Faster CPython

Contents:

FAT Python

Three-year-old Cambodian boy Oeun Sambat hugs his best friend, a four-metre long female python named Chamreun

Intro

The FAT Python project was started by Victor Stinner in October 2015 to try to solve issues of previous attempts of “static optimizers” for Python. The main feature are efficient guards using versionned dictionaries to check if something was modified. Guards are used to decide if the specialized bytecode of a function can be used or not.

Python FAT is expected to be FAT... maybe FAST if we are lucky. FAT because it will use two versions of some functions where one version is specialised to specific argument types, a specific environment, optimized when builtins are not mocked, etc.

See the fatoptimizer documentation which is the main part of FAT Python.

The FAT Python project is made of multiple parts:

Announcements and status reports:

Getting started

Compile Python 3.6 patched with PEP 509, PEP 510 and PEP 511:

hg clone http://hg.python.org/sandbox/fatpython
cd fatpython
./configure --with-pydebug CFLAGS="-O0" && make

Install fat:

git clone https://github.com/haypo/fat
cd fat
../python setup.py build
cp -v build/lib*/fat.*so ../Lib
cd ..

Install fatoptimizer:

git clone https://github.com/haypo/fatoptimizer
(cd Lib; ln -s ../fatoptimizer/fatoptimizer .)

fatoptimizer is registed by the site module if -X fat command line option is used. Extract of Lib/site.py:

if 'fat' in sys._xoptions:
    import fatoptimizer
    fatoptimizer._register()

Check that fatoptimizer is registered with:

$ ./python -X fat -c 'import sys; print(sys.implementation.optim_tag)'
fat-opt

You must get fat-opt (and not opt).

How can you contribute?

The fatoptimizer project needs the most love. Currently, the optimizer is not really smart. There is a long TODO list. Pick a simple optimization, try to implement it, send a pull request on GitHub. At least, any kind of feedback is useful ;-)

If you know the C API of Python, you may also review the implementation of the PEPs:

But these PEPs are still work-in-progress, so the implementation can still change.

Play with FAT Python

See Getting started to compile FAT Python.

Disable peephole optimizer

The -o noopt command line option disables the Python peephole optimizer:

$ ./python -o noopt -c 'import dis; dis.dis(compile("1+1", "test", "exec"))'
  1           0 LOAD_CONST               0 (1)
              3 LOAD_CONST               0 (1)
              6 BINARY_ADD
              7 POP_TOP
              8 LOAD_CONST               1 (None)
             11 RETURN_VALUE

Specialized code calling builtin function

Test fatoptimizer on builtin function:

$ ./python -X fat
>>> def func(): return len("abc")
...

>>> import dis
>>> dis.dis(func)
  1           0 LOAD_GLOBAL              0 (len)
              3 LOAD_CONST               1 ('abc')
              6 CALL_FUNCTION            1 (1 positional, 0 keyword pair)
              9 RETURN_VALUE

>>> import fat
>>> fat.get_specialized(func)
[(<code object func at 0x7f9d3155b1e0, file "<stdin>", line 1>,
[<fat.GuardBuiltins object at 0x7f9d39191198>])]

>>> dis.dis(fat.get_specialized(func)[0][0])
  1           0 LOAD_CONST               1 (3)
              3 RETURN_VALUE

The specialized code is removed when the function is called if the builtin function is replaced (here by declaring a len() function in the global namespace):

>>> len=lambda obj: "mock"
>>> func()
'mock'
>>> fat.func_get_specialized(func)
[]

Microbenchmark

Run a microbenchmark on specialized code:

$ ./python -m timeit -s 'def f(): return len("abc")' 'f()'
10000000 loops, best of 3: 0.122 usec per loop

$ ./python -X fat -m timeit -s 'def f(): return len("abc")' 'f()'
10000000 loops, best of 3: 0.0932 usec per loop

Python must be optimized to run a benchmark: use ./configure && make clean && make if you previsouly compiled it in debug mode.

You should compare specialized code to an unpatched Python 3.6 to run a fair benchmark (to also measure the overhead of PEP 509, 510 and 511 patches).

Run optimized code without registering fatoptimizer

You have to compile optimized .pyc files:

# the optimizer is slow, so add -v to enable fatoptimizer logs for more fun
./python -X fat -v -m compileall

# why does compileall not compile encodings/*.py?
./python -X fat -m py_compile Lib/encodings/{__init__,aliases,latin_1,utf_8}.py

Finally, enjoy optimized code with no registered optimized:

$ ./python -o fat-opt -c 'import sys; print(sys.implementation.optim_tag, sys.get_code_transformers())'
fat-opt []

Remember that you cannot import .py files in this case, only .pyc:

$ echo 'print("Hello World!")' > hello.py
$ ENV/bin/python -o fat-opt -c 'import hello'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ImportError: missing AST transformers for 'hello.py': optim_tag='fat-opt', transformers tag='noopt'

Origins of FAT Python

Everything in Python is mutable

Problem

开发者之所以喜欢用 Python,很大一部分原因是因为 Python 很灵活。 比如在单元测试里面,经常会用 unittest.mock ,这样可以覆盖一些内置函数,覆盖类的方法,各种灵活的用法。

Developers like Python because it’s possible to modify (almost) everything. This feature is heavily used in unit tests with unittest.mock which can override builtin function, override class methods, modify “constants, etc.

Most optimization rely on assumptions. For example, inlining rely on the fact that the inlined function is not modified. Implement optimization in respect of the Python semantics require to implement various assumptions.

Builtin functions

Python 提供的很多内置函数。 每一个用 Python 写的代码里面都会用到这些内置函数,但是通常却没有意识到这些函数已经被覆盖掉了。 实际上,这种情况非常容易发生.

Python provides a lot of builtins functions. All Python applications rely on them, and usually don’t expect that these functions are overriden. In practice, it is very easy to override them.

len() 函数来举个例子:

Example overriden the builtin len() function:

import builtins

def func(obj):
    print("length: %s" % len(obj))

func("abc")
builtins.len = lambda obj: "mock!"
func("abc")

Output:

length: 3
length: mock!

从上面的代码里面我们可以看到 len() 函数被加载到了 func 里面,按照 LEGB` 的原则, 会先在 ``blobals 的命名空间中查找 len, 失败后会尝试在 builtins 里面继续查找。

Technically, the len() function is loaded in func() with the LOAD_GLOBAL instruction which first tries to lookup in frame globals namespace, and then lookup in the frame builtins namespace.

下面这个例子里面尝试用 global 命名空间里面的 len 函数来覆盖调 buildin 里面的函数:

Example overriding the len() builtin function with a len() function injected in the global namespace:

def func(obj):
    print("length: %s" % len(obj))

func("abc")
len = lambda obj: "mock!"
func("abc")

Output:

length: 3
length: mock!

