From 3d8f26ee119059ddb05b53290a636760219ad09e Mon Sep 17 00:00:00 2001
From: Chris Duncan <chris@zoso.dev>
Date: Tue, 14 Jan 2025 14:44:32 -0800
Subject: [PATCH] Replace m[0-1] with nonce vector.

---
 src/shaders/compute.wgsl | 30 ++++++++++++++----------------
 1 file changed, 14 insertions(+), 16 deletions(-)

diff --git a/src/shaders/compute.wgsl b/src/shaders/compute.wgsl
index 4df2fb0..28c4a2a 100644
--- a/src/shaders/compute.wgsl
+++ b/src/shaders/compute.wgsl
@@ -35,8 +35,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	/**
 	* Initialize (nonce||blockhash) concatenation
 	*/
-	var m0: u32 = ubo.random.x ^ id.x;
-	var m1: u32 = ubo.random.y ^ id.y;
+	var nonce: vec2<u32> = vec2(ubo.random.x ^ id.x, ubo.random.y ^ id.y);
 	var m2: u32 = ubo.blockhash[0u].x;
 	var m3: u32 = ubo.blockhash[0u].y;
 	var m4: u32 = ubo.blockhash[0u].z;
@@ -94,7 +93,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v0 = v0 + v4 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + v4.x < v0.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v0 = v0 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + m0 < v0.x);
+	v0 = v0 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + nonce.x < v0.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v12 ^ v0;
@@ -644,7 +643,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = v1 + v6 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + v6.x < v1.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v1 = v1 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + m0 < v1.x);
+	v1 = v1 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + nonce.x < v1.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v12 ^ v1;
@@ -833,7 +832,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = v1 + v5 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + v5.x < v1.x);
 
 	// a = a + m[sigma[r][2*i+1]]
-	v1 = v1 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + m0 < v1.x);
+	v1 = v1 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + nonce.x < v1.x);
 
 	// d = rotr64(d ^ a, 16)
 	xor = v13 ^ v1;
@@ -1383,7 +1382,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v2 = v2 + v7 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + v7.x < v2.x);
 
 	// a = a + m[sigma[r][2*i+1]]
-	v2 = v2 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + m0 < v2.x);
+	v2 = v2 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + nonce.x < v2.x);
 
 	// d = rotr64(d ^ a, 16)
 	xor = v13 ^ v2;
@@ -1471,7 +1470,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v0 = v0 + v4 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + v4.x < v0.x);
 
 	// a = a + m[sigma[r][2*i+1]]
-	v0 = v0 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + m0 < v0.x);
+	v0 = v0 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + nonce.x < v0.x);
 
 	// d = rotr64(d ^ a, 16)
 	xor = v12 ^ v0;
@@ -1878,7 +1877,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v2 = v2 + v6 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + v6.x < v2.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v2 = v2 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + m0 < v2.x);
+	v2 = v2 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v2.x + nonce.x < v2.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v14 ^ v2;
@@ -2302,7 +2301,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v0 = v0 + v5 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + v5.x < v0.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v0 = v0 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + m0 < v0.x);
+	v0 = v0 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + nonce.x < v0.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v15 ^ v0;
@@ -2659,7 +2658,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v0 = v0 + v5 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + v5.x < v0.x);
 
 	// a = a + m[sigma[r][2*i+1]]
-	v0 = v0 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + m0 < v0.x);
+	v0 = v0 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + nonce.x < v0.x);
 
 	// d = rotr64(d ^ a, 16)
 	xor = v15 ^ v0;
@@ -2940,7 +2939,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = v3 + v7 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + v7.x < v3.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v3 = v3 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + m0 < v3.x);
+	v3 = v3 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + nonce.x < v3.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v15 ^ v3;
@@ -3465,7 +3464,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v3 = v3 + v4 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + v4.x < v3.x);
 
 	// a = a + m[sigma[r][2*i+1]]
-	v3 = v3 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + m0 < v3.x);
+	v3 = v3 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v3.x + nonce.x < v3.x);
 
 	// d = rotr64(d ^ a, 16)
 	xor = v14 ^ v3;
@@ -3494,7 +3493,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v0 = v0 + v4 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + v4.x < v0.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v0 = v0 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + m0 < v0.x);
+	v0 = v0 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v0.x + nonce.x < v0.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v12 ^ v0;
@@ -4044,7 +4043,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = v1 + v6 + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + v6.x < v1.x);
 
 	// a = a + m[sigma[r][2*i+0]]
-	v1 = v1 + vec2(m0, m1) + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + m0 < v1.x);
+	v1 = v1 + nonce + select(vec2<u32>(0u), vec2<u32>(0u, 1u), v1.x + nonce.x < v1.x);
 
 	// d = rotr64(d ^ a, 32)
 	xor = v12 ^ v1;
@@ -4171,8 +4170,7 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	*/
 	if ((BLAKE2B_IV32_1 ^ v0.y ^ v8.y) > threshold && atomicLoad(&work.found) == 0u) {
 		atomicStore(&work.found, 1u);
-		work.nonce.x = m0;
-		work.nonce.y = m1;
+		work.nonce = nonce;
 	}
 	return;
 }
-- 
2.34.1