diff --git a/Makefile b/Makefile
index 87fe795aa..804c1bbec 100644
--- a/Makefile
+++ b/Makefile
@@ -359,6 +359,10 @@ ifdef LLAMA_SERVER_SSL
 	MK_LDFLAGS += -lssl -lcrypto
 endif
 
+ifndef GGML_NO_CPU_AARCH64
+	MK_CPPFLAGS += -DGGML_USE_CPU_AARCH64
+endif
+
 # warnings
 WARN_FLAGS = \
 	-Wall \
@@ -940,10 +944,6 @@ ggml/src/ggml-cuda/%.o: \
 	$(MCC) $(CXXFLAGS) $(MUSAFLAGS) -x musa -mtgpu -c -o $@ $<
 endif # GGML_MUSA
 
-ifndef GGML_NO_CPU_AARCH64
-	MK_CPPFLAGS += -DGGML_USE_CPU_AARCH64
-endif
-
 ifdef GGML_METAL
 	MK_CPPFLAGS += -DGGML_USE_METAL
 	MK_LDFLAGS  += -framework Foundation -framework Metal -framework MetalKit
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index 4a97bfc32..5cdf59f25 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -49,6 +49,14 @@
 
 #define UNUSED GGML_UNUSED
 
+#if defined(_MSC_VER)
+#define m512bh(p) p
+#define m512i(p) p
+#else
+#define m512bh(p) (__m512bh)(p)
+#define m512i(p) (__m512i)(p)
+#endif
+
 // precomputed f32 table for f16 (256 KB) (ggml-impl.h)
 float ggml_table_f32_f16[1 << 16];