Builtins are references in multiple places:

Builtins 在很多地方都有被提及:

  • the builtins module
  • frames have a f_builtins attribute (builtins dictionary)
  • the global PyInterpreterState structure has a builtins attribute (builtins dictionary)
  • frame globals have a __builtins__ variable (builtins dictionary, or builtins module when __name__ equals __main__)

Function code

还可以把整个函数的行为都作出修改:

It is possible to modify at runtime the bytecode of a function to modify completly its behaviour. Example:

def func(x, y):
    return x + y

print("1+2 = %s" % func(1, 2))

def mock(x, y):
    return 'mock'

func.__code__ = mock.__code__
print("1+2 = %s" % func(1, 2))

Output:

1+2 = 3
1+2 = mock

Local variables

还可以用比较 hack 的方法来让一个函数来修改这个函数外面的 local variable

Technically, it is possible to modify local variable of a function outside the function.

举个例子: 我们用 hack() 来修改调用它的函数的 local variable, 变量名为 x

Example of a function hack() which modifies the x local variable of its caller:

import sys
import ctypes

def hack():
    # Get the frame object of the caller
    frame = sys._getframe(1)
    frame.f_locals['x'] = "hack!"
    # Force an update of locals array from locals dict
    ctypes.pythonapi.PyFrame_LocalsToFast(ctypes.py_object(frame),
                                          ctypes.c_int(0))

def func():
    x = 1
    hack()
    print(x)

func()

Output:

hack!

Modification made from other modules

一个 Python 的模块 A 可以被模块 B 修改

A Python module A can be modified by a Python module B.

Multithreading

同时运行两个 Python 线程的时候,线程 B 可以修改线程 A 的共享资源。 即使是像 local variables 这样只有线程 A 可以访问的资源,仍然可以被 B 修改

When two Python threads are running, the thread B can modify shared resources of thread A, or even resources which are supposed to only be access by the thread A like local variables.

线程 B 可以修改函数代码,覆盖内置函数,修改 local variables

The thread B can modify function code, override builtin functions, modify local variables, etc.

Python Imports and Python Modules

The Python import path sys.path is initialized by multiple environment variables (ex: PYTHONPATH and PYTHONHOME), modified by the site module and can be modified anytime at runtime (by modifying sys.path directly).

Moreover, it is possible to modify sys.modules which is the “cache” between a module fully qualified name and the module object. For example, sys.modules['sys'] should be sys. It is posible to remove modules from sys.modules to force to reload a module. It is possible to replace a module in sys.modules.

The eventlet modules injects monkey-patched modules in sys.modules to convert I/O blocking operations to asynchronous operations using an event loop.

Solutions

Make strong assumptions, ignore changes

If the optimizer is an opt-in options, users are aware that the optimizer can make some compromises on the Python semantics to implement more aggressive optimizations.

Static analysis

Analyze the code to ensure that functions don’t mutate everything, for example ensure that a function is pure.

Dummy example:

def func(x, y):
    return x + y

This function func() is pure: it has no side effect. This function will not override builtins, not modify local variables of the caller, etc. It is safe to call this function from anywhere.

It is possible to analyze the code to check that an optimization can be enabled.

Use guards checked at runtime

For some optimizations, a static analysis cannot ensure that all assumptions required by an optimization will respected. Adding guards allows to check assumptions during the execution to use the optimized code or fallback to the original code.

Optimizations

See also fatoptimizer optimizations.

Inline function calls

Example:

def _get_sep(path):
    if isinstance(path, bytes):
        return b'/'
    else:
        return '/'

def isabs(s):
    """Test whether a path is absolute"""
    sep = _get_sep(s)
    return s.startswith(sep)

Inline _get_sep() into isabs() and simplify the code for the str type:

def isabs(s: str):
    return s.startswith('/')

It can be implemented as a simple call to the C function PyUnicode_Tailmatch().

Note: Inlining uses more memory and disk because the original function should be kept. Except if the inlined function is unreachable (ex: “private function”?).

Links:

  • Issue #10399: AST Optimization: inlining of function calls

Move invariants out of the loop

Example:

def func(obj, lines):
    for text in lines:
        print(obj.cleanup(text))

Become:

def func(obj, lines):
    local_print = print
    obj_cleanup = obj.cleanup
    for text in lines:
        local_print(obj_cleanup(text))

Local variables are faster than global variables and the attribute lookup is only done once.

C functions using only C types

Optimizations:

  • Avoid reference counting
  • Memory allocations on the heap
  • Release the GIL

Example:

def demo():
    s = 0
    for i in range(10):
        s += i
    return s

In specialized code, it may be possible to use basic C types like char or int instead of Python codes which can be allocated on the stack, instead of allocating objects on the heap. i and s variables are integers in the range [0; 45] and so a simple C type int (or even char) can be used:

PyObject *demo(void)
{
    int s, i;
    Py_BEGIN_ALLOW_THREADS
    s = 0;
    for(i=0; i<10; i++)
        s += i;
    Py_END_ALLOW_THREADS
    return PyLong_FromLong(s);
}

Note: if the function is slow, we may need to check sometimes if a signal was received.

Release the GIL

Many methods of builtin types don’t need the GIL. Example: "abc".startswith("def").

Replace calls to pure functions with the result

Examples:

  • len('abc') becomes 3
  • "python2.7".startswith("python") becomes True
  • math.log(32) / math.log(2) becomes 5.0

Can be implemented in the AST optimizer.

Constant propagation

Propagate constant values of variables. Example:

Original Constant propagation
def func()
    x = 1
    y = x
    return y
def func()
    x = 1
    y = 1
    return 1

Implemented in fatoptimizer.

Read also the Wikipedia article on copy propagation.

Constant folding

Compute simple operations at the compilation. Usually, at least arithmetic operations (a+b, a-b, a*b, etc.) are computed. Example:

Original Constant folding
def func()
    return 1 + 1
def func()
    return 2

Implemented in fatoptimizer and the CPython peephole optimizer.

See also

Peephole optimizer

See CPython peephole optimizer.

Loop unrolling

Example:

for i in range(4):
    print(i)

The loop body can be duplicated (twice in this example) to reduce the cost of a loop:

for i in range(0,4,2):
    print(i)
    print(i+1)
i = 3

Or the loop can be removed by duplicating the body for all loop iterations:

i=0
print(i)
i=1
print(i)
i=2
print(i)
i=3
print(i)

Combined with other optimizations, the code can be simplified to:

print('0')
print('1')
print('2')
i = 3
print('3')

Implemented in fatoptimizer

Read also the Wikipedia article on loop unrolling.

Dead code elimination

  • Replace if 0: code with pass
  • if DEBUG: print("debug") where DEBUG is known to be False

Implemented in fatoptimizer and the CPython peephole optimizer.

See also Wikipedia Dead code elimination article.

Load globals and builtins when the module is loaded

Load globals when the module is loaded? Ex: load “print” name when the module is loaded.

Example:

def hello():
    print("Hello World")

Become:

local_print = print

def hello():
    local_print("Hello World")

Useful if hello() is compiled to C code.

fatoptimizer implements a “copy builtins to constants optimization” optimization.

Don’t create Python frames

Inlining and other optimizations don’t create Python frames anymore. It can be a serious issue to debug programs: tracebacks are an important feature of Python.

At least in debug mode, frames should be created.

PyPy supports lazy creation of frames if an exception is raised.

