April 2011 - Tachyon-list - mailman.iro.umontreal.ca

Except from Compiler Construction by Nicolas Wirth related to a discussion in the last meeting
by Erick Lavoie 29 Apr '11

29 Apr '11

"Lastly, we must also point out that the more optimizations with sizeable effects that can be incorporated in a compiler, the poorer its original version must have been. In this connection, the cumbersome structures of many commercial compilers, whose origin is difficult to fathom, lead to surprisingly poor initial performance, which makes optimizing features seem absolutely indispensible." [1] The "General Considerations" on page 100 is worth reading for its taxonomy of optimizations and point of view on optimizations. Erick [1] Compiler Construction, Nicolas Wirth, http://www-old.oberon.ethz.ch/WirthPubl/CBEAll.pdf

1 0

Nicolas Wirth's Theory and Techniques of Compiler Construction book
by Erick Lavoie 29 Apr '11

29 Apr '11

His book is available for free from a quick review could be used as an alternative to the Appel book. http://www-old.oberon.ethz.ch/WirthPubl/CBEAll.pdf Erick // <https://docs.google.com/viewer?url=http%3A%2F%2Fwww-old.oberon.ethz.ch%2FWi…>

1 0

Food for thought: Nicolas Wirth's work
by Erick Lavoie 29 Apr '11

29 Apr '11

Niklaus Wirth is widely known as the creator of several programming languages. What is less well known is that he also personally wrote several compilers, including the first single-pass compiler for Modula-2 that later evolved into the initial compiler for Oberon5. These compilers distinguished themselves by their particularly simple design -- they didn't aspire to tickle the last possible bit of achievable performance out of a piece of code, but aimed to provide adequate code quality at a price-point of reasonable compilation speed and compiler size. At the time, this was in stark contrast to almost all other research in compilers, which generally had been characterized by an enormous and ever-increasing complexity of optimizations to the detriment of compilation speed and overall compiler size. In order to find the optimal cost/benefit ratio, Wirth used a highly intuitive metric, the origin of which is unknown to me but that may very well be Wirth's own invention. He used the compiler's self-compilation speed as a measure of the compiler's quality. Considering that Wirth's compilers were written in the languages they compiled, and that compilers are substantial and non-trivial pieces of software in their own right, this introduced a highly practical benchmark that directly contested a compiler's complexity against its performance. Under the self- compilation speed benchmark, only those optimizations were allowed to be incorporated into a compiler that accelerated it by so much that the intrinsic cost of the new code addition was fully compensated. Found through The Computer Language Benchmarks Game site: http://shootout.alioth.debian.org/dont-jump-to-conclusions.php Original paper: Oberon: The Overlooked Jewel <https://docs.google.com/viewer?url=http%3A%2F%2Fwww.ics.uci.edu%2F%7Efranz%…>(pdf) Michael Franz, in L. Boszormenyi, J. Gutknecht, G. Pomberger "The School of Niklaus Wirth" 2000

1 0

Google Summer of Code ?
by Erick Lavoie 28 Apr '11

28 Apr '11

It might be interesting for next year: http://code.google.com/soc/ Erick

1 0

Regex Compilation
by chevalma＠iro.umontreal.ca 27 Apr '11

27 Apr '11

It might be interesting to look at the re2 regex engine source code. They claim: "RE2 is a fast, safe, thread-friendly alternative to backtracking regular expression engines like those used in PCRE, Perl, and Python. It is a C++ library". http://code.google.com/p/re2/ They also have a link to a page about implementing regular expressions efficiently: http://swtch.com/~rsc/regexp/ Probably worth checking out. - Maxime

2 1

Today's meeting
by Marc Feeley 21 Apr '11

21 Apr '11

Here are the notes I took at today's meeting to which I added a few details. Marc - Benchmarks + profiling => Alex + Bruno - identify JS features missing from Tachyon and implement them 1) benchmarks: Dromaeo/SunSpider/V8 2) find other relevant programs (envjs, jQuery, ...) - identify performance bottlenecks in the compiler and generated code 1) develop the profiling tools (statistical, execution frequency, ...) 2) extend current benchmarking infrastructure 3) test against SpiderMonkey, V8 and IE - Stdlib => Olivier + Maxime - RegExp - Date in JS, what can we steal? - Math - Floating point => Eric + Erick - Basic - parse FP numbers - encoding of x87 FPU instructions - flonum representation - add float type and primitives to IR - Advanced - encoding of SSE instructions - SSE register allocation

1 0

Re: [Tachyon-list] Rencontre mercredi
by Marc Feeley 20 Apr '11

20 Apr '11

On 2011-04-17, at 11:49 AM, St-Aubin Alexandre wrote: > Bonjour M. Feeley, > > Ce message est pour vous aviser que mon examen final du cours IFT2245 - Systèmes d'exploitation aura lieu ce mercredi entre 14h30 et 17h30 (devrais-je à l'avenir envoyer ce genre de message sur la mailing list Tachyon?). > > Alexandre St-Aubin OK, then can we meet Thursday 2 PM? Marc

6 5

Eric Thivierge
by Marc Feeley 15 Apr '11

15 Apr '11

Dear Tachyoners... we now have a new member, Eric Thivierge, who will be working on the Tachyon project this summer (as part of his IFT3150 "project" course, and then as a summer job). Eric has an exam next week on wednesday, so I think we should move our meeting to Thursday to accommodate him. Can everyone make it at 2 PM Thursday April 21? Marc

3 3

Re: [Tachyon-list] My image loading test files
by Marc Feeley 15 Apr '11

15 Apr '11

[Ahhhrrrgggg! I sent the wrong files... here is a repeat of the message with the correct files... this time from my new GMail account] Maxime, here are the files I used for my image loader tests. They demonstrate the different possibilities for linking (importing C functions to the image, such as gcCollect, and exporting functions to C, such as tachyon_main). The file image.S is the format that should be generated by the image dumper (except for the comments of course). It was written to be portable, and I have tested it on Linux, MacOS X and Windows (MinGW and CYGWIN). That way, an image can be generated on any of those systems without any further tweaking. Marc

1 0

My image loading test files
by Marc Feeley 14 Apr '11

14 Apr '11

Maxime, here are the files I used for my image loader tests. They demonstrate the different possibilities for linking (importing C functions to the image, such as gcCollect, and exporting functions to C, such as tachyon_main). The file image.S is the format that should be generated by the image dumper (except for the comments of course). It was written to be portable, and I have tested it on Linux, MacOS X and Windows (MinGW and CYGWIN). That way, an image can be generated on any of those systems without any further tweaking. Marc

1 0