PyPy (Posts by Antonio Cuni)

#pypy IRC moves to Libera.Chat

Antonio Cuni — Mon, 31 May 2021 10:00:00 GMT

Following the example of many other FOSS projects, the PyPy team has decided to move its official #pypy IRC channel from Freenode to Libera.Chat: irc.libera.chat/pypy

The core devs will no longer be present on the Freenode channel, so we recommend to join the new channel as soon as possible.

wikimedia.org has a nice guide on how to setup your client to migrate from Freenode to Libera.Chat.

New HPy blog

Antonio Cuni — Mon, 29 Mar 2021 14:00:00 GMT

Regular readers of this blog already know about HPy, a project which aims to develop a new C API for Python to make it easier/faster to support C extensions on alternative Python implementations, including PyPy.

The HPy team just published the first post of HPy new blog, so if you are interested in its development, make sure to check it out!

HPy kick-off sprint report

Antonio Cuni — Wed, 18 Dec 2019 13:38:00 GMT

Recently Antonio, Armin and Ronan had a small internal sprint in the beautiful city of Gdańsk to kick-off the development of HPy. Here is a brief report of what was accomplished during the sprint.

What is HPy?

The TL;DR answer is "a better way to write C extensions for Python".

The idea of HPy was born during EuroPython 2019 in Basel, where there was an informal meeting which included core developers of PyPy, CPython (Victor Stinner and Mark Shannon) and Cython (Stefan Behnel). The ideas were later also discussed with Tim Felgentreff of GraalPython, to make sure they would also be applicable to this very different implementation, Windel Bouwman of RustPython is following the project as well.

All of us agreed that the current design of the CPython C API is problematic for various reasons and, in particular, because it is too tied to the current internal design of CPython. The end result is that:

alternative implementations of Python (such as PyPy, but not only) have a hard time loading and executing existing C extensions;
CPython itself is unable to change some of its internal implementation details without breaking the world. For example, as of today it would be impossible to switch from using reference counting to using a real GC, which in turns make it hard for example to remove the GIL, as gilectomy attempted.

HPy tries to address these issues by following two major design guidelines:

objects are referenced and passed around using opaque handles, which are similar to e.g., file descriptors in spirit. Multiple, different handles can point to the same underlying object, handles can be duplicated and each handle must be released independently of any other duplicate.
The internal data structures and C-level layout of objects are not visible nor accessible using the API, so each implementation if free to use what fits best.

The other major design goal of HPy is to allow incremental transition and porting, so existing modules can migrate their codebase one method at a time. Moreover, Cython is considering to optionally generate HPy code, so extension module written in Cython would be able to benefit from HPy automatically.

More details can be found in the README of the official HPy repository.

Target ABI

When compiling an HPy extension you can choose one of two different target ABIs:

HPy/CPython ABI: in this case, hpy.h contains a set of macros and static inline functions. At compilation time this translates the HPy API into the standard C-API. The compiled module will have no performance penalty, and it will have a "standard" filename like foo.cpython-37m-x86_64-linux-gnu.so.
Universal HPy ABI: as the name implies, extension modules compiled this way are "universal" and can be loaded unmodified by multiple Python interpreters and versions. Moreover, it will be possible to dynamically enable a special debug mode which will make it easy to find e.g., open handles or memory leaks, without having to recompile the extension.

Universal modules can also be loaded on CPython, thanks to the hpy_universal module which is under development. An extra layer of indirection enables loading extensions compiled with the universal ABI. Users of hpy_universal will face a small performance penalty compared to the ones using the HPy/CPython ABI.

This setup gives several benefits:

Extension developers can use the extra debug features given by the Universal ABI with no need to use a special debug version of Python.
Projects which need the maximum level of performance can compile their extension for each relevant version of CPython, as they are doing now.
Projects for which runtime speed is less important will have the choice of distributing a single binary which will work on any version and implementation of Python.

A simple example

The HPy repo contains a proof of concept module. Here is a simplified version which illustrates what a HPy module looks like:

#include "hpy.h"

HPy_DEF_METH_VARARGS(add_ints)
static HPy add_ints_impl(HPyContext ctx, HPy self, HPy *args, HPy_ssize_t nargs)
{
    long a, b;
    if (!HPyArg_Parse(ctx, args, nargs, "ll", &a, &b))
        return HPy_NULL;
    return HPyLong_FromLong(ctx, a+b);
}


static HPyMethodDef PofMethods[] = {
    {"add_ints", add_ints, HPy_METH_VARARGS, ""},
    {NULL, NULL, 0, NULL}
};

static HPyModuleDef moduledef = {
    HPyModuleDef_HEAD_INIT,
    .m_name = "pof",
    .m_doc = "HPy Proof of Concept",
    .m_size = -1,
    .m_methods = PofMethods
};


HPy_MODINIT(pof)
static HPy init_pof_impl(HPyContext ctx)
{
    HPy m;
    m = HPyModule_Create(ctx, &moduledef);
    if (HPy_IsNull(m))
        return HPy_NULL;
    return m;
}

People who are familiar with the current C-API will surely notice many similarities. The biggest differences are:

Instead of PyObject *, objects have the type HPy, which as explained above represents a handle.
You need to explicitly pass an HPyContext around: the intent is primary to be future-proof and make it easier to implement things like sub- interpreters.
HPy_METH_VARARGS is implemented differently than CPython's METH_VARARGS: in particular, these methods receive an array of HPy and its length, instead of a fully constructed tuple: passing a tuple makes sense on CPython where you have it anyway, but it might be an unnecessary burden for alternate implementations. Note that this is similar to the new METH_FASTCALL which was introduced in CPython.
HPy relies a lot on C macros, which most of the time are needed to support the HPy/CPython ABI compilation mode. For example, HPy_DEF_METH_VARARGS expands into a trampoline which has the correct C signature that CPython expects (i.e., PyObject (*)(PyObject *self, *PyObject *args)) and which calls add_ints_impl.

Sprint report and current status

After this long preamble, here is a rough list of what we accomplished during the week-long sprint and the days immediatly after.

On the HPy side, we kicked-off the code in the repo: at the moment of writing the layout of the directories is a bit messy because we moved things around several times, but we identified several main sections:

A specification of the API which serves both as documentation and as an input for parts of the projects which are automatically generated. Currently, this lives in public_api.h.
A set of header files which can be used to compile extension modules: depending on whether the flag -DHPY_UNIVERSAL_ABI is passed to the compiler, the extension can target the HPy/CPython ABI or the HPy Universal ABI
A CPython extension module called hpy_universal which makes it possible to import universal modules on CPython
A set of tests which are independent of the implementation and are meant to be an "executable specification" of the semantics. Currently, these tests are run against three different implementations of the HPy API:
- the headers which implements the "HPy/CPython ABI"
- the hpy_universal module for CPython
- the hpy_universal module for PyPy (these tests are run in the PyPy repo)

Moreover, we started a PyPy branch in which to implement the hpy_univeral module: at the moment of writing PyPy can pass all the HPy tests apart the ones which allow conversion to and from PyObject *. Among the other things, this means that it is already possible to load the very same binary module in both CPython and PyPy, which is impressive on its own :).