Python bytecode

CPython peephole optimizer

Implementation: Python/peephole.c

Optmizations:

Latest enhancement:

changeset:   68375:14205d0fee45
user:        Antoine Pitrou <solipsis@pitrou.net>
date:        Fri Mar 11 17:27:02 2011 +0100
files:       Lib/test/test_peepholer.py Misc/NEWS Python/peephole.c
description:
Issue #11244: The peephole optimizer is now able to constant-fold
arbitrarily complex expressions.  This also fixes a 3.2 regression where
operations involving negative numbers were not constant-folded.

Should be rewritten as an AST optimizer.

Python C API

Intro

CPython comes with a C API called the “Python C API”. The most common type is PyObject* and functions are prefixed with Py (and _Py for private functions but you must not use them!).

Historical design choices

CPython was created in 1991 by Guido van Rossum. Some design choices made sense in 1991 but don’t make sense anymore in 2015. For example, the GIL was a simple and safe choice to implement multithreading in CPython. But in 2015, smartphones have 2 or 4 cores, and desktop PC have between 4 and 8 cores. The GIL restricts peek performances on multithreaded applications, even when it’s possible to release the GIL.

GIL

CPython uses a Global Interpreter Lock called “GIL” to avoid concurrent accesses to CPython internal structures (shared resources like global variables) to ensure that Python internals remain consistent.

See also Kill the GIL.

Reference counting and garbage collector

The C structure of all Python objects inherit from the PyObject structure which contains the field Py_ssize_t ob_refcnt;. This is a simple counter initialized to 1 when the object is created, increased each time that a variable has a strong reference to the object, and decreased each time that a strong reference is removed. The object is removed when the counter reached 0.

In some cases, two objects are linked together. For example, A has a strong reference to B which has a strong reference to A. Even if A and B are no more referenced outside, these objects are not destroyed because their reference counter is still equal to 1. A garbage collector is responsible to find and break reference cycles.

See also the PEP 442: Safe object finalization implemented in Python 3.4 which helps to break reference cycles.

AST Optimizers

Intro

An AST optimizer rewrites the Abstract Syntax Tree (AST) of a Python module to produce a more efficient code.

Currently in CPython 3.5, only basic optimizations are implemented by rewriting the bytecode: CPython peephole optimizer.

Old AST optimizer project

See old AST optimizer.

fatoptimizer

fatoptimizer project: AST optimizer implementing multiple optimizations and can specialize functions using guards of the fat module.

pythran AST

pythran.analysis.PureFunctions of pythran project, depend on ArgumentEffects and GlobalEffects analysis: automatically detect pure functions.

Old AST Optimizer

See also AST optimizers.

https://bitbucket.org/haypo/astoptimizer/ was a first attempt to optimize Python. This project was rejected by the Python community because it breaks the Python semantics. For example, it replaces len("abc") with 3. It checks that len() was not overriden in the module, but it doesn’t check that the builtin len() function was not overriden.

Threads on the Python-Dev mailing list:

The project was created in September 2012. It is now dead and replaced with the fatoptimizer project.

Introduction

astoptimizer is an optimizer for Python code working on the Abstract Syntax Tree (AST, high-level representration). It does as much work as possible at compile time.

The compiler is static, it is not a just-in-time (JIT) compiler, and so don’t expect better performances than psyco or PyPy for example. Optimizations depending on the type of functions parameters cannot be done for examples. Only optimizations on immutable types (constants) are done.

Website: http://pypi.python.org/pypi/astoptimizer

Source code hosted at: https://bitbucket.org/haypo/astoptimizer

Optimizations

  • Call builtin functions if arguments are constants (need “builtin_funcs” feature). Examples:
    • len("abc") => 3
    • ord("A") => 65
  • Call methods of builtin types if the object and arguments are constants. Examples:
    • u"h\\xe9ho".encode("utf-8") => b"h\\xc3\\xa9ho"
    • "python2.7".startswith("python") => True
    • (32).bit_length() => 6
    • float.fromhex("0x1.8p+0") => 1.5
  • Call functions of math and string modules for functions without border effect. Examples:
    • math.log(32) / math.log(2) => 5.0
    • string.atoi("5") => 5
  • Format strings for str%args and print(arg1, arg2, ...) if arguments are constants and the format string is valid. Examples:
    • "x=%s" % 5 => "x=5"
    • print(1.5) => print("1.5")
  • Simplify expressions. Examples:
    • not(x in y) => x not in y
    • 4 and 5 and x and 6 => x and 6
    • if a: if b: print("true") => if a and b: print("true")
  • Optimize loops (range => xrange needs “builtin_funcs” features). Examples:
    • while True: pass => while 1: pass
    • for x in range(3): print(x) => x = 0; print(x); x = 1; print(x); x = 2; print(x)
    • for x in range(1000): print(x) => for x in xrange(1000): print(x) (Python 2)
  • Optimize iterators, list, set and dict comprehension, and generators (need “builtin_funcs” feature). Examples:
    • iter(set()) => iter(())
    • frozenset("") => frozenset()
    • (x for x in "abc" if False) => (None for x in ())
    • [x*10 for x in range(1, 4)] => [10, 20, 30]
    • (x*2 for x in "abc" if True) => (x*2 for x in ("a", "b", "c"))
    • list(x for x in iterable) => list(iterable)
    • tuple(x for x in "abc") => ("a", "b", "c")
    • list(x for x in range(3)) => [0, 1, 2]
    • [x for x in ""] => []
    • [x for x in iterable] => list(iterable)
    • set([x for x in "abc"]) => {"a", "b", "c"} (Python 2.7+) or set(("a", "b", "c"))
  • Replace list with tuple (need “builtin_funcs” feature). Examples:
    • for x in [a, b, c]: print(x) => for x in (a, b, c): print(x)
    • x in [1, 2, 3] => x in (1, 2, 3)
    • list([x, y, z]) => [x, y, z]
    • set([1, 2, 3]) => {1, 2, 3} (Python 2.7+)
  • Evaluate unary and binary operators, subscript and comparaison if all arguments are constants. Examples:
    • 1 + 2 * 3 => 7
    • not True => False
    • "abc" * 3 => "abcabcabc"
    • "abcdef"[:3] => "abc"
    • (2, 7, 3)[1] => 7
    • frozenset("ab") | frozenset("bc") => frozenset("abc")
    • None is None => True
    • "2" in "python2.7" => True
    • x in [1, 2, 3] => x in {1, 2, 3} (Python 3) or x in (1, 2, 3) (Python 2)
    • def f(): return 2 if 4 < 5 else 3 => def f(): return 2
  • Remove empty loop. Example:
    • for i in (1, 2, 3): pass => i = 3
  • Remove dead code. Examples:
    • def f(): return 1; return 2 => def f(): return 1
    • def f(a, b): s = a+b; 3; return s => def f(a, b): s = a+b; return s
    • if DEBUG: print("debug") => pass with DEBUG declared as False
    • while 0: print("never executed") => pass

Use astoptimizer in your project

To enable astoptimizer globally on your project, add the following lines at the very begining of your application:

import astoptimizer
config = astoptimizer.Config('builtin_funcs', 'pythonbin')
# customize the config here
astoptimizer.patch_compile(config)

On Python 3.3, imports will then use the patched compile() function and so all modules will be optimized. With older versions, the compileall module (ex: compileall.compile_dir()) can be used to compile an application with optimizations enabled.

