diff --git a/src/r_draw.cpp b/src/r_draw.cpp
index e8cfde488..23835fe4f 100644
--- a/src/r_draw.cpp
+++ b/src/r_draw.cpp
@@ -546,12 +546,24 @@ void R_InitViewBuffer(INT32 width, INT32 height)
 
 	if (temp_dc.buf)
 	{
+#if defined(__SSE__)
+		aligned_free(temp_dc.buf);
+#else
 		Z_Free(temp_dc.buf);
+#endif
 	}
 
 	memset(&temp_dc, 0, sizeof(temp_dc));
 
-	temp_dc.buf = static_cast<UINT8*>(Z_Calloc(bufsize, PU_STATIC, NULL));
+#if defined(__SSE__)
+	while (bufsize & 15)
+		bufsize++;
+	temp_dc.buf = static_cast<UINT8*>(aligned_alloc(16, bufsize));
+#else
+	temp_dc.buf = static_cast<UINT8*>(Z_Malloc(bufsize, PU_STATIC, NULL));
+#endif
+
+	memset(temp_dc.buf, 0, bufsize);
 
 	linesize     = vid.width;      // killough 11/98
 	renderscreen = vid.screens[0]; // haleyjd 07/02/14
diff --git a/src/v_video.c b/src/v_video.c
index 34a88cecf..ccc6fcdac 100644
--- a/src/v_video.c
+++ b/src/v_video.c
@@ -3873,12 +3873,19 @@ UINT8 GetColorLUTDirect(colorlookup_t *lut, UINT8 r, UINT8 g, UINT8 b)
 void V_Init(void)
 {
 	INT32 i;
-	const INT32 screensize = vid.rowbytes * vid.height;
+	INT32 screensize = vid.rowbytes * vid.height;
 
 	for (i = 0; i < NUMSCREENS; i++)
 	{
 		if (vid.screens[i])
+		{
+#if defined(__SSE__)
+			aligned_free(vid.screens[i]);
+#else
 			free(vid.screens[i]);
+#endif
+		}
+
 		vid.screens[i] = NULL;
 	}
 
@@ -3887,7 +3894,15 @@ void V_Init(void)
 	{
 		for (i = 0; i < NUMSCREENS; i++)
 		{
+			// we need to allocate these relative to their cpu restrictions to not trigger segfaults
+			// TODO: add support for sve and neon
+#if defined(__SSE__)
+			while (screensize & 15)
+				screensize++;
+			vid.screens[i] = aligned_alloc(16, screensize);
+#else
 			vid.screens[i] = malloc(screensize);
+#endif
 			memset(vid.screens[i], 0, screensize);
 		}
 	}