Finally, we wanted a real-life use case to show how to port a module to HPy and to do benchmarks. After some searching, we choose ultrajson, for the following reasons:

it is a real-world extension module which was written with performance in mind
when parsing a JSON file it does a lot of calls to the Python API to construct the various parts of the result message
it uses only a small subset of the Python API

This repo contains the HPy port of ultrajson. This commit shows an example of what the porting looks like.

ujson_hpy is also a very good example of incremental migration: so far only ujson.loads is implemented using the HPy API, while ujson.dumps is still implemented using the old C-API, and both can coexist nicely in the same compiled module.

Benchmarks

Once we have a fully working ujson_hpy module, we can finally run benchmarks! We tested several different versions of the module:

ujson: this is the vanilla implementation of ultrajson using the C-API. On PyPy this is executed by the infamous cpyext compatibility layer, so we expect it to be much slower than on CPython
ujson_hpy: our HPy port compiled to target the HPy/CPython ABI. We expect it to be as fast as ujson
ujson_hpy_universal: same as above but compiled to target the Universal HPy ABI. We expect it to be slightly slower than ujson on CPython, and much faster on PyPy.

Finally, we also ran the benchmark using the builtin json module. This is not really relevant to HPy, but it might still be an interesting as a reference data point.

The benchmark is very simple and consists of parsing a big JSON file 100 times. Here is the average time per iteration (in milliseconds) using the various versions of the module, CPython 3.7 and the latest version of the hpy PyPy branch:

	CPython	PyPy
ujson	154.32	633.97
ujson_hpy	152.19
ujson_hpy_universal	168.78	207.68
json	224.59	135.43

As expected, the benchmark proves that when targeting the HPy/CPython ABI, HPy doesn't impose any performance penalty on CPython. The universal version is ~10% slower on CPython, but gives an impressive 3x speedup on PyPy! It it worth noting that the PyPy hpy module is not fully optimized yet, and we expect to be able to reach the same performance as CPython for this particular example (or even more, thanks to our better GC).

All in all, not a bad result for two weeks of intense hacking :)

It is also worth noting than PyPy's builtin json module does really well in this benchmark, thanks to the recent optimizations that were described in an earlier blog post.

Conclusion and future directions

We think we can be very satisfied about what we have got so far. The development of HPy is quite new, but these early results seem to indicate that we are on the right track to bring Python extensions into the future.

At the moment, we can anticipate some of the next steps in the development of HPy:

Think about a proper API design: what we have done so far has been a "dumb" translation of the API we needed to run ujson. However, one of the declared goal of HPy is to improve the design of the API. There will be a trade-off between the desire of having a clean, fresh new API and the need to be not too different than the old one, to make porting easier. Finding the sweet spot will not be easy!
Implement the "debug" mode, which will help developers to find bugs such as leaking handles or using invalid handles.
Instruct Cython to emit HPy code on request.
Eventually, we will also want to try to port parts of numpy to HPy to finally solve the long-standing problem of sub-optimal numpy performance in PyPy.

Stay tuned!

PyPy v7.0.0: triple release of 2.7, 3.5 and 3.6-alpha

Antonio Cuni — Mon, 11 Feb 2019 10:55:00 GMT

The PyPy team is proud to release the version 7.0.0 of PyPy, which includes three different interpreters:

PyPy2.7, which is an interpreter supporting the syntax and the features of Python 2.7

PyPy3.5, which supports Python 3.5

PyPy3.6-alpha: this is the first official release of PyPy to support 3.6 features, although it is still considered alpha quality.

All the interpreters are based on much the same codebase, thus the triple release.
Until we can work with downstream providers to distribute builds with PyPy, we have made packages for some common packages available as wheels.
The GC hooks , which can be used to gain more insights into its performance, has been improved and it is now possible to manually manage the GC by using a combination of gc.disable and gc.collect_step. See the GC blog post.
We updated the cffi module included in PyPy to version 1.12, and the cppyy backend to 1.4. Please use these to wrap your C and C++ code, respectively, for a JIT friendly experience.
As always, this release is 100% compatible with the previous one and fixed several issues and bugs raised by the growing community of PyPy users. We strongly recommend updating.
The PyPy3.6 release and the Windows PyPy3.5 release are still not production quality so your mileage may vary. There are open issues with incomplete compatibility and c-extension support.
The utf8 branch that changes internal representation of unicode to utf8 did not make it into the release, so there is still more goodness coming. You can download the v7.0 releases here:

https://pypy.org/download.html

We would like to thank our donors for the continued support of the PyPy project. If PyPy is not quite good enough for your needs, we are available for direct consulting work.
We would also like to thank our contributors and encourage new people to join the project. PyPy has many layers and we need help with all of them: PyPy and RPython documentation improvements, tweaking popular modules to run on pypy, or general help with making RPython's JIT even better.

What is PyPy?

PyPy is a very compliant Python interpreter, almost a drop-in replacement for CPython 2.7, 3.5 and 3.6. It's fast (PyPy and CPython 2.7.x performance comparison) due to its integrated tracing JIT compiler.
We also welcome developers of other dynamic languages to see what RPython can do for them.
The PyPy release supports:

x86 machines on most common operating systems (Linux 32/64 bits, Mac OS X 64 bits, Windows 32 bits, OpenBSD, FreeBSD)

big- and little-endian variants of PPC64 running Linux,

s390x running Linux

Unfortunately at the moment of writing our ARM buildbots are out of service, so for now we are not releasing any binary for the ARM architecture.

What else is new?

PyPy 6.0 was released in April, 2018. There are many incremental improvements to RPython and PyPy, the complete listing is here.

Please update, and continue to help us make PyPy better.

Cheers, The PyPy team

PyPy for low-latency systems

Antonio Cuni — Thu, 03 Jan 2019 14:21:00 GMT

PyPy for low-latency systems

Recently I have merged the gc-disable branch, introducing a couple of features which are useful when you need to respond to certain events with the lowest possible latency. This work has been kindly sponsored by Gambit Research (which, by the way, is a very cool and geeky place where to work, in case you are interested). Note also that this is a very specialized use case, so these features might not be useful for the average PyPy user, unless you have the same problems as described here.

The PyPy VM manages memory using a generational, moving Garbage Collector. Periodically, the GC scans the whole heap to find unreachable objects and frees the corresponding memory. Although at a first look this strategy might sound expensive, in practice the total cost of memory management is far less than e.g. on CPython, which is based on reference counting. While maybe counter-intuitive, the main advantage of a non-refcount strategy is that allocation is very fast (especially compared to malloc-based allocators), and deallocation of objects which die young is basically for free. More information about the PyPy GC is available here.

