基于cython的即时编译器cyjit,欢迎大家提建议
我正在写一个使用cython code作为后端的即时编译器名为cyjit,将python code 转换为cython code再编译为c extension导入.设计上主要参考numba.jit的思路,使用decorate来指定要编译的function,例如: from cyjit import jit @jit('int(int,int)') def add(a,b): return a+b add(1,2)#compiled @jit('int(int,int)', locals=''' int c ''') def add1(a,b): c=add(a,b)# fast invoked return c add1(1,2) 目前还不支持类型推导,需要手动使用c的语法对局部变量进行定义。 编译过程是在jit函数中完成的,后续计划将编译过程移到函数运行时完成,实现重载。 目前支持编译cache,第一次运行需要编译,时间稍慢,再次运行直接导入编译好的extension,速度就很快了。 欢迎大家fork,pull,提建议。 https://github.com/liuzhenhai/cyjit -- https://mail.python.org/mailman/listinfo/python-list
Re: 基于cython的即时编译器cyjit,欢迎大家提建议
On 11/06/2014 10:37, 1989lzhh wrote: 我正在写一个使用cython code作为后端的即时编译器名为cyjit,将python code 转换为cython code再编译为c extension导入.设计上主要参考numba.jit的思路, 使用decorate来指定要编译的function,例如: from cyjit import jit @jit('int(int,int)') def add(a,b): return a+b add(1,2)#compiled @jit('int(int,int)', locals=''' int c ''') def add1(a,b): c=add(a,b)# fast invoked return c add1(1,2) 目前还不支持类型推导,需要手动使用c的语法对局部变量进行定义。 编译过程是在jit函数中完成的,后续计划将编译过程移到函数运行时完成,实现 重载。 目前支持编译cache,第一次运行需要编译,时间稍慢,再次运行直接导入编译好 的extension,速度就很快了。 欢迎大家fork,pull,提建议。 https://github.com/liuzhenhai/cyjit You might say that but I couldn't possibly comment. -- https://mail.python.org/mailman/listinfo/python-list
Re: 基于cython的即时编译器cyjit,欢迎大家提建议
You might say that but I couldn't possibly comment. You could run the message through Google Translate. It's not publication quality translation, but serves the needs in this instance. (Gmail offers to translate the OP's message for me.) Here's what GT produced (successfully translates the Chinese, but destroys the code structure in the process - what's wrong with those people at Google? wink): I'm writing a cython code using the compiler as a backend instant named cyjit, the python code Convert cython code is then compiled c extension import. Designed primarily reference numba. jit ideas, Use decorate to specify compile function, for example: from cyjit import jit @ Jit ('int (int, int)') def add (a, b): return a + b add (1,2) # compiled @ Jit ('int (int, int)', locals ='' ' int c '' ') def add1 (a, b): c = add (a, b) # fast invoked return c add1 (1,2) Currently does not support the type of derivation, C syntax to use local variables defined manually. Jit compilation process is done in the function of Follow-up plans to move to complete the compilation process runtime functions to achieve overloading. Currently supports compilation cache, you need to compile the first run, slower time, Run again compiled directly into the extension, the speed very quickly. Welcome to fork, pull, and suggestions. https://github.com/liuzhenhai/ cyjit The concept looks like of interesting. Skip -- https://mail.python.org/mailman/listinfo/python-list
Re: 基于cython的即时编译器cyjit,欢迎大家提建议
sorry,wrong version post 发自我的 iPhone 在 Jun 12, 2014,0:16,mm0fmf n...@mailinator.com 写道: On 11/06/2014 10:37, 1989lzhh wrote: 我正在写一个使用cython code作为后端的即时编译器名为cyjit,将python code 转换为cython code再编译为c extension导入.设计上主要参考numba.jit的思路, 使用decorate来指定要编译的function,例如: from cyjit import jit @jit('int(int,int)') def add(a,b): return a+b add(1,2)#compiled @jit('int(int,int)', locals=''' int c ''') def add1(a,b): c=add(a,b)# fast invoked return c add1(1,2) 目前还不支持类型推导,需要手动使用c的语法对局部变量进行定义。 编译过程是在jit函数中完成的,后续计划将编译过程移到函数运行时完成,实现 重载。 目前支持编译cache,第一次运行需要编译,时间稍慢,再次运行直接导入编译好 的extension,速度就很快了。 欢迎大家fork,pull,提建议。 https://github.com/liuzhenhai/cyjit You might say that but I couldn't possibly comment. -- https://mail.python.org/mailman/listinfo/python-list -- https://mail.python.org/mailman/listinfo/python-list
Re: 基于cython的即时编译器cyjit,欢迎大家提建议
在 Jun 12, 2014,1:16,Skip Montanaro s...@pobox.com 写道: You might say that but I couldn't possibly comment. You could run the message through Google Translate. It's not publication quality translation, but serves the needs in this instance. (Gmail offers to translate the OP's message for me.) Here's what GT produced (successfully translates the Chinese, but destroys the code structure in the process - what's wrong with those people at Google? wink): Thanks skip, I post the email into wrong mail list, I will rewrite it into English. :) I'm writing a cython code using the compiler as a backend instant named cyjit, the python code Convert cython code is then compiled c extension import. Designed primarily reference numba. jit ideas, Use decorate to specify compile function, for example: from cyjit import jit @ Jit ('int (int, int)') def add (a, b): return a + b add (1,2) # compiled @ Jit ('int (int, int)', locals ='' ' int c '' ') def add1 (a, b): c = add (a, b) # fast invoked return c add1 (1,2) Currently does not support the type of derivation, C syntax to use local variables defined manually. Jit compilation process is done in the function of Follow-up plans to move to complete the compilation process runtime functions to achieve overloading. Currently supports compilation cache, you need to compile the first run, slower time, Run again compiled directly into the extension, the speed very quickly. Welcome to fork, pull, and suggestions. https://github.com/liuzhenhai/ cyjit The translation's quality is quite good, I will edit it and post here again. Thanks The concept looks like of interesting. Skip -- https://mail.python.org/mailman/listinfo/python-list -- https://mail.python.org/mailman/listinfo/python-list