From: Chris Duncan <chris@zoso.dev>
Date: Tue, 14 Jan 2025 17:50:15 +0000 (-0800)
Subject: Replace some 32-bit scalar rotations with vector rotations.
X-Git-Tag: v2.0.0~125
X-Git-Url: https://zoso.dev/?a=commitdiff_plain;h=e658330eed9429b29ec4984c587d36577b2af2a7;p=nano-pow.git

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

diff --git a/src/shaders/compute.wgsl b/src/shaders/compute.wgsl
index 48cfd75..2349da9 100644
--- a/src/shaders/compute.wgsl
+++ b/src/shaders/compute.wgsl
@@ -145,10 +145,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -878,10 +877,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -1608,10 +1606,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -2329,10 +2326,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -3059,10 +3055,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -3804,10 +3799,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -4537,10 +4531,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -5267,10 +5260,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -5997,10 +5989,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -6727,10 +6718,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -7463,10 +7453,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	v1 = o1;
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;
@@ -8196,10 +8185,9 @@ fn main(@builtin(global_invocation_id) id: vec3<u32>) {
 	// // skip since adding 0u does nothing
 
 	// d = rotr64(d ^ a, 32)
-	xor0 = v24 ^ v0;
-	xor1 = v25 ^ v1;
-	v24 = xor1;
-	v25 = xor0;
+	v_2425 = v_2425 ^ v_01;
+	v24 = v_2425.y;
+	v25 = v_2425.x;
 
 	// c = c + d
 	v_1617.x = v16;