diff --git a/emulators/mame/files/patch-src_emu_video_rgbutil.cpp b/emulators/mame/files/patch-src_emu_video_rgbutil.cpp
new file mode 100644
index 000000000000..46b30a1fba5d
--- /dev/null
+++ b/emulators/mame/files/patch-src_emu_video_rgbutil.cpp
@@ -0,0 +1,28 @@
+--- src/emu/video/rgbutil.cpp.orig	2026-01-08 15:01:14 UTC
++++ src/emu/video/rgbutil.cpp
+@@ -362,7 +362,7 @@ const VECS16 scale_table[256] =
+ 
+ } // anonymous namespace
+ 
+-u32 rgbaint_t::bilinear_filter(const u32 &rgb00, const u32 &rgb01, const u32 &rgb10, const u32 &rgb11, u8 u, u8 v)
++u32 rgbaint_t::bilinear_filter(u32 rgb00, u32 rgb01, u32 rgb10, u32 rgb11, u8 u, u8 v) noexcept
+ {
+ 	const VECS32 zero = vec_splat_s32(0);
+ 
+@@ -405,7 +405,7 @@ u32 rgbaint_t::bilinear_filter(const u32 &rgb00, const
+ 	return result;
+ }
+ 
+-void rgbaint_t::bilinear_filter_rgbaint(const u32 &rgb00, const u32 &rgb01, const u32 &rgb10, const u32 &rgb11, u8 u, u8 v)
++void rgbaint_t::bilinear_filter_rgbaint(u32 rgb00, u32 rgb01, u32 rgb10, u32 rgb11, u8 u, u8 v) noexcept
+ {
+ 	const VECS32 zero = vec_splat_s32(0);
+ 
+@@ -439,6 +439,7 @@ void rgbaint_t::bilinear_filter_rgbaint(const u32 &rgb
+ 	color11 = vec_sr(color11, vec_splat_u32(1));
+ 	color01 = VECS32(vec_max(VECS16(color01), VECS16(color11)));
+ 	color01 = vec_msum(VECS16(color01), scale_table[v], zero);
++	color01 = vec_sr(color01, vec_splat_u32(15));
+ 
+ #ifdef __LITTLE_ENDIAN__
+ 	vec_ste(vec_splat(color01, 3), 0, &m_a);