Fast

A long-standing goal of the Crack language has been to produce a system that is fast.  Specifically, we wanted to minimize the code-test cycle.  The language has thus far fallen short in this respect.  Startup times for anything less than very small programs have tended to be over five seconds.

We were never able to obtain any solid conclusions from profiling the code, but we knew there was a lot to be gained from caching.  So over two years ago we started work on caching.  Needless to say, this was a harder problem than we imagined.

Caching is simply storing the artifacts of a successful compile so that we don't have to produce them again.  In theory, the work involved in fully parsing and validation the source as well as generation and optimization of LLVM IR code is greater than the effort involved in simple deserialization of these structures.

Getting it to work right has been very difficult, largely because of a number of shortcuts and bad design choices that I made while producing the original compiler.  My first reaction to this was, of course, to try to adapt caching to the existing code.  However, when these attempts failed I did finally do the right thing and make the big changes to the design that were necessary to get caching to work.

And now, finally, it does.  I can run code, make changes, and run again with the most complicated programs that I have and everything just works.  The modules for the changed code and all of their dependencies get rebuilt, everything else gets loaded from the cache.  Sweet!  (I'm sure there are still some bugs in there somewhere...)

So trying this out against the most complicated program that we have -- the "wurld" game engine demo in 'examples' -- without any cached files, takes 12 seconds to build on my laptop.  After caching, it takes ... 9 seconds.

9 seconds?  Really?  A 25% reduction after all of that work?  That's depressing.  So I did some more benchmarking.

One of the results of the restructuring work required for caching is that all of the jitting of a cached program now happens at the end, just before the program is run and after all of the modules are loaded from the cache.  Originally we jitted every module when we finished loading it.  The new structure makes it much easier to see how much time we're spending on jitting versus everything else.

It turns out, we were spending most of that time, 7-8 seconds of it, on jitting.

We were storing the addresses of all of our functions before running everything.  We need function addresses in order to generate source file and function names when reporting stack traces.  But in the course of looking all of these addresses up, we were also generating all of the functions in the program, whether we needed them or not.

LLVM's ExecutionEngine lets you watch for notifications for a number of different internal events.  In particular, it lets you get a notification when a function gets jitted. So I replaced the registration loop with a notification callback.  Now instead of function registration driving jitting, jitting drives function registration, and only for the functions that are used by the program.  This got us down to about 5 seconds total runtime.

The ExecutionEngine also lets you enable "lazy jitting."  By default, when you request a function address, that function and all of its dependencies get jitted.  By contrast, lazy jitting defers the jitting of a function until it is actually called.  Enabling this feature brought the total run time down to under 2.5 seconds, because the hacked version of "wurld" I was using for benchmarking was loading everything and doing initialization but then immediately terminating, so much of the code is never called anyway.

I'm not sure if there's anything more we can do to improve on this, but 2.5 seconds is well in the realm of tolerable.

It's somewhat embarrassing that the huge effort of caching yielded only a small amount of the initial gains while an hour or two of investigation and simple tweaking had such a big impact.  But on the other hand, the simple tweaking couldn't have worked without some of the changes we put in for caching.  And as it turns out, post-tweaks, caching ends up saving us 60% of the startup time, which is huge.

Going forward, after we release 1.0, I still want to start to experiment with alternate backends.  libjit, in particular, looks like a promising alternative to LLVM for our JIT backend.  But as it stands, in terms of performance, I think we're in pretty good shape for a 1.0 release.

Crack 0.9 Released

We are pleased to announce the release of Crack 0.9.  The only major user-visible change in this release is the introduction of support for threads.  We've also done a great deal of work towards caching, but unfortunately this feature is still not ready for general use.

Other, lesser features include an IRC module (crack.net.irc) and a protobuf module (crack.protobuf).

Please note that the new tarball is not available on the googlecode download page: new uploads have been disabled by Google.  Our downloads are now available at http://crack-lang.org/downloads/

Enjoy!

Threads

I'm happy to say that Crack now supports threading.  Thread support required two basic ingredients: a pthread extension (wrapped in a high-level Crack module) and the atomic_int type for reference counting.

Normal integer operations are not sufficient to support reference counting in a multithreaded environment.  Compilers and CPUs perform all kinds of tricks in order to improve performance, and this can result in two different threads seeing two different values for the same variable at any given point in time.  Atomic operations solve this problem.  They prevent the compiler from doing any special tricks to elide or reorder the operation, and require the CPU to lock the bus and synchronize the operation across all views of memory.

To expose atomic operations in the language, I've created an atomic_int type which supports a limited set of integer operations (mainly += and -=) and synchronized conversion to plain integer types.  The reference count of crack.lang.Object is now of type atomic_int, giving us atomic reference counts.

For actual thread creation and more general synchronization, I've implemented the crack.threads module.  This provides Crack classes encapsulating functions from a new pthread extension, so we have Thread, Mutex and Condition classes.  There is also a higher level Queue generic which provides one of the most common modes of inter-thread communication.

The new threading code is currently available only from the repository, but we're about due for a release, so it should be available soon in version 0.9 of the crack SDK.