See also the issue #17515: Add sys.setasthook() to allow to use a custom AST optimizer.

Example

Example with the high-level function optimize_code:

from astoptimizer import optimize_code
code = "print(1+1)"
code = optimize_code(code)
exec(code)

Example the low-level functions optimize_ast:

from astoptimizer import Config, parse_ast, optimize_ast, compile_ast
config = Config('builtin_funcs', 'pythonbin')
code = "print(1+1)"
tree = parse_ast(code)
tree = optimize_ast(tree, config)
code = compile_ast(tree)
exec(code)

See also demo.py script.

Configuration

Unsafe optimizations are disabled by default. Use the Config() class to enable more optimizations.

Features enabled by default:

  • "builtin_types": methods of bytes, str, unicode, tuple, frozenset, int and float types
  • "math", "string": constants and functions without border effects of the math / string module

Optional features:

  • "builtin_funcs": builtin functions like abs(), str(), len(), etc. Examples:
    • len("abc") => 3
    • ord("A") => 65
    • str(123) => "123"
  • "pythonbin": Enable this feature if the optimized code will be executed by the same Python binary: so exactly the same Python version with the same build options. Allow to optimize non-BMP unicode strings on Python < 3.3. Enable the "platform" feature. Examples:
    • u"\\U0010ffff"[0] => u"\\udbff" or u"\\U0010ffff" (depending on build options, narrow or wide Unicode)
    • sys.version_info.major => 2
    • sys.maxunicode => 0x10ffff
  • "pythonenv": Enable this feature if you control the environment variables (like PYTHONOPTIMIZE) and Python command line options (like -Qnew). On Python 2, allow to optimize int/int. Enable "platform" and "pythonbin" features. Examples:
    • __debug__ => True
    • sys.flags.optimize => 0
  • "platform": optimizations specific to a platform. Examples:
    • sys.platform => "linux2"
    • sys.byteorder => "little"
    • sys.maxint => 2147483647
    • os.linesep => "\\n"
  • "struct": struct module, calcsize(), pack() and unpack() functions.
  • "cpython_tests": disable some optimizations to workaround issues with the CPython test suite. Only use it for tests.

Use Config("builtin_funcs", "pythonbin") to enable most optimizations. You may also enable "pythonenv" to enable more optimizations, but then the optimized code will depends on environment variables and Python command line options.

Use config.enable_all_optimizations() to enable all optimizations, which may generate invalid code.

Advices

Advices to help the AST optimizer:

  • Declare your constants using config.add_constant()
  • Declare your pure functions (functions with no border effect) using config.add_func()
  • Don’t use “from module import *”. If “import *” is used, builtins functions are not optimized anymore for example.

Limitations

  • Operations on mutable values are not optimized, ex: len([1, 2, 3]).
  • Unsafe optimizations are disabled by default. For example, len(“\U0010ffff”) is not optimized because the result depends on the build options of Python. Enable “builtin_funcs” and “pythonenv” features to enable more optimizations.
  • len() is not optimized if the result is bigger than 2^31-1. Enable “pythonbin” configuration feature to optimize the call for bigger objects.
  • On Python 2, operators taking a bytes string and a unicode string are not optimized if the bytes string has to be decoded from the default encoding or if the unicode string has to be encoded to the default encoding. Exception: pure ASCII strings are optimized. For example, b”abc” + u”def” is replaced with u”abcdef”, whereas u”x=%s” % b”\xe9” is not optimized.
  • On Python 3, comparaison between bytes and Unicode strings are not optimized because the comparaison may emit a warning or raise a BytesWarning exception. Bytes string are not converted to Unicode string. For example, b”abc” < “abc” and str(b”abc”) are not optimized. Converting a bytes string to Unicode is never optimized.

ChangeLog

Version 0.6 (2014-03-05)

  • Remove empty loop. Example: for i in (1, 2, 3): pass => i = 3.
  • Log removal of code
  • Fix support of Python 3.4: socket constants are now enum

Version 0.5 (2013-03-26)

  • Unroll loops (no support for break/continue yet) and list comprehension. Example: [x*10 for x in range(1, 4)] => [10, 20, 30].
  • Add Config.enable_all_optimizations() method
  • Add a more aggressive option to remove dead code (config.remove_almost_dead_code), disabled by default
  • Remove useless instructions. Example: “x=1; ‘abc’; print(x)” => “x=1; print(x)”
  • Remove empty try/except. Example: “try: pass except: pass” => “pass”

Version 0.4 (2012-12-10)

Bugfixes:

  • Don’t replace range() with xrange() if arguments cannot be converted to C long
  • Disable float.fromhex() optimization by default: float may be shadowed. Use “builtin_funcs” to enable this optimization.

Changes:

  • Add the “struct” configuration feature: functions of the struct module
  • Optimize print() on Python 2 with “from __future__ import print_function”
  • Optimize iterators, list, set and dict comprehension, and generators
  • Replace list with tuple
  • Optimize if a: if b: print("true"): if a and b: print("true")

Version 0.3.1 (2012-09-12)

Bugfixes:

  • Disable optimizations on functions and constants if a variable with the same name is set. Example: “len=ord; print(len(‘A’))”, “sys.version = ‘abc’; print(sys.version)”.
  • Don’t optimize print() function, frozenset() nor range() functions if “builtin_funcs” feature is disabled
  • Don’t remove code if it contains global or nonlocal. Example: “def f(): if 0: global x; x = 2”.

Version 0.3 (2012-09-11)

Major changes:

  • Add astoptimizer.patch_compile(config=None) function to simply hook the builtin compile() function.
  • Add “pythonbin” configuration feature.
  • Disable optimizations on builtin functions by default. Add “builtin_funcs” feature to the configuration to optimize builtin functions.
  • Remove dead code (optionnal optimization)
  • It is now posible to define a callback for warnings of the optimizer
  • Drop support of Python 2.5, it is unable to compile an AST tree to bytecode. AST objects of Python 2.5 don’t accept arguments in constructors.

Bugfixes:

  • Handle “from math import *” correctly
  • Don’t optimize operations if arguments are bytes and unicode strings. Only optimize if string arguments have the same type.
  • Disable optimizations on non-BMP unicode strings by default. Optimizations enabled with “pythonbin” feature.

Other changes:

  • More functions, methods and constants:
    • bytes, str, unicode: add more methods.
    • math module: add most remaining functions
    • string module: add some functions and all constants
  • not(a in b) => a not in b, not(a is b) => a is not b
  • a if bool else b
  • for x in range(n) => for x in xrange(n) (only on Python 2)
  • Enable more optimizations if a function is not a generator
  • Add sys.flags.<attr> and sys.version_info.<attr> constants

Version 0.2 (2012-09-02)

Major changes:

  • Check input arguments before calling an operator or a function, instead of catching errors.
  • New helper functions optimize_code() and optimize_ast() should be used instead of using directly the Optimizer class.
  • Support tuple and frozenset types

Changes:

  • FIX: add Config.max_size to check len(obj) result
  • FIX: disable non portable optimizations on non-BMP strings
  • Support Python 2.5-3.3
  • Refactor Optimizer: Optimizer.visit() now always visit children before calling the optimizer for a node, except for assignments
  • Float and complex numbers are no more restricted by the integer range of the configuration
  • More builtin functions. Examples: divmod(int, int), float(str), min(tuple), sum(tuple).
  • More method of builtin types. Examples: str.startswith(), str.find(), tuple.count(), float.is_integer().
  • math module: add math.ceil(), math.floor() and math.trunc().
  • More module constants. Examples: os.O_RDONLY, errno.EINVAL, socket.SOCK_STREAM.
  • More operators: a not in b, a is b, a is not b, +a.
  • Conversion to string: str(), str % args and print(arg1, arg2, ...).
  • Support import aliases. Examples: “import math as M; print(M.floor(1.5))” and “from math import floor as F; print(F(1.5))”.
  • Experimental support of variables (disabled by default).

