Halide入门教程11
// Halide tutorial lesson 11: Cross-compilation
// Halide教程第十一课:跨平台编译
// This lesson demonstrates how to use Halide as a cross-compiler to
// generate code for any platform from any platform.
// 本科展示了如何用Halide来充当跨平台编译器来生成不同平台的代码
// On linux, you can compile and run it like so:
// g++ lesson_11*.cpp -g -std=c++11 -I ../include -L ../bin -lHalide -lpthread -ldl -o lesson_11
// LD_LIBRARY_PATH=../bin ./lesson_11
#include "Halide.h"
#include <stdio.h>
using namespace Halide;
int main(int argc, char **argv) {
// We'll define the simple one-stage pipeline that we used in lesson 10.
Func brighter;
Var x, y;
// Declare the arguments.
Param<uint8_t> offset;
ImageParam input(type_of<uint8_t>(), 2);
std::vector<Argument> args(2);
args[0] = input;
args[1] = offset;
// Define the Func.
brighter(x, y) = input(x, y) + offset;
// Schedule it.
brighter.vectorize(x, 16).parallel(y);
// The following line is what we did in lesson 10. It compiles an
// object file suitable for the system that you're running this
// program on. For example, if you compile and run this file on
// 64-bit linux on an x86 cpu with sse4.1, then the generated code
// will be suitable for 64-bit linux on x86 with sse4.1.
// 下面这一行做的事情和第十课类似。它编译出一个适应与你当前工作平的object文件。
// 例如你当前工作在代sse4.1指令的x86 cpu的64位linux系统,它生成适应你平台的object文件
brighter.compile_to_file("lesson_11_host", args, "brighter");
// We can also compile object files suitable for other cpus and
// operating systems. You do this with an optional third argument
// to compile_to_file which specifies the target to compile for.
// 也可以编译处适合其他cpu和操作系统平台的object文件。在compile_to_file的target变量中制定
// 特定的编译目标平台即可
// Let's use this to compile a 32-bit arm android version of this code:
// 如下是一个制定安卓os和arm cpu的例子
Target target;
target.os = Target::Android; // The operating system
target.arch = Target::ARM; // The CPU architecture
target.bits = 32; // The bit-width of the architecture
std::vector<Target::Feature> arm_features; // A list of features to set
target.set_features(arm_features);
// We then pass the target as the last argument to compile_to_file.
// 只需在compile_to_file的最后一个target变量中制定编译的目标平台即可
brighter.compile_to_file("lesson_11_arm_32_android", args, "brighter", target);
// And now a Windows object file for 64-bit x86 with AVX and SSE 4.1:
// windows也类似
target.os = Target::Windows;
target.arch = Target::X86;
target.bits = 64;
std::vector<Target::Feature> x86_features;
x86_features.push_back(Target::AVX);
x86_features.push_back(Target::SSE41);
target.set_features(x86_features);
brighter.compile_to_file("lesson_11_x86_64_windows", args, "brighter", target);
// And finally an iOS mach-o object file for one of Apple's 32-bit
// ARM processors - the A6. It's used in the iPhone 5. The A6 uses
// a slightly modified ARM architecture called ARMv7s. We specify
// this using the target features field. Support for Apple's
// 64-bit ARM processors is very new in llvm, and still somewhat
// flaky.
// arm/ios也类似
target.os = Target::IOS;
target.arch = Target::ARM;
target.bits = 32;
std::vector<Target::Feature> armv7s_features;
armv7s_features.push_back(Target::ARMv7s);
target.set_features(armv7s_features);
brighter.compile_to_file("lesson_11_arm_32_ios", args, "brighter", target);
// Now let's check these files are what they claim, by examining
// their first few bytes.
// 下面是通过代码验证编译粗来的object文件是否符合对应文件格式的要求
// 32-arm android object files start with the magic bytes:
uint8_t arm_32_android_magic[] = {0x7f, 'E', 'L', 'F', // ELF format
1, // 32-bit
1, // 2's complement little-endian
1}; // Current version of elf
FILE *f = fopen("lesson_11_arm_32_android.o", "rb");
uint8_t header[32];
if (!f || fread(header, 32, 1, f) != 1) {
printf("Object file not generated\n");
return -1;
}
fclose(f);
if (memcmp(header, arm_32_android_magic, sizeof(arm_32_android_magic))) {
printf("Unexpected header bytes in 32-bit arm object file.\n");
return -1;
}
// 64-bit windows object files start with the magic 16-bit value 0x8664
// (presumably referring to x86-64)
uint8_t win_64_magic[] = {0x64, 0x86};
f = fopen("lesson_11_x86_64_windows.obj", "rb");
if (!f || fread(header, 32, 1, f) != 1) {
printf("Object file not generated\n");
return -1;
}
fclose(f);
if (memcmp(header, win_64_magic, sizeof(win_64_magic))) {
printf("Unexpected header bytes in 64-bit windows object file.\n");
return -1;
}
// 32-bit arm iOS mach-o files start with the following magic bytes:
uint32_t arm_32_ios_magic[] = {0xfeedface, // Mach-o magic bytes
12, // CPU type is ARM
11, // CPU subtype is ARMv7s
1}; // It's a relocatable object file.
f = fopen("lesson_11_arm_32_ios.o", "rb");
if (!f || fread(header, 32, 1, f) != 1) {
printf("Object file not generated\n");
return -1;
}
fclose(f);
if (memcmp(header, arm_32_ios_magic, sizeof(arm_32_ios_magic))) {
printf("Unexpected header bytes in 32-bit arm ios object file.\n");
return -1;
}
// It looks like the object files we produced are plausible for
// those targets. We'll count that as a success for the purposes
// of this tutorial. For a real application you'd then need to
// figure out how to integrate Halide into your cross-compilation
// toolchain. There are several small examples of this in the
// Halide repository under the apps folder. See HelloAndroid and
// HelloiOS here:
// https://github.com/halide/Halide/tree/master/apps/
// 经过上面的验证,halide针对那些目标平台生成的目标文件貌似可信。但是对于一个实际的引用,还必须指出如何
// 将Halide集成到跨平台的工具链道中。在Halide的github的app中有对应的例子,具体情况参见链接
printf("Success!\n");
return 0;
}
在终端中编译和运行代码:
$ g++ lesson_11*.cpp -g -std=c++11 -I ../include -L ../bin -lHalide -lpthread -ldl -o lesson_11
$ ./lesson_11
结果
补充说明:(更具体的情况点链接,去官网文档上看吧,比这里详细)
1.compile_to_file函数
EXPORT void Halide::Func::compile_to_file(
const std::string & filename_prefix, // 文件名前缀
const std::vector< Argument > & args, // 参数列表
const std::string & fn_name = "", // 函数名
const Target & target = get_target_from_environment() // 目标平台
)
2.Target结构
OS | Arch |
---|---|
OSUnknown | ArchUnknown |
Linux | X86 |
Windows | ARM |
OSX | MIPS |
Android | Hexagon |
IOS | POWERPC |
QuRT | |
NoOS |
内容太多,不在此一一列举,需要什么点链接上去查即可
http://halide-lang.org/docs/struct_halide_1_1_target.html#details