From d60609cc9f7bf71fdece096f1db2a36d30cba732 Mon Sep 17 00:00:00 2001
From: Chris Duncan <chris@zoso.dev>
Date: Tue, 14 Jan 2025 08:24:37 -0800
Subject: [PATCH] Replace more scalar addition with vector addition.

---
 src/shaders/compute.wgsl | 336 +++++++++++++++++++++++++--------------
 1 file changed, 216 insertions(+), 120 deletions(-)

diff --git a/src/shaders/compute.wgsl b/src/shaders/compute.wgsl
index 53610b6..3162f0a 100644
--- a/src/shaders/compute.wgsl
+++ b/src/shaders/compute.wgsl
@@ -496,11 +496,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -532,11 +536,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -1201,11 +1209,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+0]]
 	o0 = v0 + m2;
@@ -1234,11 +1246,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -1891,11 +1907,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -1927,11 +1947,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -2584,11 +2608,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+0]]
 	o0 = v0 + m4;
@@ -2617,11 +2645,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -3274,11 +3306,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -3310,11 +3346,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+1]]
 	o0 = v0 + m2;
@@ -3967,11 +4007,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+0]]
 	o0 = v0 + m8;
@@ -4000,11 +4044,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -4663,11 +4711,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+0]]
 	o0 = v0 + m0;
@@ -4696,11 +4748,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -5356,11 +5412,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -5392,11 +5452,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+1]]
 	o0 = v0 + m0;
@@ -6049,11 +6113,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -6085,11 +6153,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+1]]
 	o0 = v0 + m4;
@@ -6739,11 +6811,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -6775,11 +6851,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -7426,11 +7506,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+0]]
 	// // skip since adding 0u does nothing
@@ -7462,11 +7546,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
@@ -8131,11 +8219,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// a = a + m[sigma[r][2*i+0]]
 	o0 = v0 + m2;
@@ -8164,11 +8256,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v11 = (xor1 >> 24u) ^ (xor0 << 8u);
 
 	// a = a + b
-	o0 = v0 + v10;
-	o1 = v1 + v11;
-	o1 = o1 + select(0u, 1u, o0 < v0);
-	v0 = o0;
-	v1 = o1;
+	v_01.x = v0;
+	v_01.y = v1;
+	v_1011.x = v10;
+	v_1011.y = v11;
+	v_01 = v_01 + v_1011 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_01.x + v_1011.x < v_01.x);
+	v0 = v_01.x;
+	v1 = v_01.y;
+	v10 = v_1011.x;
+	v11 = v_1011.y;
 
 	// // a = a + m[sigma[r][2*i+1]]
 	// // skip since adding 0u does nothing
-- 
2.34.1