renderer: optimise shader usage further, split shaders and add more caching (#12992)

* shader: split CM rgba/rgbx into discard ones

make it branchless if we have no discards.

* shader: ensure we dont stall on vbo uv buffer

if we render a new texture before the previous was done gpu wise its
going to stall until done, call glBufferData to orphan the data.
this allows the driver to return a new memory block immediately
if the GPU is still reading from the previous one

* protocols: ensure we reset GL_PACK_ALIGNMENT

reset GL_PACK_ALIGNMENT back to the default initial value of 4

* shader: use unsigned short in VAO

loose a tiny bit of precision but gain massive bandwidth reductions.
use GL_UNSIGNED_SHORT and set it as normalized. clamp and round the UV
for uint16_t in customUv.

* shader: interleave vertex buffers

use std::array for fullverts, use a single interleaved buffer for
position and uv, should in theory improve cache locality. and also remove
the need to have two buffers around.

* shader: revert precision drop

we need the float precision because we might have 1.01 or similiar
floats entering CM shader maths, and rounding/clamping those means the
maths turns out wrong. so revert back to float, sadly higher bandwidth
usage.

* update doColorManagement api

* convert primaries to XYZ on cpu

* remove unused primaries uniform

---------

Co-authored-by: UjinT34 <ujint34@mail.ru>

13 files changed, 186 insertions, 96 deletions

diff --git a/src/protocols/Screencopy.cpp b/src/protocols/Screencopy.cpp
index d66c5342..ac7146b4 100644
--- a/src/protocols/Screencopy.cpp
+++ b/src/protocols/Screencopy.cpp
@@ -425,6 +425,7 @@ bool CScreencopyFrame::copyShm() {
         }
     }
 
+    glPixelStorei(GL_PACK_ALIGNMENT, 4);
     g_pHyprOpenGL->m_renderData.pMonitor.reset();
 
     glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
diff --git a/src/protocols/ToplevelExport.cpp b/src/protocols/ToplevelExport.cpp
index 7549425c..7b7c0a31 100644
--- a/src/protocols/ToplevelExport.cpp
+++ b/src/protocols/ToplevelExport.cpp
@@ -334,6 +334,7 @@ bool CToplevelExportFrame::copyShm(const Time::steady_tp& now) {
     }
 
     outFB.unbind();
+    glPixelStorei(GL_PACK_ALIGNMENT, 4);
     glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
 
     return true;
diff --git a/src/render/OpenGL.cpp b/src/render/OpenGL.cpp
index 84a2a061..a88d5315 100644
--- a/src/render/OpenGL.cpp
+++ b/src/render/OpenGL.cpp
@@ -43,6 +43,7 @@
 #include <fcntl.h>
 #include <gbm.h>
 #include <filesystem>
+#include <cstring>
 #include "./shaders/Shaders.hpp"
 
 using namespace Hyprutils::OS;
@@ -896,7 +897,9 @@ bool CHyprOpenGLImpl::initShaders() {
         else {
             std::vector<SFragShaderDesc> CM_SHADERS = {{
                 {SH_FRAG_CM_RGBA, "CMrgba.frag"},
+                {SH_FRAG_CM_RGBA_DISCARD, "CMrgbadiscard.frag"},
                 {SH_FRAG_CM_RGBX, "CMrgbx.frag"},
+                {SH_FRAG_CM_RGBX_DISCARD, "CMrgbadiscard.frag"},
                 {SH_FRAG_CM_BLURPREPARE, "CMblurprepare.frag"},
                 {SH_FRAG_CM_BORDER1, "CMborder.frag"},
             }};
@@ -1228,13 +1231,14 @@ void CHyprOpenGLImpl::passCMUniforms(WP<CShader> shader, const NColorManagement:
 
     shader->setUniformInt(SHADER_TARGET_TF, targetImageDescription->value().transferFunction);
 
