diff --git a/Makefile b/Makefile
index 6ed7bae8..66b37ce8 100644
--- a/Makefile
+++ b/Makefile
@@ -1,16 +1,20 @@
 CC=			gcc
 CFLAGS=		-g -Wall -O2 -Wc++-compat
 CPPFLAGS=	-DHAVE_KALLOC
-INCLUDES=	-I.
-OBJS=		kthread.o kalloc.o ksw2_extz2_sse.o ksw2_extd2_sse.o ksw2_exts2_sse.o ksw2_ll_sse.o \
-			misc.o bseq.o sketch.o sdust.o index.o chain.o align.o hit.o map.o format.o
+INCLUDES=
+OBJS=		kthread.o kalloc.o misc.o bseq.o sketch.o sdust.o index.o chain.o align.o hit.o map.o format.o ksw2_ll_sse.o
 PROG=		minimap2
 PROG_EXTRA=	sdust minimap2-lite
 LIBS=		-lm -lz -lpthread
 
+ifneq ($(cpu_dispatch),)
+	OBJS+=ksw2_extz2_sse41.o ksw2_extd2_sse41.o ksw2_exts2_sse41.o ksw2_extz2_sse2.o ksw2_extd2_sse2.o ksw2_exts2_sse2.o ksw2_dispatch.o
+else
 ifeq ($(sse2only),)
 	CFLAGS+=-msse4
 endif
+	OBJS+=ksw2_extz2_sse.o ksw2_extd2_sse.o ksw2_exts2_sse.o
+endif
 
 .SUFFIXES:.c .o
 
@@ -33,6 +37,27 @@ libminimap2.a:$(OBJS)
 sdust:sdust.c getopt.o kalloc.o kalloc.h kdq.h kvec.h kseq.h sdust.h
 		$(CC) -D_SDUST_MAIN $(CFLAGS) $< getopt.o kalloc.o -o $@ -lz
 
+ksw2_extz2_sse41.o:ksw2_extz2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -msse4 $< -o $@
+
+ksw2_extz2_sse2.o:ksw2_extz2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -mno-sse4 -msse2 $< -o $@
+
+ksw2_extd2_sse41.o:ksw2_extd2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -msse4 $< -o $@
+
+ksw2_extd2_sse2.o:ksw2_extd2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -mno-sse4 -msse2 $< -o $@
+
+ksw2_exts2_sse41.o:ksw2_exts2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -msse4 $< -o $@
+
+ksw2_exts2_sse2.o:ksw2_exts2_sse.c ksw2.h kalloc.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) -mno-sse4 -msse2 $< -o $@
+
+ksw2_dispatch.o:ksw2_dispatch.c ksw2.h
+		$(CC) -c $(CFLAGS) $(CPPFLAGS) -DKSW_CPU_DISPATCH $(INCLUDES) $< -o $@
+
 clean:
 		rm -fr gmon.out *.o a.out $(PROG) $(PROG_EXTRA) *~ *.a *.dSYM session*
 