Version 0.1 (2012-08-12)

  • First public version (to reserve the name on PyPI!)

Register-based Virtual Machine for Python

Intro

registervm is a fork of CPython 3.3 using register-based bytecode, instead of stack-code bytecode

More information: REGISTERVM.txt

Thread on the Python-Dev mailing list: Register-based VM for CPython.

The project was created in November 2012.

Status

  • Most instructions using the stack are converted to instructions using registers
  • Bytecode using registers with all optimizations enable is usually 10% faster than bytecode using the stack, according to pybench
  • registervm generates invalid code, see TODO section below, so it’s not possible yet to use it on the Python test suite

TODO

Bugs

  • Register allocator doesn’t handle correctly conditional branches: CLEAR_REG is removed on the wrong branch in test_move_instr.

  • Fail to track the stack state in if/else. Bug hidden by the register allocator in the following example:

    def func(obj):

    obj.attr = sys.modules[‘warnings’] if module is None else module

  • Don’t move globals out of if. Only out of loops? subprocess.py:

    if mswindows:
        if p2cwrite != -1:
            p2cwrite = msvcrt.open_osfhandle(p2cwrite.Detach(), 0)
    

    But do move len() out of loop for:

    def loop_move_instr():
        length = 0
        for i in range(5):
            length += len("abc") - 1
        return length
    
  • Don’t remove duplicate LOAD_GLOBAL in “LOAD_GLOBAL ...; CALL_PROCEDURE ...; LOAD_GLOBAL ...”: CALL_PROCEDURE has border effect

  • Don’t remove duplicate LOAD_NAME if a function has a border effect:

    x=1
    def modify():
        global x
        x = 2
    print(x)
    modify()
    print(x)
    

Improvments

  • Move LOAD_CONST out of loops: it was done in a previous version, but the optimization was broken by the introduction of CLEAR_REG

  • Copy constants to the frame objects so constants can be used as registers and LOAD_CONST instructions can be simplify removed

  • Enable move_load_const by default?

  • Fix moving LOAD_ATTR_REG: only do that when calling methods. See test_sieve() of test_registervm: primes.append().

    result = Result()
    while 1:
        if result.done:
            break
        func(result)
    
  • Reenable merging duplicate LOAD_ATTR

  • Register allocation for locale_alias = {...} is very very slow

  • “while 1: ... return” generates useless SETUP_LOOP

  • Reuse locals?

  • implement register version of the following instructions:

    • DELETE_ATTR
    • try/finally
    • yield from
    • CALL_FUNCTION_VAR_KW
    • CALL_FUNCTION_VAR
    • operators: a | b, a & b, a ^ b, a |= b, a &= b, a ^= b
  • DEREF:

    • add a test using free variables
    • Move LOAD_DEREF_REG out of loops
  • NAME:

    • test_list_append() of test_registervm.py
    • Move LOAD_NAME_REG out of loop
  • Handle JUMP_IF_TRUE_OR_POP: see test_getline() of test_registervm

  • Compute the number of used registers in a frame

  • Write a new test per configuration option

  • Factorize code processing arg_types, ex: disassmblers of dis and registervm modules

  • Add tests on class methods

  • Fix lnotab

Changelog

2012-12-21

  • Use RegisterTracker to merge duplicated LOAD, STORE_GLOBAL/LOAD_GLOBAL are now also simplified

2012-12-19

  • Emit POP_REG to simplify the stack tracker

2012-12-18

  • LOAD are now only moved out of loops

2012-12-14

  • Duplicated LOAD instructions can be merged without moving them
  • Rewrite the stack tracker: PUSH_REG don’t need to be moved anymore
  • Fix JUMP_IF_TRUE_OR_POP/JUMP_IF_FALSE_OR_POP to not generate invalid code
  • Don’t move LOAD_ATTR_REG out of try/except block

2012-12-11

  • Split instructions into linked-blocks

2012-11-26

  • Add a stack tracker

2012-11-20

  • Remove useless jumps
  • CALL_FUNCTION_REG and CALL_PROCEDURE_REG are fully implemented

2012-10-29

  • Remove “if (HAS_ARG(op))” check in PyEval_EvalFrameEx()

2012-10-27

  • Duplicated LOAD_CONST and LOAD_GLOBAL are merged (optimization disabled on LOAD_GLOBAL because it is buggy)

2012-10-23

  • initial commit, 0f7f49b7083c

CPython 3.3 bytecode is inefficient

  • Useless jump: JUMP_ABSOLUTE <offset+0>
  • Generate dead code: RETURN_VALUE; RETURN_VALUE (the second instruction is unreachable)
  • Duplicate constants: see TupleSlicing of pybench
  • Constant folding: see astoptimizer project
  • STORE_NAME ‘f’; LOAD_NAME ‘f’
  • STORE_GLOBAL ‘x’; LOAD_GLOBAL ‘x’

Rationale

The performance of the loop evaluating bytecode is critical in Python. For Python example, using computed-goto instead of switch to dispatch bytecode improved performances by 20%. Related issues:

Using registers of a stack reduce the number of operations, but increase the size of the code. I expect an significant speedup when all operations will use registers.

Optimizations

Optimizations:

  • Remove useless LOAD_NAME and LOAD_GLOBAL. For example: “STORE_NAME var; LOAD_NAME var”
  • Merge duplicate loads (LOAD_CONST, LOAD_GLOBAL_REG, LOAD_ATTR). For example, “lst.append(1); lst.append(1)” only gets constant “1” and the “lst.append” attribute once.

Misc:

  • Automatically detect inplace operations. For example, “x = x + y” is compiled to “BINARY_ADD_REG ‘x’, ‘x’, ‘y’” which calls PyNumber_InPlaceAdd(), instead of PyNumber_Add().
  • Move constant, global and attribute loads out of loops (to the beginning)
  • Remove useless jumps (ex: JUMP_FORWARD <relative jump to 103 (+0)>)

Algorithm

The current implementation rewrites the stack-based operations to use register-based operations instead. For example, “LOAD_GLOBAL range” is replaced with “LOAD_GLOBAL_REG R0, range; PUSH_REG R0”. This first step is inefficient because it increases the number of operations.

Then, operations are reordered: PUSH_REG and POP_REG to the end. So we can replace “PUSH_REG R0; PUSH_REG R1; STACK_OPERATION; POP_REG R2” with a single operatiton: “REGISTER_OPERATION R2, R0, R1”.