-    const auto                   targetPrimaries = targetImageDescription->getPrimaries();
-
-    const std::array<GLfloat, 8> glTargetPrimaries = {
-        targetPrimaries->value().red.x,  targetPrimaries->value().red.y,  targetPrimaries->value().green.x, targetPrimaries->value().green.y,
-        targetPrimaries->value().blue.x, targetPrimaries->value().blue.y, targetPrimaries->value().white.x, targetPrimaries->value().white.y,
+    const auto                   targetPrimaries      = targetImageDescription->getPrimaries();
+    const auto                   mat                  = targetPrimaries->value().toXYZ().mat();
+    const std::array<GLfloat, 9> glTargetPrimariesXYZ = {
+        mat[0][0], mat[1][0], mat[2][0], //
+        mat[0][1], mat[1][1], mat[2][1], //
+        mat[0][2], mat[1][2], mat[2][2], //
     };
-    shader->setUniformMatrix4x2fv(SHADER_TARGET_PRIMARIES, 1, false, glTargetPrimaries);
+    shader->setUniformMatrix3fv(SHADER_TARGET_PRIMARIES_XYZ, 1, false, glTargetPrimariesXYZ);
 
     const bool needsSDRmod = modifySDR && isSDR2HDR(imageDescription->value(), targetImageDescription->value());
     const bool needsHDRmod = !needsSDRmod && isHDR2SDR(imageDescription->value(), targetImageDescription->value());
@@ -1364,10 +1368,17 @@ void CHyprOpenGLImpl::renderTextureInternal(SP<CTexture> tex, const CBox& box, c
             m_renderData.pMonitor->inFullscreenMode()) /* Fullscreen window with pass cm enabled */;
 
     if (!skipCM && !usingFinalShader) {
-        if (texType == TEXTURE_RGBA)
-            shader = m_shaders->frag[SH_FRAG_CM_RGBA];
-        else if (texType == TEXTURE_RGBX)
-            shader = m_shaders->frag[SH_FRAG_CM_RGBX];
+        if (!data.discardActive) {
+            if (texType == TEXTURE_RGBA)
+                shader = m_shaders->frag[SH_FRAG_CM_RGBA];
+            else if (texType == TEXTURE_RGBX)
+                shader = m_shaders->frag[SH_FRAG_CM_RGBX];
+        } else {
+            if (texType == TEXTURE_RGBA)
+                shader = m_shaders->frag[SH_FRAG_CM_RGBA_DISCARD];
+            else if (texType == TEXTURE_RGBX)
+                shader = m_shaders->frag[SH_FRAG_CM_RGBA_DISCARD];
+        }
 
         shader = useShader(shader);
 
@@ -1487,19 +1498,32 @@ void CHyprOpenGLImpl::renderTextureInternal(SP<CTexture> tex, const CBox& box, c
     }
 
     glBindVertexArray(shader->getUniformLocation(SHADER_SHADER_VAO));
-    if (data.allowCustomUV && m_renderData.primarySurfaceUVTopLeft != Vector2D(-1, -1)) {
-        const float customUVs[] = {
-            m_renderData.primarySurfaceUVBottomRight.x, m_renderData.primarySurfaceUVTopLeft.y,     m_renderData.primarySurfaceUVTopLeft.x,
-            m_renderData.primarySurfaceUVTopLeft.y,     m_renderData.primarySurfaceUVBottomRight.x, m_renderData.primarySurfaceUVBottomRight.y,
-            m_renderData.primarySurfaceUVTopLeft.x,     m_renderData.primarySurfaceUVBottomRight.y,
-        };
+    glBindBuffer(GL_ARRAY_BUFFER, shader->getUniformLocation(SHADER_SHADER_VBO));
 
-        glBindBuffer(GL_ARRAY_BUFFER, shader->getUniformLocation(SHADER_SHADER_VBO_UV));
-        glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(customUVs), customUVs);
-    } else {
-        glBindBuffer(GL_ARRAY_BUFFER, shader->getUniformLocation(SHADER_SHADER_VBO_UV));
-        glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(fullVerts), fullVerts);
-    }
+    // this tells GPU can keep reading the old block for previous draws while the CPU writes to a new one.
+    // to avoid stalls if renderTextureInternal is called multiple times on same renderpass
+    // at the cost of some temporar vram usage.
+    glBufferData(GL_ARRAY_BUFFER, sizeof(fullVerts), nullptr, GL_DYNAMIC_DRAW);
+
+    auto verts = fullVerts;
+
+    if (data.allowCustomUV && m_renderData.primarySurfaceUVTopLeft != Vector2D(-1, -1)) {
+        const float u0 = m_renderData.primarySurfaceUVTopLeft.x;
+        const float v0 = m_renderData.primarySurfaceUVTopLeft.y;
+        const float u1 = m_renderData.primarySurfaceUVBottomRight.x;
+        const float v1 = m_renderData.primarySurfaceUVBottomRight.y;
+
+        verts[0].u = u0;
+        verts[0].v = v0;
+        verts[1].u = u0;
+        verts[1].v = v1;
+        verts[2].u = u1;
+        verts[2].v = v0;
+        verts[3].u = u1;
+        verts[3].v = v1;
+    }
+
+    glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(verts), verts.data());
 
