From: Chris Duncan <chris@zoso.dev>
Date: Wed, 8 Jan 2025 21:35:44 +0000 (-0800)
Subject: Inline G mix 3 of round 0.
X-Git-Url: https://zoso.dev/?a=commitdiff_plain;h=8f51c09178e9412e86e5e31bc0d3b824ec2403eb;p=libnemo.git

Inline G mix 3 of round 0.
---

diff --git a/src/lib/nano-pow/shaders/gpu-compute.ts b/src/lib/nano-pow/shaders/gpu-compute.ts
index c9918f2..a83ef2e 100644
--- a/src/lib/nano-pow/shaders/gpu-compute.ts
+++ b/src/lib/nano-pow/shaders/gpu-compute.ts
@@ -461,7 +461,92 @@ fn main(
 
 
 
-	G(&v6, &v7, &v14, &v15, &v22, &v23, &v30, &v31, 0u, 0u, 0u, 0u);
+	/**
+	* r=0, i=3(x2), a=v[6-7], b=v[14-15], c=v[22-23], d=v[30-31]
+	*/
+
+	// a = a + b
+	o0 = v6 + v14;
+	o1 = v7 + v15;
+	if (v6 > 0xFFFFFFFFu - v14) {
+		o1 = o1 + 1u;
+	}
+	v6 = o0;
+	v7 = o1;
+
+	// // a = a + m[sigma[r][2*i+0]]
+	// // skip since adding 0u does nothing
+	// o0 = v6 + 0u;
+	// o1 = v7 + 0u;
+	// if (v6 > 0xFFFFFFFFu - 0u) {
+	// 	o1 = o1 + 1u;
+	// }
+	// v6 = o0;
+	// v7 = o1;
+
+	// d = rotr64(d ^ a, 32)
+	xor0 = v30 ^ v6;
+	xor1 = v31 ^ v7;
+	v30 = xor1;
+	v31 = xor0;
+
+	// c = c + d
+	o0 = v22 + v30;
+	o1 = v23 + v31;
+	if (v22 > 0xFFFFFFFFu - v30) {
+		o1 = o1 + 1u;
+	}
+	v22 = o0;
+	v23 = o1;
+
+	// b = rotr64(b ^ c, 24)
+	xor0 = v14 ^ v22;
+	xor1 = v15 ^ v23;
+	v14 = (xor0 >> 24u) ^ (xor1 << 8u);
+	v15 = (xor1 >> 24u) ^ (xor0 << 8u);
+
+	// a = a + b
+	o0 = v6 + v14;
+	o1 = v7 + v15;
+	if (v6 > 0xFFFFFFFFu - v14) {
+		o1 = o1 + 1u;
+	}
+	v6 = o0;
+	v7 = o1;
+
+	// // a = a + m[sigma[r][2*i+1]]
+	// // skip since adding 0u does nothing
+	// o0 = v6 + 0u;
+	// o1 = v7 + 0u;
+	// if (v6 > 0xFFFFFFFFu - 0u) {
+	// 	o1 = o1 + 1u;
+	// }
+	// v6 = o0;
+	// v7 = o1;
+
+	// d = rotr64(d ^ a, 16)
+	xor0 = v30 ^ v6;
+	xor1 = v31 ^ v7;
+	v30 = (xor0 >> 16u) ^ (xor1 << 16u);
+	v31 = (xor1 >> 16u) ^ (xor0 << 16u);
+
+	// c = c + d
+	o0 = v22 + v30;
+	o1 = v23 + v31;
+	if (v22 > 0xFFFFFFFFu - v30) {
+		o1 = o1 + 1u;
+	}
+	v22 = o0;
+	v23 = o1;
+
+	// b = rotr64(b ^ c, 63)
+	xor0 = v14 ^ v22;
+	xor1 = v15 ^ v23;
+	v14 = (xor1 >> 31u) ^ (xor0 << 1u);
+	v15 = (xor0 >> 31u) ^ (xor1 << 1u);
+
+
+
 	G(&v0, &v1, &v10, &v11, &v20, &v21, &v30, &v31, 0u, 0u, 0u, 0u);
 	G(&v2, &v3, &v12, &v13, &v22, &v23, &v24, &v25, 0u, 0u, 0u, 0u);
 	G(&v4, &v5, &v14, &v15, &v16, &v17, &v26, &v27, 0u, 0u, 0u, 0u);