6、milk-v duo(CV1800B,C906内核)测试RVV 0.7.1矢量指令集

特别注意：以下所有操作必须再同一个终端命令行下进行。

在milk-v duo的SDK的目录下，预准备开发环境，参考1、milk-v duo(CV1800B,C906内核)搭建开发环境中的单步编译。

然后在milk-v duo的SDK目录外，新建一个文件夹，例如命名为rvv071，进入rvv071文件夹；
创建一个main.c文件，文件内容如下（摘抄于《RISC-V Vector Extension Intrinsic API Reference Manual》）：

#include <riscv_vector.h>
#include <stddef.h>
#include <stdio.h>
#include <math.h>

#define N 31

float input[N] = {
	-0.4325648115282207, -1.6655843782380970, 0.1253323064748307,
	0.2876764203585489, -1.1464713506814637, 1.1909154656429988,
	1.1891642016521031, -0.0376332765933176, 0.3272923614086541,
	0.1746391428209245, -0.1867085776814394, 0.7257905482933027,
	-0.5883165430141887, 2.1831858181971011, -0.1363958830865957,
	0.1139313135208096, 1.0667682113591888, 0.0592814605236053,
	-0.0956484054836690, -0.8323494636500225, 0.2944108163926404,
	-1.3361818579378040, 0.7143245518189522, 1.6235620644462707,
	-0.6917757017022868, 0.8579966728282626, 1.2540014216025324,
	-1.5937295764474768, -1.4409644319010200, 0.5711476236581780,
	-0.3998855777153632
};
float output_golden[N] = {
	1.7491401329284098, 0.1325982188803279, 0.3252281811989881,
	-0.7938091410349637, 0.3149236145048914, -0.5272704888029532,
	0.9322666565031119, 1.1646643544607362, -2.0456694357357357,
	-0.6443728590041911, 1.7410657940825480, 0.4867684246821860,
	1.0488288293660140, 1.4885752747099299, 1.2705014969484090,
	-1.8561241921210170, 2.1343209047321410, 1.4358467535865909,
	-0.9173023332875400, -1.1060770780029008, 0.8105708062681296,
	0.6985430696369063, -0.4015827425012831, 1.2687512030669628,
	-0.7836083053674872, 0.2132664971465569, 0.7878984786088954,
	0.8966819356782295, -0.1869172943544062, 1.0131816724341454,
	0.2484350696132857
};
float output[N] = {
	1.7491401329284098, 0.1325982188803279, 0.3252281811989881,
	-0.7938091410349637, 0.3149236145048914, -0.5272704888029532,
	0.9322666565031119, 1.1646643544607362, -2.0456694357357357,
	-0.6443728590041911, 1.7410657940825480, 0.4867684246821860,
	1.0488288293660140, 1.4885752747099299, 1.2705014969484090,
	-1.8561241921210170, 2.1343209047321410, 1.4358467535865909,
	-0.9173023332875400, -1.1060770780029008, 0.8105708062681296,
	0.6985430696369063, -0.4015827425012831, 1.2687512030669628,
	-0.7836083053674872, 0.2132664971465569, 0.7878984786088954,
	0.8966819356782295, -0.1869172943544062, 1.0131816724341454,
	0.2484350696132857
};
void saxpy_golden(size_t n, const float a, const float *x, float *y) {
	for (size_t i = 0; i < n; ++i) {
		y[i] = a * x[i] + y[i];
	}
}
void saxpy_vec(size_t n, const float a, const float *x, float *y) {
	size_t l;
	vfloat32m8_t vx, vy;
	for (; n > 0; n -= l) {
		l = vsetvl_e32m8(n);// 设置向量寄存器每次操作的元素个数n
		vx = vle32_v_f32m8(x, l);// 从数组x中加载l个元素到向量寄存器vx中
		x += l;
		vy = vle32_v_f32m8(y, l);// 从数组y中加载l个元素到向量寄存器vy中
		vy = vfmacc_vf_f32m8(vy, a, vx, l);// 向量寄存器a和向量寄存器vx中l个元素与vy对应相累加，结果为vy
		vse32_v_f32m8 (y, vy, l);// 将向量寄存器vy中的l个元素存到数组y中

		y += l;
	}
}
int fp_eq(float reference, float actual, float relErr)
{
	// if near zero, do absolute error instead.
	float absErr = relErr * ((fabsf(reference) > relErr) ?
	fabsf(reference) : relErr);
	return fabsf(actual - reference) < absErr;
}
int main() {
	saxpy_golden(N, 55.66, input, output_golden);
	saxpy_vec(N, 55.66, input, output);
	int pass = 1;
	for (int i = 0; i < N; i++) {
		if (!fp_eq(output_golden[i], output[i], 1e-6)) {
			printf("failed, %f=!%f\n", output_golden[i], output[i]);
			pass = 0;
		}
	}
	if (pass)
		printf("passed\n");
	return (pass == 0);
}

创建一个Makefile文件，文件内容如下（只是测试使用所以采用静态编译，如果项目使用可以将相应的库复制到板子上）：

TOOLPREFIX := riscv64-unknown-linux-musl-
# TOOLPREFIX := riscv64-unknown-linux-gnu-

CC = $(TOOLPREFIX)gcc

CFLAGS = -march=rv64gcv0p7_zfh_xtheadc -static
# CFLAGS = -march=rv64gcv0p7xthead -static

all:
	$(CC) $(CFLAGS) main.c -o test.app

然后编译，将生成的test.app文件复制到板子上，设置权限后执行，即可看到“passed”。

测试完成。

QQ：763314235，会在第一时间处理。