     if (!m_renderData.clipBox.empty() || !m_renderData.clipRegion.empty()) {
         CRegion damageClip = m_renderData.clipBox;
diff --git a/src/render/OpenGL.hpp b/src/render/OpenGL.hpp
index 857cb891..4189e32b 100644
--- a/src/render/OpenGL.hpp
+++ b/src/render/OpenGL.hpp
@@ -35,13 +35,19 @@
 struct gbm_device;
 class CHyprRenderer;
 
-inline const float fullVerts[] = {
-    1, 0, // top right
-    0, 0, // top left
-    1, 1, // bottom right
-    0, 1, // bottom left
+struct SVertex {
+    float x, y; // position
+    float u, v; // uv
 };
-inline const float fanVertsFull[] = {-1.0f, -1.0f, 1.0f, -1.0f, 1.0f, 1.0f, -1.0f, 1.0f};
+
+constexpr std::array<SVertex, 4> fullVerts = {{
+    {0.0f, 0.0f, 0.0f, 0.0f}, // top-left
+    {0.0f, 1.0f, 0.0f, 1.0f}, // bottom-left
+    {1.0f, 0.0f, 1.0f, 0.0f}, // top-right
+    {1.0f, 1.0f, 1.0f, 1.0f}, // bottom-right
+}};
+
+inline const float               fanVertsFull[] = {-1.0f, -1.0f, 1.0f, -1.0f, 1.0f, 1.0f, -1.0f, 1.0f};
 
 enum eDiscardMode : uint8_t {
     DISCARD_OPAQUE = 1,
@@ -98,7 +104,9 @@ enum ePreparedFragmentShader : uint8_t {
     SH_FRAG_BORDER1,
     SH_FRAG_GLITCH,
     SH_FRAG_CM_RGBA,
+    SH_FRAG_CM_RGBA_DISCARD,
     SH_FRAG_CM_RGBX,
+    SH_FRAG_CM_RGBX_DISCARD,
 
     SH_FRAG_LAST,
 };
diff --git a/src/render/Shader.cpp b/src/render/Shader.cpp
index 635e1328..5f62232c 100644
--- a/src/render/Shader.cpp
+++ b/src/render/Shader.cpp
@@ -127,19 +127,19 @@ void CShader::getUniformLocations() {
     m_uniformLocations[SHADER_TEX_TYPE]    = getUniform("texType");
 