As we said, the total cost of memory managment is less on PyPy than on CPython, and it's one of the reasons why PyPy is so fast. However, one big disadvantage is that while on CPython the cost of memory management is spread all over the execution of the program, on PyPy it is concentrated into GC runs, causing observable pauses which interrupt the execution of the user program.
To avoid excessively long pauses, the PyPy GC has been using an incremental strategy since 2013. The GC runs as a series of "steps", letting the user program to progress between each step.

The following chart shows the behavior of a real-world, long-running process:

The orange line shows the total memory used by the program, which increases linearly while the program progresses. Every ~5 minutes, the GC kicks in and the memory usage drops from ~5.2GB to ~2.8GB (this ratio is controlled by the PYPY_GC_MAJOR_COLLECT env variable).
The purple line shows aggregated data about the GC timing: the whole collection takes ~1400 individual steps over the course of ~1 minute: each point represent the maximum time a single step took during the past 10 seconds. Most steps take ~10-20 ms, although we see a horrible peak of ~100 ms towards the end. We have not investigated yet what it is caused by, but we suspect it is related to the deallocation of raw objects.

These multi-millesecond pauses are a problem for systems where it is important to respond to certain events with a latency which is both low and consistent. If the GC kicks in at the wrong time, it might causes unacceptable pauses during the collection cycle.

Let's look again at our real-world example. This is a system which continuously monitors an external stream; when a certain event occurs, we want to take an action. The following chart shows the maximum time it takes to complete one of such actions, aggregated every minute:

You can clearly see that the baseline response time is around ~20-30 ms. However, we can also see periodic spikes around ~50-100 ms, with peaks up to ~350-450 ms! After a bit of investigation, we concluded that most (although not all) of the spikes were caused by the GC kicking in at the wrong time.

The work I did in the gc-disable branch aims to fix this problem by introducing two new features to the gc module:

gc.disable(), which previously only inhibited the execution of finalizers without actually touching the GC, now disables the GC major collections. After a call to it, you will see the memory usage grow indefinitely.

gc.collect_step() is a new function which you can use to manually execute a single incremental GC collection step.

It is worth to specify that gc.disable() disables only the major collections, while minor collections still runs. Moreover, thanks to the JIT's virtuals, many objects with a short and predictable lifetime are not allocated at all. The end result is that most objects with short lifetime are still collected as usual, so the impact of gc.disable() on memory growth is not as bad as it could sound.

Combining these two functions, it is possible to take control of the GC to make sure it runs only when it is acceptable to do so. For an example of usage, you can look at the implementation of a custom GC inside pypytools. The peculiarity is that it also defines a "with nogc():" context manager which you can use to mark performance-critical sections where the GC is not allowed to run.

The following chart compares the behavior of the default PyPy GC and the new custom GC, after a careful placing of nogc() sections:

The yellow line is the same as before, while the purple line shows the new system: almost all spikes have gone, and the baseline performance is about 10% better. There is still one spike towards the end, but after some investigation we concluded that it was not caused by the GC.

Note that this does not mean that the whole program became magically faster: we simply moved the GC pauses in some other place which is not shown in the graph: in this specific use case this technique was useful because it allowed us to shift the GC work in places where pauses are more acceptable.

All in all, a pretty big success, I think. These functionalities are already available in the nightly builds of PyPy, and will be included in the next release: take this as a New Year present :)

Antonio Cuni and the PyPy team

Inside cpyext: Why emulating CPython C API is so Hard

Antonio Cuni — Fri, 21 Sep 2018 16:32:00 GMT

cpyext is PyPy's subsystem which provides a compatibility layer to compile and run CPython C extensions inside PyPy. Often people ask why a particular C extension doesn't work or is very slow on PyPy. Usually it is hard to answer without going into technical details. The goal of this blog post is to explain some of these technical details, so that we can simply link here instead of explaining again and again :).
From a 10.000 foot view, cpyext is PyPy's version of "Python.h". Every time you compile an extension which uses that header file, you are using cpyext. This includes extension explicitly written in C (such as numpy) and extensions which are generated from other compilers/preprocessors (e.g. Cython).
At the time of writing, the current status is that most C extensions "just work". Generally speaking, you can simply pip install them, provided they use the public, official C API instead of poking at private implementation details. However, the performance of cpyext is generally poor. A Python program which makes heavy use of cpyext extensions is likely to be slower on PyPy than on CPython.
Note: in this blog post we are talking about Python 2.7 because it is still the default version of PyPy: however most of the implementation of cpyext is shared with PyPy3, so everything applies to that as well.

C API Overview

In CPython, which is written in C, Python objects are represented as PyObject*, i.e. (mostly) opaque pointers to some common "base struct".
CPython uses a very simple memory management scheme: when you create an object, you allocate a block of memory of the appropriate size on the heap. Depending on the details, you might end up calling different allocators, but for the sake of simplicity, you can think that this ends up being a call to malloc(). The resulting block of memory is initialized and casted to to PyObject*: this address never changes during the object lifetime, and the C code can freely pass it around, store it inside containers, retrieve it later, etc.
Memory is managed using reference counting. When you create a new reference to an object, or you discard a reference you own, you have to increment or decrement the reference counter accordingly. When the reference counter goes to 0, it means that the object is no longer used and can safely be destroyed. Again, we can simplify and say that this results in a call to free(), which finally releases the memory which was allocated by malloc().
Generally speaking, the only way to operate on a PyObject* is to call the appropriate API functions. For example, to convert a given PyObject* to a C integer, you can use PyInt_AsLong(); to add two objects together, you can call PyNumber_Add().
Internally, PyPy uses a similar approach. All Python objects are subclasses of the RPython W_Root class, and they are operated by calling methods on the space singleton, which represents the interpreter.
At first, it looks very easy to write a compatibility layer: just make PyObject* an alias for W_Root, and write simple RPython functions (which will be translated to C by the RPython compiler) which call the space accordingly:

def PyInt_AsLong(space, o):
    return space.int_w(o)

def PyNumber_Add(space, o1, o2):
    return space.add(o1, o2)

Actually, the code above is not too far from the real implementation. However, there are tons of gory details which make it much harder than it looks, and much slower unless you pay a lot of attention to performance.

The PyPy GC

To understand some of cpyext challenges, you need to have at least a rough idea of how the PyPy GC works.
Contrarily to the popular belief, the "Garbage Collector" is not only about collecting garbage: instead, it is generally responsible for all memory management, including allocation and deallocation.
Whereas CPython uses a combination of malloc/free/refcounting to manage memory, the PyPy GC uses a completely different approach. It is designed assuming that a dynamic language like Python behaves the following way:

You create, either directly or indirectly, lots of objects.

