197 lines
4.9 KiB
Markdown
197 lines
4.9 KiB
Markdown
# nuitka 1.2.1 的性能测试
|
||
大家好！ 这里是shenjack！
|
||
这篇文章就是 [B站](还没发出来呢) 专栏的原文！
|
||
|
||
# 综述
|
||
|
||
## 环境
|
||
|
||
- 硬件
|
||
- CPU: `AMD R5 5600X`
|
||
- 内存: `64G DDR4 3200 (32×2)`
|
||
|
||
|
||
- 软件
|
||
- 
|
||
- 
|
||
- 
|
||
- `gcc 12.2.0`
|
||
- `clang 14.0.6`
|
||
- `nuitka v1.2.1`
|
||
- `Python 3.8.10`
|
||
- [DR](https://github.com/shenjackyuanjie/Difficult-Rocket) commit `e6a2b7b36cc98e508efbc61c4646181f5817fc7c`
|
||
|
||
## 编译选项
|
||
|
||
- 核心编译选项
|
||
- 环境
|
||
- `--standalone`
|
||
- 将环境一并打包进完成的文件夹中
|
||
- ```
|
||
--standalone Enable standalone mode for output. This allows you to
|
||
transfer the created binary to other machines without
|
||
it using an existing Python installation. This also
|
||
means it will become big. It implies these option: "--
|
||
follow-imports" and "--python-flag=no_site". Defaults
|
||
to off.```
|
||
|
||
- 编译器
|
||
|
||
- `--clang`
|
||
- 强制使用 `clang` 编译器
|
||
- ```
|
||
--clang Enforce the use of clang. On Windows this requires a
|
||
working Visual Studio version to piggy back on.
|
||
Defaults to off.```
|
||
|
||
- `--mingw64` (常量)
|
||
- 强制使用 `mingw64` 编译器
|
||
- ```
|
||
--mingw64 Enforce the use of MinGW64 on Windows. Defaults to
|
||
off.```
|
||
|
||
- `--jobs=24` (常量)
|
||
- 最多同时运行的 C 编译器 `(clang|mingw64)` 数量
|
||
- ```
|
||
-j N, --jobs=N Specify the allowed number of parallel C compiler
|
||
jobs. Defaults to the system CPU count.```
|
||
|
||
- `--lto=` `yes|no|auto(None)`
|
||
- 是否启用 链接
|
||
- ```
|
||
--lto=choice Use link time optimizations (MSVC, gcc, clang).
|
||
Allowed values are "yes", "no", and "auto" (when it's
|
||
known to work). Defaults to "auto".```
|
||
|
||
- `--disable-ccache`
|
||
- 不使用 `ccache` 或者 `clcache` 的缓存
|
||
- ```
|
||
--disable-ccache Do not attempt to use ccache (gcc, clang, etc.) or
|
||
clcache (MSVC, clangcl).```
|
||
|
||
|
||
## 结论综述
|
||
|
||
补充：`lto=yes` 的情况下经常有报错
|
||
|
||
### 有 `Ccache` (全 hit 无实际编译)
|
||
|
||
`clang lto=no` 效率最高
|
||
|
||
`mingw lto=no` 效率低于 `lto=yes`，但是更加稳定（不会报错）
|
||
|
||
### 无 `Ccache` (测试编译器实际效率)
|
||
|
||
`clang lto=no` 效率最高
|
||
|
||
`mingw lto=no` 同样效率相对高
|
||
|
||
### 总结
|
||
|
||
尽量使用 `lto=no` 配合 `ccache` 以及 `clang`
|
||
|
||
|
||
# 原始数据
|
||
|
||
## 因为视频使用CPU编码，所以很多数据没法看到，这里使用的是后面再次测试/微信群内存档的数据
|
||
## 视频应该会再次测试，数据可能不同，请谅解(obs更新完没注意编码啊啊啊啊)
|
||
|
||
<table>
|
||
<tr>
|
||
<th>编译器</th> <th colspan="2">有ccache</th> <th colspan="2">无ccache</th>
|
||
</tr>
|
||
<tr>
|
||
<th>编译器</th> <th>lto=no</th> <th>lto=yes</th> <th>lto=no</th> <th>lto=yes</th>
|
||
</tr>
|
||
<tr>
|
||
<td>clang</td> <td>50s</td> <td>86s</td> <td>83s</td> <td>67s</td>
|
||
</tr>
|
||
<tr>
|
||
<td>gcc</td> <td>59s</td> <td>641s</td> <td>227s</td> <td>630s</td>
|
||
</tr>
|
||
</table>
|
||
|
||
|
||
- `clang`
|
||
|
||
- `with ccache`
|
||
|
||
- `lto=no`
|
||
|
||
- `视频内数据`
|
||
- `--clang`
|
||
- `--mingw64`
|
||
- `--lto=no`
|
||
- 耗时: 50000ms (50s)
|
||
- 
|
||
|
||
- `lto=yes`
|
||
|
||
- `后期测试数据`
|
||
- `--clang`
|
||
- `--mingw64`
|
||
- `--lto=yes`
|
||
- 耗时 86000ms (86s)
|
||
- 
|
||
|
||
- `no ccache`
|
||
|
||
- `lto=no`
|
||
|
||
- `视频内数据`
|
||
- `--clang`
|
||
- `--mingw64`
|
||
- `--lto=no`
|
||
- `--disable-ccache`
|
||
- 耗时 83000ms (83s)
|
||
- 
|
||
|
||
- `lto=yes`
|
||
|
||
- `后期测试数据`
|
||
- `--clang`
|
||
- `--mingw64`
|
||
- `--lto=yes`
|
||
- `--disable-ccache`
|
||
- 耗时 67000ms (67s)
|
||
- 
|
||
|
||
- `mingw64`
|
||
|
||
- `with ccache`
|
||
|
||
- `lto=no`
|
||
|
||
- `后期测试数据`
|
||
- `--mingw64`
|
||
- `--lto=no`
|
||
- 耗时 59000ms (59s)
|
||
- 
|
||
|
||
- `lto=yes`
|
||
|
||
- `后期测试数据`
|
||
- `--mingw64`
|
||
- `--lto=yes`
|
||
- 耗时 641000ms (641s) (10.68min)
|
||
- 
|
||
|
||
- `no ccache`
|
||
|
||
- `lto=no`
|
||
|
||
- `视频内数据`
|
||
- `--mingw64`
|
||
- `--lto=no`
|
||
- `--disable-ccache`
|
||
- 耗时 227000ms (227s) (3.78min)
|
||
- 
|
||
|
||
- `lto=yes`
|
||
|
||
- `后期测试数据`
|
||
- `--mingw64`
|
||
- `--lto=yes`
|
||
- `--disable-ccache`
|
||
- 耗时 630000ms (630s) (10.5min)
|
||
-  |