     // shader has #include "CM.glsl"
-    m_uniformLocations[SHADER_SKIP_CM]           = getUniform("skipCM");
-    m_uniformLocations[SHADER_SOURCE_TF]         = getUniform("sourceTF");
-    m_uniformLocations[SHADER_TARGET_TF]         = getUniform("targetTF");
-    m_uniformLocations[SHADER_SRC_TF_RANGE]      = getUniform("srcTFRange");
-    m_uniformLocations[SHADER_DST_TF_RANGE]      = getUniform("dstTFRange");
-    m_uniformLocations[SHADER_TARGET_PRIMARIES]  = getUniform("targetPrimaries");
-    m_uniformLocations[SHADER_MAX_LUMINANCE]     = getUniform("maxLuminance");
-    m_uniformLocations[SHADER_SRC_REF_LUMINANCE] = getUniform("srcRefLuminance");
-    m_uniformLocations[SHADER_DST_MAX_LUMINANCE] = getUniform("dstMaxLuminance");
-    m_uniformLocations[SHADER_DST_REF_LUMINANCE] = getUniform("dstRefLuminance");
-    m_uniformLocations[SHADER_SDR_SATURATION]    = getUniform("sdrSaturation");
-    m_uniformLocations[SHADER_SDR_BRIGHTNESS]    = getUniform("sdrBrightnessMultiplier");
-    m_uniformLocations[SHADER_CONVERT_MATRIX]    = getUniform("convertMatrix");
+    m_uniformLocations[SHADER_SKIP_CM]              = getUniform("skipCM");
+    m_uniformLocations[SHADER_SOURCE_TF]            = getUniform("sourceTF");
+    m_uniformLocations[SHADER_TARGET_TF]            = getUniform("targetTF");
+    m_uniformLocations[SHADER_SRC_TF_RANGE]         = getUniform("srcTFRange");
+    m_uniformLocations[SHADER_DST_TF_RANGE]         = getUniform("dstTFRange");
+    m_uniformLocations[SHADER_TARGET_PRIMARIES_XYZ] = getUniform("targetPrimariesXYZ");
+    m_uniformLocations[SHADER_MAX_LUMINANCE]        = getUniform("maxLuminance");
+    m_uniformLocations[SHADER_SRC_REF_LUMINANCE]    = getUniform("srcRefLuminance");
+    m_uniformLocations[SHADER_DST_MAX_LUMINANCE]    = getUniform("dstMaxLuminance");
+    m_uniformLocations[SHADER_DST_REF_LUMINANCE]    = getUniform("dstRefLuminance");
+    m_uniformLocations[SHADER_SDR_SATURATION]       = getUniform("sdrSaturation");
+    m_uniformLocations[SHADER_SDR_BRIGHTNESS]       = getUniform("sdrBrightnessMultiplier");
+    m_uniformLocations[SHADER_CONVERT_MATRIX]       = getUniform("convertMatrix");
     //
     m_uniformLocations[SHADER_TEX]                 = getUniform("tex");
     m_uniformLocations[SHADER_ALPHA]               = getUniform("alpha");
@@ -208,7 +208,7 @@ void CShader::getUniformLocations() {
 }
 
 void CShader::createVao() {
-    GLuint shaderVao = 0, shaderVbo = 0, shaderVboUv = 0;
+    GLuint shaderVao = 0, shaderVbo = 0;
 
     glGenVertexArrays(1, &shaderVao);
     glBindVertexArray(shaderVao);
@@ -216,30 +216,26 @@ void CShader::createVao() {
     if (m_uniformLocations[SHADER_POS_ATTRIB] != -1) {
         glGenBuffers(1, &shaderVbo);
         glBindBuffer(GL_ARRAY_BUFFER, shaderVbo);
-        glBufferData(GL_ARRAY_BUFFER, sizeof(fullVerts), fullVerts, GL_STATIC_DRAW);
+        glBufferData(GL_ARRAY_BUFFER, sizeof(fullVerts), fullVerts.data(), GL_DYNAMIC_DRAW);
         glEnableVertexAttribArray(m_uniformLocations[SHADER_POS_ATTRIB]);
-        glVertexAttribPointer(m_uniformLocations[SHADER_POS_ATTRIB], 2, GL_FLOAT, GL_FALSE, 0, nullptr);
+        glVertexAttribPointer(m_uniformLocations[SHADER_POS_ATTRIB], 2, GL_FLOAT, GL_FALSE, sizeof(SVertex), (void*)offsetof(SVertex, x));
     }
 
     // UV VBO (dynamic, may be updated per frame)
-    if (m_uniformLocations[SHADER_TEX_ATTRIB] != -1) {
-        glGenBuffers(1, &shaderVboUv);
-        glBindBuffer(GL_ARRAY_BUFFER, shaderVboUv);
-        glBufferData(GL_ARRAY_BUFFER, sizeof(fullVerts), fullVerts, GL_DYNAMIC_DRAW); // Initial dummy UVs
+    if (m_uniformLocations[SHADER_TEX_ATTRIB] != -1 && shaderVbo != 0) {
+        glBindBuffer(GL_ARRAY_BUFFER, shaderVbo);
         glEnableVertexAttribArray(m_uniformLocations[SHADER_TEX_ATTRIB]);
-        glVertexAttribPointer(m_uniformLocations[SHADER_TEX_ATTRIB], 2, GL_FLOAT, GL_FALSE, 0, nullptr);
+        glVertexAttribPointer(m_uniformLocations[SHADER_TEX_ATTRIB], 2, GL_FLOAT, GL_FALSE, sizeof(SVertex), (void*)offsetof(SVertex, u));
     }
 
     glBindVertexArray(0);
     glBindBuffer(GL_ARRAY_BUFFER, 0);
 
