From: Chris Duncan <chris@zoso.dev>
Date: Tue, 14 Jan 2025 16:33:13 +0000 (-0800)
Subject: Replace more scalar addition with vector addition.
X-Git-Tag: v2.0.0~131
X-Git-Url: https://zoso.dev/?a=commitdiff_plain;h=ee940cdb9d1f34367efb800654862fbef1100402;p=nano-pow.git

Replace more scalar addition with vector addition.
---

diff --git a/src/shaders/compute.wgsl b/src/shaders/compute.wgsl
index d3b1628..c8b6f83 100644
--- a/src/shaders/compute.wgsl
+++ b/src/shaders/compute.wgsl
@@ -715,11 +715,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -755,11 +759,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -1451,11 +1459,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -1491,11 +1503,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -2187,11 +2203,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -2224,11 +2244,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -2917,11 +2941,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -2954,11 +2982,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -3650,11 +3682,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -3690,11 +3726,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -4383,11 +4423,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -4423,11 +4467,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -5116,11 +5164,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -5153,11 +5205,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -5849,11 +5905,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -5889,11 +5949,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -6582,11 +6646,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -6619,11 +6687,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -7315,11 +7387,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -7355,11 +7431,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -8045,11 +8125,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -8085,11 +8169,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;
@@ -8781,11 +8869,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = xor0;
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 24)
 	xor0 = v14 ^ v16;
@@ -8821,11 +8913,15 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v27 = (xor1 >> 16u) ^ (xor0 << 16u);
 
 	// c = c + d
-	o0 = v16 + v26;
-	o1 = v17 + v27;
-	o1 = o1 + select(0u, 1u, o0 < v16);
-	v16 = o0;
-	v17 = o1;
+	v_1617.x = v16;
+	v_1617.y = v17;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_1617 = v_1617 + v_2627 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v_1617.x + v_2627.x < v_1617.x);
+	v16 = v_1617.x;
+	v17 = v_1617.y;
+	v26 = v_2627.x;
+	v27 = v_2627.y;
 
 	// b = rotr64(b ^ c, 63)
 	xor0 = v14 ^ v16;