diff --git a/Makefile b/Makefile
index 263976b..4b86020 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 CC=			gcc
-CFLAGS=		-g -Wall -O2 -Wc++-compat -march=native
+CFLAGS=		-g -Wall -O2 -Wc++-compat
 CPPFLAGS=	-DHAVE_KALLOC
 INCLUDES=	-I.
 OBJS=		kthread.o kalloc.o ksw2_extz2_sse.o ksw2_extd2_sse.o misc.o bseq.o \
@@ -8,6 +8,10 @@ PROG=		minimap2
 PROG_EXTRA=	sdust
 LIBS=		-lm -lz -lpthread
 
+ifeq ($(sse2only),)
+	CFLAGS+=-msse4
+endif
+
 .SUFFIXES:.c .o
 
 .c.o:
diff --git a/README.md b/README.md
index 9e5dd20..982a4b9 100644
--- a/README.md
+++ b/README.md
@@ -31,6 +31,15 @@ and [longISLND][longislnd]), better chaining and the ability to produce CIGAR
 with fast extension alignment (see also [libgaba][gaba] and [ksw2][ksw2]) and
 piece-wise affine gap cost.
 
+## Installation
+
+For modern x86-64 CPUs, just type `make` in the source code directory. This
+will compile a binary `minimap2` which you can copy to your desired location.
+If you see compilation errors, try `make sse2only=1` to disable SSE4. Minimap2
+will run a little slower. At present, minimap2 does not work with non-x86 CPUs
+or ancient CPUs that do not support SSE2. SSE2 is critical to the performance
+of minimap2.
+
 ## Limitations
 
 * At the alignment phase, minimap2 performs global alignments between minimizer