-    m_uniformLocations[SHADER_SHADER_VAO]     = shaderVao;
-    m_uniformLocations[SHADER_SHADER_VBO_POS] = shaderVbo;
-    m_uniformLocations[SHADER_SHADER_VBO_UV]  = shaderVboUv;
+    m_uniformLocations[SHADER_SHADER_VAO] = shaderVao;
+    m_uniformLocations[SHADER_SHADER_VBO] = shaderVbo;
 
     RASSERT(m_uniformLocations[SHADER_SHADER_VAO] >= 0, "SHADER_SHADER_VAO could not be created");
-    RASSERT(m_uniformLocations[SHADER_SHADER_VBO_POS] >= 0, "SHADER_SHADER_VBO_POS could not be created");
-    RASSERT(m_uniformLocations[SHADER_SHADER_VBO_UV] >= 0, "SHADER_SHADER_VBO_UV could not be created");
+    RASSERT(m_uniformLocations[SHADER_SHADER_VBO] >= 0, "SHADER_SHADER_VBO_POS could not be created");
 }
 
 void CShader::setUniformInt(eShaderUniform location, GLint v0) {
@@ -390,11 +386,10 @@ void CShader::destroy() {
     if (m_program == 0)
         return;
 
-    GLuint shaderVao, shaderVbo, shaderVboUv;
+    GLuint shaderVao, shaderVbo;
 
-    shaderVao   = m_uniformLocations[SHADER_SHADER_VAO] == -1 ? 0 : m_uniformLocations[SHADER_SHADER_VAO];
-    shaderVbo   = m_uniformLocations[SHADER_SHADER_VBO_POS] == -1 ? 0 : m_uniformLocations[SHADER_SHADER_VBO_POS];
-    shaderVboUv = m_uniformLocations[SHADER_SHADER_VBO_UV] == -1 ? 0 : m_uniformLocations[SHADER_SHADER_VBO_UV];
+    shaderVao = m_uniformLocations[SHADER_SHADER_VAO] == -1 ? 0 : m_uniformLocations[SHADER_SHADER_VAO];
+    shaderVbo = m_uniformLocations[SHADER_SHADER_VBO] == -1 ? 0 : m_uniformLocations[SHADER_SHADER_VBO];
 
     if (shaderVao)
         glDeleteVertexArrays(1, &shaderVao);
@@ -402,9 +397,6 @@ void CShader::destroy() {
     if (shaderVbo)
         glDeleteBuffers(1, &shaderVbo);
 
-    if (shaderVboUv)
-        glDeleteBuffers(1, &shaderVboUv);
-
     glDeleteProgram(m_program);
     m_program = 0;
 }
diff --git a/src/render/Shader.hpp b/src/render/Shader.hpp
index 6ab8248b..9f871c0e 100644
--- a/src/render/Shader.hpp
+++ b/src/render/Shader.hpp
@@ -14,7 +14,7 @@ enum eShaderUniform : uint8_t {
     SHADER_TARGET_TF,
     SHADER_SRC_TF_RANGE,
     SHADER_DST_TF_RANGE,
-    SHADER_TARGET_PRIMARIES,
+    SHADER_TARGET_PRIMARIES_XYZ,
     SHADER_MAX_LUMINANCE,
     SHADER_SRC_REF_LUMINANCE,
     SHADER_DST_MAX_LUMINANCE,
@@ -31,8 +31,7 @@ enum eShaderUniform : uint8_t {
     SHADER_DISCARD_ALPHA,
     SHADER_DISCARD_ALPHA_VALUE,
     SHADER_SHADER_VAO,
-    SHADER_SHADER_VBO_POS,
-    SHADER_SHADER_VBO_UV,
+    SHADER_SHADER_VBO,
     SHADER_TOP_LEFT,
     SHADER_BOTTOM_RIGHT,
     SHADER_FULL_SIZE,
diff --git a/src/render/shaders/glsl/CM.glsl b/src/render/shaders/glsl/CM.glsl
index 362d7cfb..8b02c5ee 100644
--- a/src/render/shaders/glsl/CM.glsl
+++ b/src/render/shaders/glsl/CM.glsl
@@ -401,13 +401,12 @@ vec4 tonemap(vec4 color, mat3 dstXYZ) {
     return vec4(fromLMS * toLinear(vec4(ICtCpPQInv * ICtCp, 1.0), CM_TRANSFER_FUNCTION_ST2084_PQ).rgb * HDR_MAX_LUMINANCE * refScale, color[3]); 
 }
 
