vmovq_n_u8: mov data to neon register

.gitignore

@@ -50,3 +50,11 @@ modules.order
 Module.symvers
 Mkfile.old
 dkms.conf
+
+# Cmake files
+
+CMakeFiles/
+Makefile
+CMakeCache.txt
+cmake_install.cmake
+

CMakeLists.txt

@@ -0,0 +1,15 @@
+project(leetcode)
+
+set(CMAKE_VERBOSE_MAKEFILE on)
+
+#build vmovq_n_u8 with and without neon instructtion, like: movi v0.16b, #0x3
+add_compile_options(-Wall -march=armv8-a)
+
+add_executable(vmovq_n_u8_have_neon.out vmovq_n_u8.c)
+add_executable(vmovq_n_u8_no_neon.out vmovq_n_u8.c)
+
+target_compile_options(vmovq_n_u8_have_neon.out PRIVATE -O3)
+target_compile_options(vmovq_n_u8_no_neon.out PRIVATE -O0)
+
+add_executable(arm_neon_add.out arm_neon_add.c)
+cmake_minimum_required(VERSION 2.8.0)

arm_neon_add.c

@@ -0,0 +1,40 @@
+#include <stdio.h>
+#include "arm_neon.h"
+
+void pr(uint8_t *p, int n)
+{
+	int i;
+	printf("data: ");
+	for(i = 0; i < n; i++)
+	{
+		printf("%02d ", p[i]);
+	}
+	printf("\n");
+}
+
+void print_uint8x16(uint8x16_t *a, int n)
+{
+	uint8_t *p = (uint8_t *)a;
+	int i;
+	for(i = 0; i < n; i++)
+	{
+		printf("%02d ", *(p+1));
+	} 
+	printf("\n");
+}
+
+int main()
+{
+	//uint8_t origin[16] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
+	//uint8_t final[16] = {0};
+
+	uint8x16_t three = vmovq_n_u8(3);
+
+	//print_uint8x16(&three, 16);
+	//uint8x16_t data = vld1q_u8(origin);
+	//uint8x16_t result = vaddq_u8(data, three);
+	//vst1q_u8(final, result);
+
+	//pr(final, 16);
+
+}

vmovq_n_u8.c

@@ -0,0 +1,21 @@
+#include <stdio.h>
+#include "arm_neon.h"
+
+void print_uint8x16(uint8x16_t *a, int n)
+{
+	uint8_t *p = (uint8_t *)a;
+	int i;
+	for(i = 0; i < n; i++)
+	{
+		printf("%02d ", *(p+1));
+	} 
+	printf("\n");
+}
+
+int main()
+{
+
+	uint8x16_t three = vmovq_n_u8(3);
+
+	print_uint8x16(&three, 16);
+}