diff --git a/ksw2.h b/ksw2.h
index 5e43970f..fa22fe6a 100644
--- a/ksw2.h
+++ b/ksw2.h
@@ -169,5 +169,4 @@ static inline int ksw_apply_zdrop(ksw_extz_t *ez, int is_rot, int32_t H, int a,
 	}
 	return 0;
 }
-
 #endif
diff --git a/ksw2_dispatch.c b/ksw2_dispatch.c
new file mode 100644
index 00000000..681460eb
--- /dev/null
+++ b/ksw2_dispatch.c
@@ -0,0 +1,43 @@
+#ifdef KSW_CPU_DISPATCH
+#include <stdlib.h>
+#include "ksw2.h"
+
+void ksw_extz2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez)
+{
+	extern void ksw_extz2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez);
+	extern void ksw_extz2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez);
+	if (__builtin_cpu_supports("sse4.1"))
+		ksw_extz2_sse41(km, qlen, query, tlen, target, m, mat, q, e, w, zdrop, flag, ez);
+	else if (__builtin_cpu_supports("sse2"))
+		ksw_extz2_sse2(km, qlen, query, tlen, target, m, mat, q, e, w, zdrop, flag, ez);
+	else abort();
+}
+
+void ksw_extd2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez)
+{
+	extern void ksw_extd2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez);
+	extern void ksw_extd2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez);
+	if (__builtin_cpu_supports("sse4.1"))
+		ksw_extd2_sse41(km, qlen, query, tlen, target, m, mat, q, e, q2, e2, w, zdrop, flag, ez);
+	else if (__builtin_cpu_supports("sse2"))
+		ksw_extd2_sse2(km, qlen, query, tlen, target, m, mat, q, e, q2, e2, w, zdrop, flag, ez);
+	else abort();
+}
+
+void ksw_exts2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez)
+{
+	extern void ksw_exts2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez);
+	extern void ksw_exts2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez);
+	if (__builtin_cpu_supports("sse4.1"))
+		ksw_exts2_sse41(km, qlen, query, tlen, target, m, mat, q, e, q2, noncan, zdrop, flag, ez);
+	else if (__builtin_cpu_supports("sse2"))
+		ksw_exts2_sse2(km, qlen, query, tlen, target, m, mat, q, e, q2, noncan, zdrop, flag, ez);
+	else abort();
+}
+#endif
diff --git a/ksw2_extd2_sse.c b/ksw2_extd2_sse.c
index 56cd8cd8..9a64e49e 100644
--- a/ksw2_extd2_sse.c
+++ b/ksw2_extd2_sse.c
@@ -10,8 +10,18 @@
 #include <smmintrin.h>
 #endif
 
+#ifdef KSW_CPU_DISPATCH
+#ifdef __SSE4_1__
+void ksw_extd2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez)
+#else
+void ksw_extd2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez)
+#endif
+#else
 void ksw_extd2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
 				   int8_t q, int8_t e, int8_t q2, int8_t e2, int w, int zdrop, int flag, ksw_extz_t *ez)
+#endif // ~KSW_CPU_DISPATCH
 {
 #define __dp_code_block1 \
 	z = _mm_load_si128(&s[t]); \
diff --git a/ksw2_exts2_sse.c b/ksw2_exts2_sse.c
index 6decd2de..7a64905b 100644
--- a/ksw2_exts2_sse.c
+++ b/ksw2_exts2_sse.c
@@ -10,8 +10,18 @@
 #include <smmintrin.h>
 #endif
 
+#ifdef KSW_CPU_DISPATCH
+#ifdef __SSE4_1__
+void ksw_exts2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez)
+#else
+void ksw_exts2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
+				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez)
+#endif
+#else
 void ksw_exts2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat,
 				   int8_t q, int8_t e, int8_t q2, int8_t noncan, int zdrop, int flag, ksw_extz_t *ez)
+#endif // ~KSW_CPU_DISPATCH
 {
 #define __dp_code_block1 \
 	z = _mm_load_si128(&s[t]); \
diff --git a/ksw2_extz2_sse.c b/ksw2_extz2_sse.c
index f21f1845..18a3d2b5 100644
--- a/ksw2_extz2_sse.c
+++ b/ksw2_extz2_sse.c
@@ -9,7 +9,15 @@
 #include <smmintrin.h>
 #endif
 
+#ifdef KSW_CPU_DISPATCH
+#ifdef __SSE4_1__
+void ksw_extz2_sse41(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez)
+#else
+void ksw_extz2_sse2(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez)
+#endif
+#else
 void ksw_extz2_sse(void *km, int qlen, const uint8_t *query, int tlen, const uint8_t *target, int8_t m, const int8_t *mat, int8_t q, int8_t e, int w, int zdrop, int flag, ksw_extz_t *ez)
+#endif // ~KSW_CPU_DISPATCH
 {
 #define __dp_code_block1 \
 	z = _mm_add_epi8(_mm_load_si128(&s[t]), qe2_); \
diff --git a/main.c b/main.c
index 720e7434..e838560d 100644
--- a/main.c
+++ b/main.c
@@ -6,7 +6,7 @@
 #include "mmpriv.h"
 #include "getopt.h"
 
-#define MM_VERSION "2.1-r335-dirty"
+#define MM_VERSION "2.1-r337-dirty"
 
 #ifdef __linux__
 #include <sys/resource.h>