-vec4 doColorManagement(vec4 pixColor, int srcTF, int dstTF, mat4x2 dstPrimaries) {
+vec4 doColorManagement(vec4 pixColor, int srcTF, int dstTF, mat3 dstxyz) {
     pixColor.rgb /= max(pixColor.a, 0.001);
     pixColor.rgb = toLinearRGB(pixColor.rgb, srcTF);
     pixColor.rgb = convertMatrix * pixColor.rgb;
     pixColor = toNit(pixColor, srcTFRange);
     pixColor.rgb *= pixColor.a;
-    mat3 dstxyz = primaries2xyz(dstPrimaries);
     pixColor = tonemap(pixColor, dstxyz);
     pixColor = fromLinearNit(pixColor, dstTF, dstTFRange);
     if ((srcTF == CM_TRANSFER_FUNCTION_SRGB || srcTF == CM_TRANSFER_FUNCTION_GAMMA22) && dstTF == CM_TRANSFER_FUNCTION_ST2084_PQ) {
diff --git a/src/render/shaders/glsl/CMborder.frag b/src/render/shaders/glsl/CMborder.frag
index 3c9540a7..079f940d 100644
--- a/src/render/shaders/glsl/CMborder.frag
+++ b/src/render/shaders/glsl/CMborder.frag
@@ -6,7 +6,7 @@ in vec2 v_texcoord;
 
 uniform int sourceTF; // eTransferFunction
 uniform int targetTF; // eTransferFunction
-uniform mat4x2 targetPrimaries;
+uniform mat3 targetPrimariesXYZ;
 
 uniform vec2 fullSizeUntransformed;
 uniform float radiusOuter;
@@ -90,7 +90,7 @@ void main() {
     pixColor = getColorForCoord(v_texcoord);
     pixColor.rgb *= pixColor[3];
 
-    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimaries);
+    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
 
     pixColor *= alpha * additionalAlpha;
 
diff --git a/src/render/shaders/glsl/CMrgba.frag b/src/render/shaders/glsl/CMrgba.frag
index 1e4e024d..b6696649 100644
--- a/src/render/shaders/glsl/CMrgba.frag
+++ b/src/render/shaders/glsl/CMrgba.frag
@@ -7,14 +7,9 @@ uniform sampler2D tex;
 
 uniform int sourceTF; // eTransferFunction
 uniform int targetTF; // eTransferFunction
-uniform mat4x2 targetPrimaries;
+uniform mat3 targetPrimariesXYZ;
 
 uniform float alpha;
-
-uniform bool discardOpaque;
-uniform bool discardAlpha;
-uniform float discardAlphaValue;
-
 uniform bool applyTint;
 uniform vec3 tint;
 
@@ -25,14 +20,8 @@ layout(location = 0) out vec4 fragColor;
 void main() {
     vec4 pixColor = texture(tex, v_texcoord);
 
-    if (discardOpaque && pixColor.a * alpha == 1.0)
-        discard;
-
-    if (discardAlpha && pixColor.a <= discardAlphaValue)
-        discard;
-
     // this shader shouldn't be used when skipCM == 1
-    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimaries);
+    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
 
     if (applyTint)
         pixColor.rgb *= tint;
diff --git a/src/render/shaders/glsl/CMrgbadiscard.frag b/src/render/shaders/glsl/CMrgbadiscard.frag
new file mode 100644
index 00000000..9be2ed97
--- /dev/null
+++ b/src/render/shaders/glsl/CMrgbadiscard.frag
@@ -0,0 +1,44 @@
+#version 300 es
+#extension GL_ARB_shading_language_include : enable
+
+precision highp float;
+in vec2 v_texcoord;
+uniform sampler2D tex;
+
+uniform int sourceTF; // eTransferFunction
+uniform int targetTF; // eTransferFunction
+uniform mat3 targetPrimariesXYZ;
+
+uniform float alpha;
+
+uniform bool discardOpaque;
+uniform bool discardAlpha;
+uniform float discardAlphaValue;
+
+uniform bool applyTint;
+uniform vec3 tint;
+
+#include "rounding.glsl"
+#include "CM.glsl"
+
+layout(location = 0) out vec4 fragColor;
+void main() {
+    vec4 pixColor = texture(tex, v_texcoord);
+
+    if (discardOpaque && pixColor.a * alpha == 1.0)
+        discard;
+
+    if (discardAlpha && pixColor.a <= discardAlphaValue)
+        discard;
+
+    // this shader shouldn't be used when skipCM == 1
+    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
+
+    if (applyTint)
+        pixColor.rgb *= tint;
+
+    if (radius > 0.0)
+        pixColor = rounding(pixColor);
+
+    fragColor = pixColor * alpha;
+}
diff --git a/src/render/shaders/glsl/CMrgbx.frag b/src/render/shaders/glsl/CMrgbx.frag
index e2b1a838..d37328de 100644
--- a/src/render/shaders/glsl/CMrgbx.frag
+++ b/src/render/shaders/glsl/CMrgbx.frag
@@ -7,14 +7,9 @@ uniform sampler2D tex;
 
