From: Chris Duncan <chris@zoso.dev>
Date: Tue, 14 Jan 2025 18:31:48 +0000 (-0800)
Subject: Replace more 32-bit scalar rotations with vector rotations.
X-Git-Url: https://zoso.dev/?a=commitdiff_plain;h=b03be9a1ab8c9dfca564a3540f74ae13a16750e9;p=nano-pow.git

Replace more 32-bit scalar rotations with vector rotations.
---

diff --git a/src/shaders/compute.wgsl b/src/shaders/compute.wgsl
index d202d42..e025ff9 100644
--- a/src/shaders/compute.wgsl
+++ b/src/shaders/compute.wgsl
@@ -236,10 +236,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -964,10 +967,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -1698,10 +1704,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -2417,10 +2426,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -3157,10 +3169,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -3897,10 +3912,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -4628,10 +4646,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -5362,10 +5383,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -6093,10 +6117,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -6830,10 +6857,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -7564,10 +7594,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;
@@ -8292,10 +8325,13 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v26 ^ v2;
-	xor1 = v27 ^ v3;
-	v26 = xor1;
-	v27 = xor0;
+	v_23.x = v2;
+	v_23.y = v3;
+	v_2627.x = v26;
+	v_2627.y = v27;
+	v_2627 = v_2627 ^ v_23;
+	v26 = v_2627.y;
+	v27 = v_2627.x;
 
 	// c = c + d
 	v_1819.x = v18;