Most of these objects are temporary and very short-lived. Think e.g. of doing a + b + c: you need to allocate an object to hold the temporary result of a + b, then it dies very quickly because you no longer need it when you do the final + c part.

Only small fraction of the objects survive and stay around for a while.

So, the strategy is: make allocation as fast as possible; make deallocation of short-lived objects as fast as possible; find a way to handle the remaining small set of objects which actually survive long enough to be important.
This is done using a Generational GC: the basic idea is the following:

We have a nursery, where we allocate "young objects" very quickly.

When the nursery is full, we start what we call a "minor collection".

We do a quick scan to determine the small set of objects which survived so far

We move these objects out of the nursery, and we place them in the area of memory which contains the "old objects". Since the address of the objects changes, we fix all the references to them accordingly.

now the nursery contains only objects which "died young". We can discard all of them very quickly, reset the nursery, and use the same area of memory to allocate new objects from now.

In practice, this scheme works very well and it is one of the reasons why PyPy is much faster than CPython. However, careful readers have surely noticed that this is a problem for cpyext. On one hand, we have PyPy objects which can potentially move and change their underlying memory address; on the other hand, we need a way to represent them as fixed-address PyObject* when we pass them to C extensions. We surely need a way to handle that.

`PyObject*` in PyPy

Another challenge is that sometimes, PyObject* structs are not completely opaque: there are parts of the public API which expose to the user specific fields of some concrete C struct. For example the definition of PyTypeObject which exposes many of the tp_* slots to the user. Since the low-level layout of PyPy W_Root objects is completely different than the one used by CPython, we cannot simply pass RPython objects to C; we need a way to handle the difference.
So, we have two issues so far: objects can move, and incompatible low-level layouts. cpyext solves both by decoupling the RPython and the C representations. We have two "views" of the same entity, depending on whether we are in the PyPy world (the movable W_Root subclass) or in the C world (the non-movable PyObject*).
PyObject* are created lazily, only when they are actually needed. The vast majority of PyPy objects are never passed to any C extension, so we don't pay any penalty in that case. However, the first time we pass a W_Root to C, we allocate and initialize its PyObject* counterpart.
The same idea applies also to objects which are created in C, e.g. by calling PyObject_New(). At first, only the PyObject* exists and it is exclusively managed by reference counting. As soon as we pass it to the PyPy world (e.g. as a return value of a function call), we create its W_Root counterpart, which is managed by the GC as usual.
Here we start to see why calling cpyext modules is more costly in PyPy than in CPython. We need to pay some penalty for all the conversions between W_Root and PyObject*.
Moreover, the first time we pass a W_Root to C we also need to allocate the memory for the PyObject* using a slowish "CPython-style" memory allocator. In practice, for all the objects which are passed to C we pay more or less the same costs as CPython, thus effectively "undoing" the speedup guaranteed by PyPy's Generational GC under normal circumstances.

Maintaining the link between `W_Root` and `PyObject*`

We now need a way to convert between W_Root and PyObject* and vice-versa; also, we need to to ensure that the lifetime of the two entities are in sync. In particular:

as long as the W_Root is kept alive by the GC, we want the PyObject* to live even if its refcount drops to 0;

as long as the PyObject* has a refcount greater than 0, we want to make sure that the GC does not collect the W_Root.

The PyObject* ⇨ W_Root link is maintained by the special field ob_pypy_link which is added to all PyObject*. On a 64 bit machine this means that all PyObject* have 8 bytes of overhead, but then the conversion is very quick, just reading the field.
For the other direction, we generally don't want to do the same: the assumption is that the vast majority of W_Root objects will never be passed to C, and adding an overhead of 8 bytes to all of them is a waste. Instead, in the general case the link is maintained by using a dictionary, where W_Root are the keys and PyObject* the values.
However, for a few selected W_Root subclasses we do maintain a direct link using the special _cpy_ref field to improve performance. In particular, we use it for W_TypeObject (which is big anyway, so a 8 bytes overhead is negligible) and W_NoneObject. None is passed around very often, so we want to ensure that the conversion to PyObject* is very fast. Moreover it's a singleton, so the 8 bytes overhead is negligible as well.
This means that in theory, passing an arbitrary Python object to C is potentially costly, because it involves doing a dictionary lookup. We assume that this cost will eventually show up in the profiler: however, at the time of writing there are other parts of cpyext which are even more costly (as we will show later), so the cost of the dict lookup is never evident in the profiler.

Crossing the border between RPython and C

There are two other things we need to care about whenever we cross the border between RPython and C, and vice-versa: exception handling and the GIL.
In the C API, exceptions are raised by calling PyErr_SetString() (or one of many other functions which have a similar effect), which basically works by creating an exception value and storing it in some global variable. The function then signals that an exception has occurred by returning an error value, usually NULL.
On the other hand, in the PyPy interpreter, exceptions are propagated by raising the RPython-level OperationError exception, which wraps the actual app-level exception values. To harmonize the two worlds, whenever we return from C to RPython, we need to check whether a C API exception was raised and if so turn it into an OperationError.
We won't dig into details of how the GIL is handled in cpyext. For the purpose of this post, it is enough to know that whenever we enter C land, we store the current thread id into a global variable which is accessible also from C; conversely, whenever we go back from RPython to C, we restore this value to 0.
Similarly, we need to do the inverse operations whenever you need to cross the border between C and RPython, e.g. by calling a Python callback from C code.
All this complexity is automatically handled by the RPython function generic_cpy_call. If you look at the code you see that it takes care of 4 things:

Handling the GIL as explained above.

Handling exceptions, if they are raised.

Converting arguments from W_Root to PyObject*.

Converting the return value from PyObject* to W_Root.

So, we can see that calling C from RPython introduce some overhead. Can we measure it?
Assuming that the conversion between W_Root and PyObject* has a reasonable cost (as explained by the previous section), the overhead introduced by a single border-cross is still acceptable, especially if the callee is doing some non-negligible amount of work.
However this is not always the case. There are basically three problems that make (or used to make) cpyext super slow:

Paying the border-crossing cost for trivial operations which are called very often, such as Py_INCREF.

Crossing the border back and forth many times, even if it's not strictly needed.

Paying an excessive cost for argument and return value conversions.

The next sections explain in more detail each of these problems.