Move invariant out of the loop: it is possible to move constants out of the loop. For example, LOAD_CONST_REG are moved to the beginning. We might also move LOAD_GLOBAL_REG and LOAD_ATTR_REG to the beginning.

Later, a new AST to bytecote compiler can be implemented to emit directly operations using registers.

Example

Simple function computing the factorial of n:

def fact_iter(n):
    f = 1
    for i in range(2, n+1):
        f *= i
    return f

Stack-based bytecode (20 instructions):

      0 LOAD_CONST           1 (const#1)
      3 STORE_FAST           'f'
      6 SETUP_LOOP           <relative jump to 46 (+37)>
      9 LOAD_GLOBAL          0 (range)
     12 LOAD_CONST           2 (const#2)
     15 LOAD_FAST            'n'
     18 LOAD_CONST           1 (const#1)
     21 BINARY_ADD
     22 CALL_FUNCTION        2 (2 positional, 0 keyword pair)
     25 GET_ITER
>>   26 FOR_ITER             <relative jump to 45 (+16)>
     29 STORE_FAST           'i'
     32 LOAD_FAST            'f'
     35 LOAD_FAST            'i'
     38 INPLACE_MULTIPLY
     39 STORE_FAST           'f'
     42 JUMP_ABSOLUTE        <jump to 26>
>>   45 POP_BLOCK
>>   46 LOAD_FAST            'f'
     49 RETURN_VALUE

Register-based bytecode (13 instructions):

      0 LOAD_CONST_REG       'f', 1 (const#1)
      5 LOAD_CONST_REG       R0, 2 (const#2)
     10 LOAD_GLOBAL_REG      R1, 'range' (name#0)
     15 SETUP_LOOP           <relative jump to 57 (+39)>
     18 BINARY_ADD_REG       R2, 'n', 'f'
     25 CALL_FUNCTION_REG    4, R1, R1, R0, R2
     36 GET_ITER_REG         R1, R1
>>   41 FOR_ITER_REG         'i', R1, <relative jump to 56 (+8)>
     48 INPLACE_MULTIPLY_REG 'f', 'i'
     53 JUMP_ABSOLUTE        <jump to 41>
>>   56 POP_BLOCK
>>   57 RETURN_VALUE_REG     'f'

The body of the main loop of this function is composed of 1 instructions instead of 5.

Comparative table

Example     |S|r|R|            Stack                 |         Register
------------+-+-+-+----------------------------------+----------------------------------------------------
append(2)   |4|1|2| LOAD_FAST 'append'               | LOAD_CONST_REG R1, 2 (const#2)
            | | | | LOAD_CONST 2 (const#2)           | ...
            | | | | CALL_FUNCTION (1 positional)     | ...
            | | | | POP_TOP                          | CALL_PROCEDURE_REG 'append', (1 positional), R1
------------+-+-+-+----------------------------------+----------------------------------------------------
l[0] = 3    |4|1|2| LOAD_CONST 3 (const#1)           | LOAD_CONST_REG R0, 3 (const#1)
            | | | | LOAD_FAST 'l'                    | LOAD_CONST_REG R3, 0 (const#4)
            | | | | LOAD_CONST 0 (const#4)           | ...
            | | | | STORE_SUBSCR                     | STORE_SUBSCR_REG 'l', R3, R0
------------+-+-+-+----------------------------------+----------------------------------------------------
x = l[0]    |4|1|2| LOAD_FAST 'l'                    | LOAD_CONST_REG R3, 0 (const#4)
            | | | | LOAD_CONST 0 (const#4)           | ...
            | | | | BINARY_SUBSCR                    | ...
            | | | | STORE_FAST 'x'                   | BINARY_SUBSCR_REG 'x', 'l', R3
------------+-+-+-+----------------------------------+----------------------------------------------------
s.isalnum() |4|1|2| LOAD_FAST 's'                    | LOAD_ATTR_REG R5, 's', 'isalnum' (name#3)
            | | | | LOAD_ATTR 'isalnum' (name#3)     | ...
            | | | | CALL_FUNCTION (0 positional)     | ...
            | | | | POP_TOP                          | CALL_PROCEDURE_REG R5, (0 positional)
------------+-+-+-+----------------------------------+----------------------------------------------------
o.a = 2     |3|1|2| LOAD_CONST 2 (const#3)           | LOAD_CONST_REG R2, 2 (const#3)
            | | | | LOAD_FAST 'o'                    | ...
            | | | | STORE_ATTR 'a' (name#2)          | STORE_ATTR_REG 'o', 'a' (name#2), R2
------------+-+-+-+----------------------------------+----------------------------------------------------
x = o.a     |3|1|1| LOAD_FAST 'o'                    | LOAD_ATTR_REG 'x', 'o', 'a' (name#2)
            | | | | LOAD_ATTR 'a' (name#2)           |
            | | | | STORE_FAST 'x'                   |
------------+-+-+-+----------------------------------+----------------------------------------------------

Columns:

  • “S”: Number of stack-based instructions
  • “r”: Number of stack-based instructions exclusing instructions moved out of loops (ex: LOAD_CONST_REG)
  • “R”: Total number of stack-based instructions (including instructions moved out of loops)

Read-only Python

Intro

A first attempt to implement guards was the readonly PoC (fork of CPython 3.5) which registered callbacks to notify all guards. The problem is that modifying a watched dictionary gets a complexity of O(n) where n is the number of registered guards.

readonly adds a modified flag to types and a readonly property to dictionaries. The guard was notified with the modified key to decide to disable or not the optimization.

More information: READONLY.txt

Thread on the python-ideas mailing list: Make Python code read-only.

The project was mostly developed in May 2014. The project is now dead, replaced with FAT Python.

READONLY

This fork on CPython 3.5 adds a machinery to be notified when the Python code is modified. Modules, classes (types) and functions are tracked. At the first modification, a callback is called with the object and the modified attribute.

This machinery should help static optimizers. See this article for more information: http://haypo-notes.readthedocs.org/faster_cpython.html

Examples of such optimizers:

  • astoptimizer project: replace a function call by its result during the AST compilation
  • Learn types of function paramters and local variables, and then compile Python (byte)code to machine code specialized for these types (like Cython)

Issues with read-only code

  • Currently, it’s not possible to allow again to modify a module, class or function to keep my implementation simple. With a registry of callbacks, it may be possible to enable again modification and call code to disable optimizations.
  • PyPy implements this but thanks to its JIT, it can optimize again the modified code during the execution. Writing a JIT is very complex, I’m trying to find a compromise between the fast PyPy and the slow CPython. Add a JIT to CPython is out of my scope, it requires too much modifications of the code.
  • With read-only code, monkey-patching cannot be used anymore. It’s annoying to run tests. An obvious solution is to disable read-only mode to run tests, which can be seen as unsafe since tests are usually used to trust the code.
  • The sys module cannot be made read-only because modifying sys.stdout and sys.ps1 is a common use case.
  • The warnings module tries to add a __warningregistry__ global variable in the module where the warning was emited to not repeat warnings that should only be emited once. The problem is that the module namespace is made read-only before this variable is added. A workaround would be to maintain these dictionaries in the warnings module directly, but it becomes harder to clear the dictionary when a module is unloaded or reloaded. Another workaround is to add __warningregistry__ before making a module read-only.
  • Lazy initialization of module variables does not work anymore. A workaround is to use a mutable type. It can be a dict used as a namespace for module modifiable variables.
  • The interactive interpreter sets a “_” variable in the builtins namespace. I have no workaround for this. The “_” variable is no more created in read-only mode. Don’t run the interactive interpreter in read-only mode.
  • It is not possible yet to make the namespace of packages read-only. For example, “import encodings.utf_8” adds the symbol “utf_8” to the encodings namespace. A workaround is to load all submodules before making the namespace read-only. This cannot be done for some large modules. For example, the encodings has a lot of submodules, only a few are needed.