 uniform int sourceTF; // eTransferFunction
 uniform int targetTF; // eTransferFunction
-uniform mat4x2 targetPrimaries;
+uniform mat3 targetPrimariesXYZ;
 
 uniform float alpha;
-
-uniform bool discardOpaque;
-uniform bool discardAlpha;
-uniform float discardAlphaValue;
-
 uniform bool applyTint;
 uniform vec3 tint;
 
@@ -25,14 +20,8 @@ layout(location = 0) out vec4 fragColor;
 void main() {
     vec4 pixColor = vec4(texture(tex, v_texcoord).rgb, 1.0);
 
-    if (discardOpaque && pixColor.a * alpha == 1.0)
-        discard;
-
-    if (discardAlpha && pixColor.a <= discardAlphaValue)
-        discard;
-
     // this shader shouldn't be used when skipCM == 1
-    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimaries);
+    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
 
     if (applyTint)
         pixColor.rgb *= tint;
diff --git a/src/render/shaders/glsl/CMrgbxdiscard.frag b/src/render/shaders/glsl/CMrgbxdiscard.frag
new file mode 100644
index 00000000..a4c05d00
--- /dev/null
+++ b/src/render/shaders/glsl/CMrgbxdiscard.frag
@@ -0,0 +1,44 @@
+#version 300 es
+#extension GL_ARB_shading_language_include : enable
+
+precision highp float;
+in vec2 v_texcoord;
+uniform sampler2D tex;
+
+uniform int sourceTF; // eTransferFunction
+uniform int targetTF; // eTransferFunction
+uniform mat3 targetPrimariesXYZ;
+
+uniform float alpha;
+
+uniform bool discardOpaque;
+uniform bool discardAlpha;
+uniform float discardAlphaValue;
+
+uniform bool applyTint;
+uniform vec3 tint;
+
+#include "rounding.glsl"
+#include "CM.glsl"
+
+layout(location = 0) out vec4 fragColor;
+void main() {
+    vec4 pixColor = vec4(texture(tex, v_texcoord).rgb, 1.0);
+
+    if (discardOpaque && pixColor.a * alpha == 1.0)
+        discard;
+
+    if (discardAlpha && pixColor.a <= discardAlphaValue)
+        discard;
+
+    // this shader shouldn't be used when skipCM == 1
+    pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
+
+    if (applyTint)
+        pixColor.rgb *= tint;
+
+    if (radius > 0.0)
+        pixColor = rounding(pixColor);
+
+    fragColor = pixColor * alpha;
+}
diff --git a/src/render/shaders/glsl/shadow.frag b/src/render/shaders/glsl/shadow.frag
index b6fdf6ee..71e96ddb 100644
--- a/src/render/shaders/glsl/shadow.frag
+++ b/src/render/shaders/glsl/shadow.frag
@@ -8,7 +8,7 @@ in vec2 v_texcoord;
 uniform int skipCM;
 uniform int sourceTF; // eTransferFunction
 uniform int targetTF; // eTransferFunction
-uniform mat4x2 targetPrimaries;
+uniform mat3 targetPrimariesXYZ;
 
 uniform vec2 topLeft;
 uniform vec2 bottomRight;
@@ -93,7 +93,7 @@ void main() {
     pixColor.rgb *= pixColor[3];
 
 	if (skipCM == 0)
-        pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimaries);
+        pixColor = doColorManagement(pixColor, sourceTF, targetTF, targetPrimariesXYZ);
 
 	fragColor = pixColor;
 }
\ No newline at end of file