Avoiding unnecessary roundtrips

Prior to the 2017 Cape Town Sprint, cpyext was horribly slow, and we were well aware of it: the main reason was that we never really paid too much attention to performance. As explained in the blog post, emulating all the CPython quirks is basically a nightmare, so better to concentrate on correctness first.
However, we didn't really know why it was so slow. We had theories and assumptions, usually pointing at the cost of conversions between W_Root and PyObject*, but we never actually measured it.
So, we decided to write a set of cpyext microbenchmarks to measure the performance of various operations. The result was somewhat surprising: the theory suggests that when you do a cpyext C call, you should pay the border-crossing costs only once, but what the profiler told us was that we were paying the cost of generic_cpy_call several times more than what we expected.
After a bit of investigation, we discovered this was ultimately caused by our "correctness-first" approach. For simplicity of development and testing, when we started cpyext we wrote everything in RPython: thus, every single API call made from C (like the omnipresent PyArg_ParseTuple(), PyInt_AsLong(), etc.) had to cross back the C-to-RPython border. This was especially daunting for very simple and frequent operations like Py_INCREF and Py_DECREF, which CPython implements as a single assembly instruction!
Another source of slow down was the implementation of PyTypeObject slots. At the C level, these are function pointers which the interpreter calls to do certain operations, e.g. tp_new to allocate a new instance of that type.
As usual, we have some magic to implement slots in RPython; in particular, _make_wrapper does the opposite of generic_cpy_call: it takes a RPython function and wraps it into a C function which can be safely called from C, handling the GIL, exceptions and argument conversions automatically.
This was very handy during the development of cpyext, but it might result in some bad nonsense; consider what happens when you call the following C function:

static PyObject* foo(PyObject* self, PyObject* args)
{
    PyObject* result = PyInt_FromLong(1234);
    return result;
}

you are in RPython and do a cpyext call to foo: RPython-to-C;
foo calls PyInt_FromLong(1234), which is implemented in RPython: C-to-RPython;
the implementation of PyInt_FromLong indirectly calls PyIntType.tp_new, which is a C function pointer: RPython-to-C;
however, tp_new is just a wrapper around an RPython function, created by _make_wrapper: C-to-RPython;
finally, we create our RPython W_IntObject(1234); at some point during the RPython-to-C crossing, its PyObject* equivalent is created;
after many layers of wrappers, we are again in foo: after we do return result, during the C-to-RPython step we convert it from PyObject* to W_IntObject(1234).

Phew! After we realized this, it was not so surprising that cpyext was very slow :). And this was a simplified example, since we are not passing a PyObject* to the API call. When we do, we need to convert it back and forth at every step. Actually, I am not even sure that what I described was the exact sequence of steps which used to happen, but you get the general idea.
The solution is simple: rewrite as much as we can in C instead of RPython, to avoid unnecessary roundtrips. This was the topic of most of the Cape Town sprint and resulted in the cpyext-avoid-roundtrip branch, which was eventually merged.
Of course, it is not possible to move everything to C: there are still operations which need to be implemented in RPython. For example, think of PyList_Append: the logic to append an item to a list is complex and involves list strategies, so we cannot replicate it in C. However, we discovered that a large subset of the C API can benefit from this.
Moreover, the C API is huge. While we invented this new way of writing cpyext code, we still need to convert many of the functions to the new paradigm. Sometimes the rewrite is not automatic or straighforward. cpyext is a delicate piece of software, so it happens often that we make a mistake and end up staring at a segfault in gdb.
However, the most important takeaway is that the performance improvements we got from this optimization are impressive, as we will detail later.

Conversion costs

The other potential big source of slowdown is the conversion of arguments between W_Root and PyObject*.
As explained earlier, the first time you pass a W_Root to C, you need to allocate its PyObject* counterpart. Suppose you have a foo function defined in C, which takes a single int argument:

for i in range(N):
    foo(i)

To run this code, you need to create a different PyObject* for each value of i: if implemented naively, it means calling N times malloc() and free(), which kills performance.
CPython has the very same problem, which is solved by using a free list to allocate ints. So, what we did was to simply steal the code from CPython and do the exact same thing. This was also done in the cpyext-avoid-roundtrip branch, and the benchmarks show that it worked perfectly.
Every type which is converted often to PyObject* must have a very fast allocator. At the moment of writing, PyPy uses free lists only for ints and tuples: one of the next steps on our TODO list is certainly to use this technique with more types, like float.
Conversely, we also need to optimize the conversion from PyObject* to W_Root: this happens when an object is originally allocated in C and returned to Python. Consider for example the following code:

import numpy as np
myarray = np.random.random(N)
for i in range(len(arr)):
    myarray[i]

At every iteration, we get an item out of the array: the return type is a an instance of numpy.float64 (a numpy scalar), i.e. a PyObject'*: this is something which is implemented by numpy entirely in C, so completely opaque to cpyext. We don't have any control on how it is allocated, managed, etc., and we can assume that allocation costs are the same as on CPython.
As soon as we return these PyObject* to Python, we need to allocate their W_Root equivalent. If you do it in a small loop like in the example above, you end up allocating all these W_Root inside the nursery, which is a good thing since allocation is super fast (see the section above about the PyPy GC).
However, we also need to keep track of the W_Root to PyObject* link. Currently, we do this by putting all of them in a dictionary, but it is very inefficient, especially because most of these objects die young and thus it is wasted work to do that for them. Currently, this is one of the biggest unresolved problem in cpyext, and it is what causes the two microbenchmarks allocate_int and allocate_tuple to be very slow.
We are well aware of the problem, and we have a plan for how to fix it. The explanation is too technical for the scope of this blog post as it requires a deep knowledge of the GC internals to be understood, but the details are here.

C API quirks

Finally, there is another source of slowdown which is beyond our control. Some parts of the CPython C API are badly designed and expose some of the implementation details of CPython.
The major example is reference counting. The Py_INCREF / Py_DECREF API is designed in such a way which forces other implementation to emulate refcounting even in presence of other GC management schemes, as explained above.
Another example is borrowed references. There are API functions which do not incref an object before returning it, e.g. PyList_GetItem(). This is done for performance reasons because we can avoid a whole incref/decref pair, if the caller needs to handle the returned item only temporarily: the item is kept alive because it is in the list anyway.
For PyPy, this is a challenge: thanks to list strategies, lists are often represented in a compact way. For example, a list containing only integers is stored as a C array of long. How to implement PyList_GetItem? We cannot simply create a PyObject* on the fly, because the caller will never decref it and it will result in a memory leak.
The current solution is very inefficient. The first time we do a PyList_GetItem, we convert the whole list to a list of PyObject*. This is bad in two ways: the first is that we potentially pay a lot of unneeded conversion cost in case we will never access the other items of the list. The second is that by doing that we lose all the performance benefit granted by the original list strategy, making it slower for the rest of the pure-python code which will manipulate the list later.
PyList_GetItem is an example of a bad API because it assumes that the list is implemented as an array of PyObject*: after all, in order to return a borrowed reference, we need a reference to borrow, don't we?
Fortunately, (some) CPython developers are aware of these problems, and there is an ongoing project to design a better C API which aims to fix exactly this kind of problem.
Nonetheless, in the meantime we still need to implement the current half-broken APIs. There is no easy solution for that, and it is likely that we will always need to pay some performance penalty in order to implement them correctly.
However, what we could potentially do is to provide alternative functions which do the same job but are more PyPy friendly: for example, we could think of implementing PyList_GetItemNonBorrowed or something like that: then, C extensions could choose to use it (possibly hidden inside some macro and #ifdef) if they want to be fast on PyPy.