STATUS

  • Python API:
    • new function.__modified__ and type.__modified__ properties: False by default, becomes True when the object is modified
    • new module.is_modified() method
    • new module.set_initialized() method
  • C API:
    • PyDictObject: new “int ma_readonly;” field
    • PyTypeObject: a new “int tp_modified;” field
    • PyFunctionObject: new “int func_module;” and “int func_initialized;” fields
    • PyModuleObject: new “int md_initialized;” field

Modified modules, classes and functions

  • It’s common to modify the following attributes of the sys module:
    • sys.ps1, sys.ps2
    • sys.stdin, sys.stdout, sys.stderr
  • “import encodings.latin_1” sets “latin_1” attribute in the namespace of the “encodings” module.
  • The interactive interpreter sets the “_” variable in builtins.
  • warnings: global variable __warningregistry__ set in modules
  • functools.wraps() modifies the wrapper to copy attributes of the wrapped function

TODO

  • builtins modified in initstdio(): builtins.open modified
  • sys modified in initstdio(): sys.__stdin__ modified
  • structseq: types are created modified; same issue with _ast types (Python-ast.c)
  • module, type and function __dict__:
    • Drop dict.setreadonly()
    • Decide if it’s better to use dict.setreadonly() or a new subclass (ex: “dict_maybe_readonly” or “namespace”).
    • Read only dict: add a new ReadOnlyError instead of ValueError?
    • sysmodule.c: PyDict_DelItemString(FlagsType.tp_dict, “__new__”) doesn’t mark FlagsType as modified
    • Getting func.__dict__ / module.__dict__ marks the function/module as modified, this is wrong. Use instead a mapping marking the function as modified when the mapping is modified.
    • module.__dict__ is read-only: similar issue for functions.
  • Import submodule. Example: “import encodings.utf_8” modifies “encoding” to set a new utf_8 attribute

TODO: Specialized functions

Environment

  • module and type attribute values:
    • (“module”, “os”, OS_CHECKSUM)
    • (“attribute”, “os.path”)
    • (“module”, “path”, PATH_CHECKSUM)
    • (“attribute”, “path.isabs”)
    • (“function”, “path.isabs”)
  • function attributes
  • set of function parameter types (passed as indexed or keyword arguments)

Read-only state

Scenario:

  • 1: application.py is compiled. Function A depends on os.path.isabs, function B depends on project.DEBUG
  • 2: application is started, “import os.path”
  • 3: os.path.isabs is modified
  • 4: optimized application.py is loaded
  • 5: project.DEBUG is modified

When the function is created, os.path.isabs was already modified compared to the OS_CHECKSUM.

Example of environments

  • The function calls “os.path.isabs”:
    • rely on “os.path” attribute
    • rely on “os.path.isabs” attribute
    • rely on “os.path.isabs” function attributes (except __doc__)
  • The function “def mysum(x, y):” has two parameters
    • x type is int and y type is int
    • or: x type is str and y type is str
    • (“type is”: check the exact type, not a subclass)
  • The function uses “project.DEBUG” constant
    • rely on “project.DEBUG” attribute

Content of a function

  • classic attributes: doc, etc.
  • multiple versions of the code:
    • required environment of the code
    • bytecode

Create a function

  • build the environment
  • register on module, type and functions modification

Callback when then environment is modified

xxx

Call a function

xxx

History of Python optimizations

  • 2002: Creation of the psyco project by Armin Rigo
  • 2003-05-05: psyco 1.0 released
  • Spring 1997: Creation of Jython project (initially called JPython) by Jim Hugunin
  • xxxx-xx-xx: Creation of IronPython project by xxx
  • Creation of PyPy, spin-off of psyco
  • mid-2007: PyPy 1.0 released.
  • 2009-03: Creation of Unladen Swallow project by xxx
  • xxxx-xx-xx: Creation of Pyston project by xxx
  • 2012-09: Creation of the AST optimizer project by Victor Stinner
  • 2012-11: Creation of the registervm project by Victor Stinner
  • 2014-05: Creation of read-only Python PoC by Victor Stinner
  • 2015-10: Creation of the FAT Python project by Victor Stinner

Misc

Ideas

  • PyPy CALL_METHOD instructor
  • Lazy formatting of Exception message: in most cases, the message is not used. AttributeError(message) => AttributeError(attr=name), lazy formatting for str(exc) and exc.args.

Plan

Other idea:

Status

See also the status of individual projects:

Done

  • astoptimizer project exists: astoptimizer.
  • Fork of CPython 3.5: be notified when the Python code is changed: modules, types and functions are tracked. My fork of CPython 3.5: readonly; read READONLY.txt documentation.

Note

“readonly” is no more a good name for the project. The name comes from a first implementation using ead-only code.

To do

  • Learn types
  • Enhance astoptimizer to use the type information
  • Emit machine code

Why Python is slow?

Why the CPython implementation is slower than PyPy?

  • everything is stored as an object, even simple types like integers or characters. Computing the sum of two numbers requires to “unbox” objects, compute the sum, and “box” the result.
  • Python maintains different states: thread state, interperter state, frames, etc. These informations are available in Python. The common usecase is to display a traceback in case of a bug. PyPy builds frames on demand.
  • Cost of maintaince the reference counter: Python programs rely on the garbage collector
  • ceval.c uses a virtual stack instead of CPU registers

Why the Python language is slower than C?

  • modules are mutable, classes are mutable, etc. Because of that, it is not possible to inline code nor replace a function call by its result (ex: len(“abc”)).
  • The types of function parameters and variables are unknown. Example of missing optimizations:
    • “obj.attr” instruction cannot be moved out of a loop: “obj.attr” may return a different result at each call, or execute arbitrary Python code
    • x+0 raises a TypeError for “abc”, whereas it is a noop for int (it can be replaced with just x)
    • conditional code becomes dead code when types are known
  • obj.method creates a temporary bounded method

Why improving CPython instead of writing a new implementation?

  • There are already a lot of other Python implementations. Some examples: PyPy, Jython, IronPython, Pyston.
  • CPython remains the reference implementation: new features are first implemented in CPython. For example, PyPy doesn’t support Python 3 yet.
  • Important third party modules rely heavily on CPython implementation details, especially the Python C API. Examples: numpy and PyQt.

Why not a JIT?

  • write a JIT is much more complex, it requires deep changes in CPython; CPython code is old (+20 years)
  • cost to “warn up” the JIT: Mercurial project is concerned by the Python startup time
  • Store generated machine code?

