diff --git a/lib/include/srslte/phy/utils/mat.h b/lib/include/srslte/phy/utils/mat.h
index 48d3b776d..d960590c4 100644
--- a/lib/include/srslte/phy/utils/mat.h
+++ b/lib/include/srslte/phy/utils/mat.h
@@ -30,6 +30,7 @@
 #include "srslte/phy/utils/simd.h"
 #include "srslte/config.h"
 
+
 /*
  * Generic Macros
  */
@@ -65,6 +66,7 @@ SRSLTE_API float srslte_mat_2x2_cn(cf_t h00,
 
 
 #ifdef LV_HAVE_SSE
+#include <smmintrin.h>
 
 /* SSE implementation for complex reciprocal */
 SRSLTE_API __m128 srslte_mat_cf_recip_sse(__m128 a);
@@ -88,6 +90,8 @@ SRSLTE_API void srslte_mat_2x2_mmse_sse(__m128 y0, __m128 y1,
 
 #ifdef LV_HAVE_AVX
 
+#include <immintrin.h>
+
 /* AVX implementation for complex reciprocal */
 SRSLTE_API __m256 srslte_mat_cf_recip_avx(__m256 a);
 
diff --git a/lib/src/phy/utils/mat.c b/lib/src/phy/utils/mat.c
index 3c81d4a13..439daa2ce 100644
--- a/lib/src/phy/utils/mat.c
+++ b/lib/src/phy/utils/mat.c
@@ -25,7 +25,6 @@
  */
 
 #include <complex.h>
-#include <immintrin.h>
 #include <math.h>
 
 #include "srslte/phy/utils/mat.h"
@@ -116,6 +115,7 @@ inline float srslte_mat_2x2_cn(cf_t h00, cf_t h01, cf_t h10, cf_t h11) {
 }
 
 #ifdef LV_HAVE_SSE
+#include <smmintrin.h>
 
 /* SSE implementation for complex reciprocal */
 inline __m128 srslte_mat_cf_recip_sse(__m128 a) {
@@ -186,6 +186,7 @@ inline void srslte_mat_2x2_mmse_sse(__m128 y0, __m128 y1, __m128 h00, __m128 h01
 #endif /* LV_HAVE_SSE */
 
 #ifdef LV_HAVE_AVX
+#include <immintrin.h>
 
 /* AVX implementation for complex reciprocal */
 inline __m256 srslte_mat_cf_recip_avx(__m256 a) {
diff --git a/lib/src/phy/utils/vector.c b/lib/src/phy/utils/vector.c
index 742b9573f..d78f5d707 100644
--- a/lib/src/phy/utils/vector.c
+++ b/lib/src/phy/utils/vector.c
@@ -35,6 +35,16 @@
 #include "srslte/phy/utils/vector_simd.h"
 #include "srslte/phy/utils/bit.h"
 
+
+#ifdef LV_HAVE_SSE
+#include <smmintrin.h>
+#endif
+
+#ifdef LV_HAVE_AVX
+#include <immintrin.h>
+#endif
+
+
 #ifdef HAVE_VOLK
 #include "volk/volk.h"
 #endif