Current performance

During the whole blog post we claimed cpyext is slow. How slow it is, exactly?
We decided to concentrate on microbenchmarks for now. It should be evident by now there are simply too many issues which can slow down a cpyext program, and microbenchmarks help us to concentrate on one (or few) at a time.
The microbenchmarks measure very simple things, like calling functions and methods with the various calling conventions (no arguments, one arguments, multiple arguments); passing various types as arguments (to measure conversion costs); allocating objects from C, and so on.
Here are the results from the old PyPy 5.8 relative and normalized to CPython 2.7, the lower the better:

PyPy was horribly slow everywhere, ranging from 2.5x to 10x slower. It is particularly interesting to compare simple.noargs, which measures the cost of calling an empty function with no arguments, and simple.onearg(i), which measures the cost calling an empty function passing an integer argument: the latter is ~2x slower than the former, indicating that the conversion cost of integers is huge.
PyPy 5.8 was the last release before the famous Cape Town sprint, when we started to look at cpyext performance seriously. Here are the performance data for PyPy 6.0, the latest release at the time of writing:

The results are amazing! PyPy is now massively faster than before, and for most benchmarks it is even faster than CPython: yes, you read it correctly: PyPy is faster than CPython at doing CPython's job, even considering all the extra work it has to do to emulate the C API. This happens thanks to the JIT, which produces speedups high enough to counterbalance the slowdown caused by cpyext.
There are two microbenchmarks which are still slower though: allocate_int and allocate_tuple, for the reasons explained in the section about Conversion costs.

Next steps

Despite the spectacular results we got so far, cpyext is still slow enough to kill performance in most real-world code which uses C extensions extensively (e.g., the omnipresent numpy).
Our current approach is something along these lines:

run a real-world small benchmark which exercises cpyext

measure and find the major bottleneck

write a corresponding microbenchmark

optimize it

repeat

On one hand, this is a daunting task because the C API is huge and we need to tackle functions one by one. On the other hand, not all the functions are equally important, and is is enough to optimize a relatively small subset to improve many different use cases.
Where a year ago we announced we have a working answer to run c-extension in PyPy, we now have a clear picture of what are the performance bottlenecks, and we have developed some technical solutions to fix them. It is "only" a matter of tackling them, one by one. It is worth noting that most of the work was done during two sprints, for a total 2-3 person-months of work.
We think this work is important for the Python ecosystem. PyPy has established a baseline for performance in pure python code, providing an answer for the "Python is slow" detractors. The techniques used to make cpyext performant will let PyPy become an alternative for people who mix C extensions with Python, which, it turns out, is just about everyone, in particular those using the various scientific libraries. Today, many developers are forced to seek performance by converting code from Python to a lower language. We feel there is no reason to do this, but in order to prove it we must be able to run both their python and their C extensions performantly, then we can begin to educate them how to write JIT-friendly code in the first place.
We envision a future in which you can run arbitrary Python programs on PyPy, with the JIT speeding up the pure Python parts and the C parts running as fast as today: the best of both worlds!

How to ignore the annoying Cython warnings in PyPy 6.0

Antonio Cuni — Fri, 27 Apr 2018 14:10:00 GMT

If you install any Cython-based module in PyPy 6.0.0, it is very likely that you get a warning like this:

>>>> import numpy
/data/extra/pypy/6.0.0/site-packages/numpy/random/__init__.py:99: UserWarning: __builtin__.type size changed, may indicate binary incompatibility. Expected 888, got 408
  from .mtrand import *

The TL;DR version is: the warning is a false alarm, and you can hide it by doing:

$ pypy -m pip install pypy-fix-cython-warning

The package does not contain any module, only a .pth file which installs a warning filter at startup.

Technical details

This happens because whenever Cython compiles a pyx file, it generates C code which does a sanity check on the C size of PyType_Type. PyPy versions up to 5.10 are buggy and report the incorrect size, so Cython includes a workaround to compare it with the incorrect value, when on PyPy.

PyPy 6 fixed the bug and now PyType_Type reports the correct size; however, Cython still tries to compare it with the old, buggy value, so it (wrongly) emits the warning.

Cython 0.28.2 includes a fix for it, so that C files generated by it no longer emit the warning. However, most packages are distributed with pre-cythonized C files. For example, numpy-1.14.2.zip include C files which were generated by Cython 0.26.1: if you compile it you still get the warning, even if you locally installed a newer version of Cython.

There is not much that we can do on the PyPy side, apart for waiting for all the Cython-based packages to do a new release which include C files generated by a newer Cython. In the mean time, installing this module will silence the warning.

How to make your code 80 times faster

Antonio Cuni — Mon, 30 Oct 2017 10:15:00 GMT

I often hear people who are happy because PyPy makes their code 2 times faster or so. Here is a short personal story which shows PyPy can go well beyond that.

DISCLAIMER: this is not a silver bullet or a general recipe: it worked in this particular case, it might not work so well in other cases. But I think it is still an interesting technique. Moreover, the various steps and implementations are showed in the same order as I tried them during the development, so it is a real-life example of how to proceed when optimizing for PyPy.

Some months ago I played a bit with evolutionary algorithms: the ambitious plan was to automatically evolve a logic which could control a (simulated) quadcopter, i.e. a PID controller (spoiler: it doesn't fly).

The idea is to have an initial population of random creatures: at each generation, the ones with the best fitness survive and reproduce with small, random variations.

However, for the scope of this post, the actual task at hand is not so important, so let's jump straight to the code. To drive the quadcopter, a Creature has a run_step method which runs at each delta_t (full code):

class Creature(object):
    INPUTS = 2  # z_setpoint, current z position
    OUTPUTS = 1 # PWM for all 4 motors
    STATE_VARS = 1
    ...

    def run_step(self, inputs):
        # state: [state_vars ... inputs]
        # out_values: [state_vars, ... outputs]
        self.state[self.STATE_VARS:] = inputs
        out_values = np.dot(self.matrix, self.state) + self.constant
        self.state[:self.STATE_VARS] = out_values[:self.STATE_VARS]
        outputs = out_values[self.STATE_VARS:]
        return outputs

inputs is a numpy array containing the desired setpoint and the current position on the Z axis;
outputs is a numpy array containing the thrust to give to the motors. To start easy, all the 4 motors are constrained to have the same thrust, so that the quadcopter only travels up and down the Z axis;
self.state contains arbitrary values of unknown size which are passed from one step to the next;
self.matrix and self.constant contains the actual logic. By putting the "right" values there, in theory we could get a perfectly tuned PID controller. These are randomly mutated between generations.

run_step is called at 100Hz (in the virtual time frame of the simulation). At each generation, we test 500 creatures for a total of 12 virtual seconds each. So, we have a total of 600,000 executions of run_step at each generation.

