16) static const float r3_data[] = { 0, 0, 0, 1 };
float32x4_t r0 = vld1q_f32(src[0].data);
float32x4_t r1 = vld1q_f32(src[1].data);
float32x4_t r2 = vld1q_f32(src[2].data);
float32x4_t r3 = vld1q_f32(r3_data);
float32x4x4_t r = { r0, r1, r2, r3 };
vst4q_f32(dst[0].data, r);
#else
mat4 m = transpose(src);
for (int i = 0; i < 3; i++)
dst[i] = m[i];
#endif
}
void mat_affine::to_mat4(muglm::mat4 &m) const
{
transpose_from_affine(m, vec);
}
mat4 mat_affine::to_mat4() const
{
mat4 m;
to_mat4(m);
return m;
}
float mat_affine::get_uniform_scale() const
{
return length(vec[0].xyz());
}
vec3 mat_affine::get_translation() const
{
// this * vec4(0, 0, 0, 1)
return { vec[0].w, vec[1].w, vec[2].w };
}
vec3 mat_affine::get_forward() const
{
// this * vec4(0, 0, -1, 0).
return { -vec[0].z, -vec[1].z, -vec[2].z };
}
vec3 mat_affine::get_right() const
{
return { vec[0].x, vec[1].x, vec[2].x };
}
vec3 mat_affine::get_up() const
{
return { vec[0].y, vec[1].y, vec[2].y };
}
}