From 627aa610fc93398ac51129096ce8fb30731af341 Mon Sep 17 00:00:00 2001
From: Andreas Grois <andi@grois.info>
Date: Sun, 11 Mar 2018 21:27:11 +0100
Subject: Split buffer in three, one per color (as OpenGL guarantees that one
 can at least have 8 SSBOs), to allow three times larger images. Also make it
 possible to ignore maximum buffer size reported by the driver.

The maximum buffer size limitation is rather strict on radeonsi, so I decided to split the buffer in three, effectively increasing the maximum image size by a factor of three.
While doing so I realized that at least on radeonsi the reported maximum buffer size seems to be off. For this reason I added a new command line switch, that allows to ignore the maximum buffer size check.
For those curious: radeonsi reports a maximum buffer size of 128 MB, but I had no problems when using three buffers of 1098 MB each.
---
 BuddhaTest/Shaders/BuddhaCompute.glsl  | 20 ++++++++++++++------
 BuddhaTest/Shaders/BuddhaFragment.glsl | 16 ++++++++++++----
 2 files changed, 26 insertions(+), 10 deletions(-)

(limited to 'BuddhaTest/Shaders')

diff --git a/BuddhaTest/Shaders/BuddhaCompute.glsl b/BuddhaTest/Shaders/BuddhaCompute.glsl
index 9d8301a..ddcbace 100644
--- a/BuddhaTest/Shaders/BuddhaCompute.glsl
+++ b/BuddhaTest/Shaders/BuddhaCompute.glsl
@@ -2,9 +2,17 @@
 //#version 430
 //layout (local_size_x = 1024) in; //to be safe, we limit our local work group size to 1024. That's the minimum a GL 4.3 capable driver must support.
 
-layout(std430, binding=2) buffer renderedData
+layout(std430, binding=2) buffer renderedDataRed
 {
-        uint counts_SSBO[];
+        uint counts_SSBORed[];
+};
+layout(std430, binding=3) buffer renderedDataGreen
+{
+        uint counts_SSBOGreen[];
+};
+layout(std430, binding=4) buffer renderedDataBlue
+{
+        uint counts_SSBOBlue[];
 };
 
 uniform uint width;
@@ -15,10 +23,10 @@ uniform uvec3 orbitLength;
 
 void addToColorOfCell(uvec2 cell, uvec3 toAdd)
 {
-    uint firstIndex = 3*(cell.x + cell.y * width);
-    atomicAdd(counts_SSBO[firstIndex],toAdd.x);
-    atomicAdd(counts_SSBO[firstIndex+1],toAdd.y);
-    atomicAdd(counts_SSBO[firstIndex+2],toAdd.z);
+    uint firstIndex = (cell.x + cell.y * width);
+    atomicAdd(counts_SSBORed[firstIndex],toAdd.x);
+    atomicAdd(counts_SSBOGreen[firstIndex],toAdd.y);
+    atomicAdd(counts_SSBOBlue[firstIndex],toAdd.z);
 }
 
 uvec2 getCell(vec2 complex)
diff --git a/BuddhaTest/Shaders/BuddhaFragment.glsl b/BuddhaTest/Shaders/BuddhaFragment.glsl
index 1556065..24fb6a7 100644
--- a/BuddhaTest/Shaders/BuddhaFragment.glsl
+++ b/BuddhaTest/Shaders/BuddhaFragment.glsl
@@ -4,9 +4,17 @@ in vec2 uv;
 
 out vec3 color;
 
-layout(std430, binding=2) buffer renderedData
+layout(std430, binding=2) buffer renderedDataRed
 {
-        uint counts_SSBO[];
+        uint counts_SSBORed[];
+};
+layout(std430, binding=3) buffer renderedDataGreen
+{
+        uint counts_SSBOGreen[];
+};
+layout(std430, binding=4) buffer renderedDataBlue
+{
+        uint counts_SSBOBlue[];
 };
 
 uniform uint width;
@@ -16,8 +24,8 @@ uvec3 getColorAt(vec2 fragCoord)
 {
     uint xIndex = uint(max(0.0,(fragCoord.x+1.0)*0.5*width));
     uint yIndex = uint(max(0.0,abs(fragCoord.y)*height));
-    uint firstIndex = 3*(xIndex + yIndex * width);
-    return uvec3(counts_SSBO[firstIndex],counts_SSBO[firstIndex+1],counts_SSBO[firstIndex+2]);
+    uint firstIndex = (xIndex + yIndex * width);
+    return uvec3(counts_SSBORed[firstIndex],counts_SSBOGreen[firstIndex],counts_SSBOBlue[firstIndex]);
 }
 
 void main(){
-- 
cgit v1.2.3