At first, I simply tried to run this code on CPython; here is the result:

$ python -m ev.main
Generation   1: ... [population = 500]  [12.06 secs]
Generation   2: ... [population = 500]  [6.13 secs]
Generation   3: ... [population = 500]  [6.11 secs]
Generation   4: ... [population = 500]  [6.09 secs]
Generation   5: ... [population = 500]  [6.18 secs]
Generation   6: ... [population = 500]  [6.26 secs]

Which means ~6.15 seconds/generation, excluding the first.

Then I tried with PyPy 5.9:

$ pypy -m ev.main
Generation   1: ... [population = 500]  [63.90 secs]
Generation   2: ... [population = 500]  [33.92 secs]
Generation   3: ... [population = 500]  [34.21 secs]
Generation   4: ... [population = 500]  [33.75 secs]

Ouch! We are ~5.5x slower than CPython. This was kind of expected: numpy is based on cpyext, which is infamously slow. (Actually, we are working on that and on the cpyext-avoid-roundtrip branch we are already faster than CPython, but this will be the subject of another blog post.)

So, let's try to avoid cpyext. The first obvious step is to use numpypy instead of numpy (actually, there is a hack to use just the micronumpy part). Let's see if the speed improves:

$ pypy -m ev.main   # using numpypy
Generation   1: ... [population = 500]  [5.60 secs]
Generation   2: ... [population = 500]  [2.90 secs]
Generation   3: ... [population = 500]  [2.78 secs]
Generation   4: ... [population = 500]  [2.69 secs]
Generation   5: ... [population = 500]  [2.72 secs]
Generation   6: ... [population = 500]  [2.73 secs]

So, ~2.7 seconds on average: this is 12x faster than PyPy+numpy, and more than 2x faster than the original CPython. At this point, most people would be happy and go tweeting how PyPy is great.

In general, when talking of CPython vs PyPy, I am rarely satisfied with a 2x speedup: I know that PyPy can do much better than this, especially if you write code which is specifically optimized for the JIT. For a real-life example, have a look at capnpy benchmarks, in which the PyPy version is ~15x faster than the heavily optimized CPython+Cython version (both have been written by me, and I tried hard to write the fastest code for both implementations).

So, let's try to do better. As usual, the first thing to do is to profile and see where we spend most of the time. Here is the vmprof profile. We spend a lot of time inside the internals of numpypy, and allocating tons of temporary arrays to store the results of the various operations.

Also, let's look at the jit traces and search for the function run: this is loop in which we spend most of the time, and it is composed of 1796 operations. The operations emitted for the line np.dot(...) + self.constant are listed between lines 1217 and 1456. Here is the excerpt which calls np.dot(...); most of the ops are cheap, but at line 1232 we see a call to the RPython function descr_dot; by looking at the implementation we see that it creates a new W_NDimArray to store the result, which means it has to do a malloc():

The implementation of the + self.constant part is also interesting: contrary the former, the call to W_NDimArray.descr_add has been inlined by the JIT, so we have a better picture of what's happening; in particular, we can see the call to __0_alloc_with_del____ which allocates the W_NDimArray for the result, and the raw_malloc which allocates the actual array. Then we have a long list of 149 simple operations which set the fields of the resulting array, construct an iterator, and finally do a call_assembler: this is the actual logic to do the addition, which was JITtted independently; call_assembler is one of the operations to do JIT-to-JIT calls:

All of this is very suboptimal: in this particular case, we know that the shape of self.matrix is always (3, 2): so, we are doing an incredible amount of work, including calling malloc() twice for the temporary arrays, just to call two functions which ultimately do a total of 6 multiplications and 6 additions. Note also that this is not a fault of the JIT: CPython+numpy has to do the same amount of work, just hidden inside C calls.

One possible solution to this nonsense is a well known compiler optimization: loop unrolling. From the compiler point of view, unrolling the loop is always risky because if the matrix is too big you might end up emitting a huge blob of code, possibly uselss if the shape of the matrices change frequently: this is the main reason why the PyPy JIT does not even try to do it in this case.

However, we know that the matrix is small, and always of the same shape. So, let's unroll the loop manually:

class SpecializedCreature(Creature):

    def __init__(self, *args, **kwargs):
        Creature.__init__(self, *args, **kwargs)
        # store the data in a plain Python list
        self.data = list(self.matrix.ravel()) + list(self.constant)
        self.data_state = [0.0]
        assert self.matrix.shape == (2, 3)
        assert len(self.data) == 8

    def run_step(self, inputs):
        # state: [state_vars ... inputs]
        # out_values: [state_vars, ... outputs]
        k0, k1, k2, q0, q1, q2, c0, c1 = self.data
        s0 = self.data_state[0]
        z_sp, z = inputs
        #
        # compute the output
        out0 = s0*k0 + z_sp*k1 + z*k2 + c0
        out1 = s0*q0 + z_sp*q1 + z*q2 + c1
        #
        self.data_state[0] = out0
        outputs = [out1]
        return outputs

In the actual code there is also a sanity check which asserts that the computed output is the very same as the one returned by Creature.run_step.

So, let's try to see how it performs. First, with CPython:

$ python -m ev.main
Generation   1: ... [population = 500]  [7.61 secs]
Generation   2: ... [population = 500]  [3.96 secs]
Generation   3: ... [population = 500]  [3.79 secs]
Generation   4: ... [population = 500]  [3.74 secs]
Generation   5: ... [population = 500]  [3.84 secs]
Generation   6: ... [population = 500]  [3.69 secs]

This looks good: 60% faster than the original CPython+numpy implementation. Let's try on PyPy:

Generation   1: ... [population = 500]  [0.39 secs]
Generation   2: ... [population = 500]  [0.10 secs]
Generation   3: ... [population = 500]  [0.11 secs]
Generation   4: ... [population = 500]  [0.09 secs]
Generation   5: ... [population = 500]  [0.08 secs]
Generation   6: ... [population = 500]  [0.12 secs]
Generation   7: ... [population = 500]  [0.09 secs]
Generation   8: ... [population = 500]  [0.08 secs]
Generation   9: ... [population = 500]  [0.08 secs]
Generation  10: ... [population = 500]  [0.08 secs]
Generation  11: ... [population = 500]  [0.08 secs]
Generation  12: ... [population = 500]  [0.07 secs]
Generation  13: ... [population = 500]  [0.07 secs]
Generation  14: ... [population = 500]  [0.08 secs]
Generation  15: ... [population = 500]  [0.07 secs]

Yes, it's not an error. After a couple of generations, it stabilizes at around ~0.07-0.08 seconds per generation. This is around 80 (eighty) times faster than the original CPython+numpy implementation, and around 35-40x faster than the naive PyPy+numpypy one.

