From 91df32232f917429c173d0d3d17a80d15dc3e484 Mon Sep 17 00:00:00 2001
From: Chuck Walbourn <chuck_walbourn@yahoo.com>
Date: Mon, 28 Nov 2022 00:03:22 -0800
Subject: [PATCH] Bias added for ARM version of 16bpp format conversions to
 match existing output

---
 DirectXTex/DirectXTexConvert.cpp | 20 +++++++++++++++++++-
 DirectXTex/DirectXTexP.h         |  1 +
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/DirectXTex/DirectXTexConvert.cpp b/DirectXTex/DirectXTexConvert.cpp
index f98fbc7..bbcc704 100644
--- a/DirectXTex/DirectXTexConvert.cpp
+++ b/DirectXTex/DirectXTexConvert.cpp
@@ -2068,7 +2068,11 @@ bool DirectX::Internal::StoreScanline(
             {
                 if (sPtr >= ePtr) break;
                 XMVECTOR v = XMVectorSwizzle<2, 1, 0, 3>(*sPtr++);
+#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || __arm__ || __aarch64__
+                v = XMVectorMultiplyAdd(v, s_Scale, g_XMOneHalf);
+#else
                 v = XMVectorMultiply(v, s_Scale);
+#endif
                 XMStoreU565(dPtr++, v);
             }
             return true;
@@ -2079,12 +2083,19 @@ bool DirectX::Internal::StoreScanline(
         if (size >= sizeof(XMU555))
         {
             static const XMVECTORF32 s_Scale = { { { 31.f, 31.f, 31.f, 1.f } } };
+#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || __arm__ || __aarch64__
+            static const XMVECTORF32 s_OneHalfXYZ = { { { 0.5f, 0.5f, 0.5f, 0.f } } };
+#endif
             XMU555 * __restrict dPtr = static_cast<XMU555*>(pDestination);
             for (size_t icount = 0; icount < (size - sizeof(XMU555) + 1); icount += sizeof(XMU555))
             {
                 if (sPtr >= ePtr) break;
                 XMVECTOR v = XMVectorSwizzle<2, 1, 0, 3>(*sPtr++);
+#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || __arm__ || __aarch64__
+                v = XMVectorMultiplyAdd(v, s_Scale, s_OneHalfXYZ);
+#else
                 v = XMVectorMultiply(v, s_Scale);
+#endif
                 XMStoreU555(dPtr, v);
                 dPtr->w = (XMVectorGetW(v) > threshold) ? 1u : 0u;
                 ++dPtr;
@@ -2360,7 +2371,11 @@ bool DirectX::Internal::StoreScanline(
             {
                 if (sPtr >= ePtr) break;
                 XMVECTOR v = XMVectorSwizzle<2, 1, 0, 3>(*sPtr++);
+#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || __arm__ || __aarch64__
+                v = XMVectorMultiplyAdd(v, s_Scale, g_XMOneHalf);
+#else
                 v = XMVectorMultiply(v, s_Scale);
+#endif
                 XMStoreUNibble4(dPtr++, v);
             }
             return true;
@@ -2436,8 +2451,11 @@ bool DirectX::Internal::StoreScanline(
             for (size_t icount = 0; icount < (size - sizeof(uint8_t) + 1); icount += sizeof(uint8_t))
             {
                 if (sPtr >= ePtr) break;
+#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || __arm__ || __aarch64__
+                const XMVECTOR v = XMVectorMultiplyAdd(*sPtr++, s_Scale, g_XMOneHalf);
+#else
                 const XMVECTOR v = XMVectorMultiply(*sPtr++, s_Scale);
-
+#endif
                 XMUNIBBLE4 nibble;
                 XMStoreUNibble4(&nibble, v);
                 *dPtr = static_cast<uint8_t>(nibble.v);
diff --git a/DirectXTex/DirectXTexP.h b/DirectXTex/DirectXTexP.h
index 38fdc44..9673713 100644
--- a/DirectXTex/DirectXTexP.h
+++ b/DirectXTex/DirectXTexP.h
@@ -70,6 +70,7 @@
 #pragma clang diagnostic ignored "-Wswitch-enum"
 #pragma clang diagnostic ignored "-Wtautological-type-limit-compare"
 #pragma clang diagnostic ignored "-Wunknown-pragmas"
+#pragma clang diagnostic ignored "-Wundef"
 #endif
 
 #ifdef _WIN32