Learn types

  • Add code in the compiler to record types of function calls. Run your program. Use recorded types.
  • Range of numbers (predict C int overflow)
  • Optional paramters: forceload=0. Dead code with forceload=0.
  • Count number of calls to the function to decide if it should be optimized or not.
  • Measure time spend in a function. It can be used to decide if it’s useful to release or not the GIL.
  • Store type information directly in the source code? Manual type annotation?

Emit machine code

  • Limited to simple types like integers?
  • Use LLVM?
  • Reuse Cython or numba?
  • Replace bytecode with C functions calls. Ex: instead of PyNumber_Add(a, b) for a+b, emit PyUnicode_Concat(a, b), long_add(a, b) or even simpler code without unbox/box
  • Calling convention: have two versions of the function? only emit the C version if it is needed?
    • Called from Python: Python C API, PyObject* func(PyObject *args, PyObject *kwargs)
    • Called from C (specialized machine code): C API, int func(char a, double d)
    • Version which doesn’t need the GIL to be locked?
  • Option to compile a whole application into machine code for proprietary software?

Example of (specialized) machine code

Python code:

def mysum(a, b):
    return a + b

Python bytecode:

0 LOAD_FAST                0 (a)
3 LOAD_FAST                1 (b)
6 BINARY_ADD
7 RETURN_VALUE

C code used to executed bytecode (without code to read bytecode and handle signals):

/* LOAD_FAST */
{
    PyObject *value = GETLOCAL(0);
    if (value == NULL) {
        format_exc_check_arg(PyExc_UnboundLocalError, ...);
        goto error;
    }
    Py_INCREF(value);
    PUSH(value);
}

/* LOAD_FAST */
{
    PyObject *value = GETLOCAL(1);
    if (value == NULL) {
        format_exc_check_arg(PyExc_UnboundLocalError, ...);
        goto error;
    }
    Py_INCREF(value);
    PUSH(value);
}

/* BINARY_ADD */
{
    PyObject *right = POP();
    PyObject *left = TOP();
    PyObject *sum;
    if (PyUnicode_CheckExact(left) &&
             PyUnicode_CheckExact(right)) {
        sum = unicode_concatenate(left, right, f, next_instr);
        /* unicode_concatenate consumed the ref to v */
    }
    else {
        sum = PyNumber_Add(left, right);
        Py_DECREF(left);
    }
    Py_DECREF(right);
    SET_TOP(sum);
    if (sum == NULL)
        goto error;
}

/* RETURN_VALUE */
{
    retval = POP();
    why = WHY_RETURN;
    goto fast_block_end;
}

Specialized and simplified C code if both arguments are Unicode strings:

/* LOAD_FAST */
PyObject *left = GETLOCAL(0);
if (left == NULL) {
    format_exc_check_arg(PyExc_UnboundLocalError, ...);
    goto error;
}
Py_INCREF(left);

/* LOAD_FAST */
PyObject *right = GETLOCAL(1);
if (right == NULL) {
    format_exc_check_arg(PyExc_UnboundLocalError, ...);
    goto error;
}
Py_INCREF(right);

/* BINARY_ADD */
PyUnicode_Append(&left, right);
Py_DECREF(right);
if (sum == NULL)
    goto error;

/* RETURN_VALUE */
retval = left;
why = WHY_RETURN;
goto fast_block_end;

Test if the specialized function can be used

Write code to choose between the bytecode evaluation and the machine code.

Preconditions:

  • Check if os.path.isabs() was modified:
    • current namespace was modified? (os name cannot be replaced)
    • namespace of the os.path module was modified?
    • os.path.isabs function was modified?
    • compilation: checksum of the os.py and posixpath.py?
  • Check the exact type of arguments
    • x type is str: in C, PyUnicode_CheckExact(x)
    • list of int: check the whole array before executing code? fallback in the specialized code to handle non int items?
  • Callback to use the slow-path if something is modified?
  • Disable optimizations when tracing is enabled
  • Online benchmark to decide if preconditions and optimized code is faster than the original code?

Kill the GIL?

See the Global Interpreter Lock.

Why does CPython need a global lock?

Incomplete list:

  • Python memory allocation is not thread safe (it should be easy to make it thread safe)
  • The reference counter of each object is protected by the GIL.
  • CPython has a lot of global C variables. Examples:
    • interp is a structure which contains variables of the Python interpreter: modules, list of Python threads, builtins, etc.
    • int singletons (-5..255)
    • str singletons (Python 3: latin1 characters)
  • Some third party C libraries and even functions the C standard library are not thread safe: the GIL works around this limitation.

Kill the GIL

  • Require deep changes of CPython code
  • The current Python C API is too specific to CPython implementation details: need a new API. Maybe the stable ABI?
  • Modify third party modules to use the stable ABI to avoid relying on CPython implementation details like reference couting
  • Replace reference counting with something else? Atomic operations?
  • Use finer locks on some specific operations (release the GIL)? like operations on builtin types which don’t need to execute arbitrary Python code. Counter example: dict where keys are objects different than int and str.

See also pyparallel.

Implementations of Python

Faster Python implementations

Fully Python compliant

Other

  • Replace stack-based bytecode with register-based bytecode: old registervm project

Fully Python compliant??

  • psyco: JIT. The author of pysco, Armin Rigo, co-created the PyPy project.

Subset of Python to C++

Subset of Python

Language very close to Python

  • Cython: “Cython is a programming language based on Python, with extra syntax allowing for optional static type declarations.”

Benchmarks

See also:

PEP 509: Add a private version to dict

Moved to python.org: PEP 509 – Add a private version to dict

PEP 510: Specialized functions with guards

Moved to python.org: PEP 510 – Specialized functions with guards.

PEP 511: API for AST transformers

Moved to python.org: PEP 511 – API for AST transformers

Random notes about PyPy

What is the problem with PyPy?

PyPy is fast, much faster than CPython, but it’s still not widely used by users. What is the problem? Or what are the problems?

  • Bad support of the Python C API: PyPy was written from scratch and uses different memory structures for objects. The cpyext module emulates the Python C API but it’s slow.
  • New features are first developped in CPython. In january 2016, PyPy only supports Python 2.7 and 3.2, whereas CPython is at the version 3.5. It’s hard to have a single code base for Python 2.7 and 3.2, Python 3.3 reintroduced u'...' syntax for example.
  • Not all modules are compatible with PyPy: see PyPy Compatibility Wiki. For example, numpy is not compatible with PyPy, but there is a project under development: pypy/numy. PyGTK, PyQt, PySide and wxPython libraries are not compatible with PyPy; these libraries heavily depend on the Python C API. GUI applications (ex: gajim, meld) using these libraries don’t work on PyPy :-( Hopefully, a lot of popular modules are compatible with PyPy (ex: Django, Twisted).
  • PyPy is slower than CPython on some use cases. Django: “PyPy runs the templating engine faster than CPython, but so far DB access is slower for some drivers.” (source of the quote)

If I understood correctly, Pyjston will have same problems than PyPy since it doesn’t support the Python C API neither. Same issue for Pyjion?

Talks

Talks about Python optimizations:

  • PyCon UK 2014 - When performance matters ... <http://www.egenix.com/library/presentations/PyCon-UK-2014-When-performance-matters/> by Marc-Andre Lemburg