Let's look at the trace again: it no longer contains expensive calls, and certainly no more temporary malloc() s. The core of the logic is between lines 386-416, where we can see that it does fast C-level multiplications and additions: float_mul and float_add are translated straight into mulsd and addsd x86 instructions.

As I said before, this is a very particular example, and the techniques described here do not always apply: it is not realistic to expect an 80x speedup on arbitrary code, unfortunately. However, it clearly shows the potential of PyPy when it comes to high-speed computing. And most importantly, it's not a toy benchmark which was designed specifically to have good performance on PyPy: it's a real world example, albeit small.

You might be also interested in the talk I gave at last EuroPython, in which I talk about a similar topic: "The Joy of PyPy JIT: abstractions for free" (abstract, slides and video).

How to reproduce the results

$ git clone https://github.com/antocuni/evolvingcopter
$ cd evolvingcopter
$ {python,pypy} -m ev.main --no-specialized --no-numpypy
$ {python,pypy} -m ev.main --no-specialized
$ {python,pypy} -m ev.main

(Cape of) Good Hope for PyPy

Antonio Cuni — Wed, 18 Oct 2017 13:31:00 GMT

Hello from the other side of the world (for most of you)!

With the excuse of coming to PyCon ZA during the last two weeks Armin, Ronan, Antonio and sometimes Maciek had a very nice and productive sprint in Cape Town, as pictures show :). We would like to say a big thank you to Kiwi.com, which sponsored part of the travel costs via its awesome Sourcelift program to help Open Source projects.

Armin, Anto and Ronan at Cape Point

Armin, Ronan and Anto spent most of the time hacking at cpyext, our CPython C-API compatibility layer: during the last years, the focus was to make it working and compatible with CPython, in order to run existing libraries such as numpy and pandas. However, we never paid too much attention to performance, so the net result is that with the latest released version of PyPy, C extensions generally work but their speed ranges from "slow" to "horribly slow".

For example, these very simple microbenchmarks measure the speed of calling (empty) C functions, i.e. the time you spend to "cross the border" between RPython and C. (Note: this includes the time spent doing the loop in regular Python code.) These are the results on CPython, on PyPy 5.8, and on our newest in-progress version:

$ python bench.py     # CPython
noargs      : 0.41 secs
onearg(None): 0.44 secs
onearg(i)   : 0.44 secs
varargs     : 0.58 secs

$ pypy-5.8 bench.py   # PyPy 5.8
noargs      : 1.01 secs
onearg(None): 1.31 secs
onearg(i)   : 2.57 secs
varargs     : 2.79 secs

$ pypy bench.py       # cpyext-refactor-methodobject branch
noargs      : 0.17 secs
onearg(None): 0.21 secs
onearg(i)   : 0.22 secs
varargs     : 0.47 secs

So yes: before the sprint, we were ~2-6x slower than CPython. Now, we are faster than it! To reach this result, we did various improvements, such as:

teach the JIT how to look (a bit) inside the cpyext module;

write specialized code for calling METH_NOARGS, METH_O and METH_VARARGS functions; previously, we always used a very general and slow logic;

implement freelists to allocate the cpyext versions of int and tuple objects, as CPython does;

the cpyext-avoid-roundtrip branch: crossing the RPython/C border is slowish, but the real problem was (and still is for many cases) we often cross it many times for no good reason. So, depending on the actual API call, you might end up in the C land, which calls back into the RPython land, which goes to C, etc. etc. (ad libitum).

The branch tries to fix such nonsense: so far, we fixed only some cases, which are enough to speed up the benchmarks shown above. But most importantly, we now have a clear path and an actual plan to improve cpyext more and more. Ideally, we would like to reach a point in which cpyext-intensive programs run at worst at the same speed of CPython.

The other big topic of the sprint was Armin and Maciej doing a lot of work on the unicode-utf8 branch: the goal of the branch is to always use UTF-8 as the internal representation of unicode strings. The advantages are various:

decoding a UTF-8 stream is super fast, as you just need to check that the stream is valid;

encoding to UTF-8 is almost a no-op;

UTF-8 is always more compact representation than the currently used UCS-4. It's also almost always more compact than CPython 3.5 latin1/UCS2/UCS4 combo;

smaller representation means everything becomes quite a bit faster due to lower cache pressure.

Before you ask: yes, this branch contains special logic to ensure that random access of single unicode chars is still O(1), as it is on both CPython and the current PyPy.
We also plan to improve the speed of decoding even more by using modern processor features, like SSE and AVX. Preliminary results show that decoding can be done 100x faster than the current setup.

In summary, this was a long and profitable sprint, in which we achieved lots of interesting results. However, what we liked even more was the privilege of doing commits from awesome places such as the top of Table Mountain:

Our sprint venue today #pypy pic.twitter.com/o38IfTYmAV
— Ronan Lamy (@ronanlamy) 4 ottobre 2017

The panorama we looked at instead of staring at cpyext code

Binary wheels for PyPy

Antonio Cuni — Wed, 26 Jul 2017 16:53:00 GMT

Hi,

this is a short blog post, just to announce the existence of this Github repository, which contains binary PyPy wheels for some selected packages. The availability of binary wheels means that you can install the packages much more quickly, without having to wait for compilation.

At the moment of writing, these packages are available:

numpy
scipy
pandas
psutil
netifaces

For now, we provide only wheels built on Ubuntu, compiled for PyPy 5.8.
In particular, it is worth noting that they are not manylinux1 wheels, which means they could not work on other Linux distributions. For more information, see the explanation in the README of the above repo.

Moreover, the existence of the wheels does not guarantee that they work correctly 100% of the time. they still depend on cpyext, our C-API emulation layer, which is still work-in-progress, although it has become better and better during the last months. Again, the wheels are there only to save compilation time.

To install a package from the wheel repository, you can invoke pip like this:

$ pip install --extra-index https://antocuni.github.io/pypy-wheels/ubuntu numpy

Happy installing!

PyPy (Posts by Antonio Cuni)

#pypy IRC moves to Libera.Chat

New HPy blog

HPy kick-off sprint report

What is HPy?

Target ABI

A simple example

Sprint report and current status

Benchmarks

Conclusion and future directions

PyPy v7.0.0: triple release of 2.7, 3.5 and 3.6-alpha

What is PyPy?

What else is new?

PyPy for low-latency systems

PyPy for low-latency systems

Inside cpyext: Why emulating CPython C API is so Hard

C API Overview

The PyPy GC

PyObject* in PyPy

Maintaining the link between W_Root and PyObject*

Crossing the border between RPython and C

Avoiding unnecessary roundtrips

Conversion costs

C API quirks

Current performance

Next steps

How to ignore the annoying Cython warnings in PyPy 6.0

Technical details

How to make your code 80 times faster

How to reproduce the results

(Cape of) Good Hope for PyPy

Binary wheels for PyPy

`PyObject*` in PyPy

Maintaining the link between `W_Root` and `PyObject*`