similar to: [LLVMdev] LLVM footprint

On Mon, 3 Dec 2007, Benjamin Smedberg wrote: > What is the expected footprint of a tool using the LLVM JIT? Right now it's ~1.5 to 2M for one platform, at least on darwin. > I have created a simple project that uses the LLVM C++ API to JIT calls > to XPCOM method signature... it works well, but the component DLL is > very large (Linux x86-74, 5.8MB optimized and stripped). Is

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Chris Lattner wrote: > Finally, there is still a lot that can be done to reduce code size. For > example, building a JIT links in the .s file printers in, and they have > non-trivial size (big string tables etc). It would be great to refactor > the code to avoid things like this. I wouldn't be surprised if we could > shrink the

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I'm a LLVM newbie who's started experimenting with the JIT, and I have a couple questions relating to LLVM and the runtime JIT: I'm going to post them as separate messages to avoid getting threads tangled up. I want to call a C++ function from a function that was JITted at runtime. I'm starting with HowToUseJIT.cpp as a base for

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Anton Korobeynikov wrote: > Benjamin, > >> It seems that LLVM bytecode doesn't have a builtin type for "pointer-sized >> integer". > You should use TargetData to provide a size of pointer. Sorry, perhaps I was not clear. When JITting code it's not hard to figure out the size of a pointer and use the equivalent

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Joachim Durchholz wrote: | Subject says it all. | | Looking in /usr/include/gnu, I find stubs-64.h and stubs.h, but no | stubs-32.h. | This is probably related to my machine running a 64-bit install of | Ubuntu. I'm not sure whether this should be rectified in Ubuntu or LLVM. You don't have i386 and x86-64 devel packages installed. Ask your

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I've been reading the headers and http://llvm.org/releases/2.1/docs/ProgrammersManual.html and I'm still confused about a pretty fundamental point... who is expected to clean up various objects when we're finished with them, and when? Let's say I've created a module, retrieved a bunch of types, created a function, a basic block,

Benjamin, > It seems that LLVM bytecode doesn't have a builtin type for "pointer-sized > integer". You should use TargetData to provide a size of pointer. -- With best regards, Anton Korobeynikov. Faculty of Mathematics & Mechanics, Saint Petersburg State University.

Subject says it all. Looking in /usr/include/gnu, I find stubs-64.h and stubs.h, but no stubs-32.h. This is probably related to my machine running a 64-bit install of Ubuntu. I'm not sure whether this should be rectified in Ubuntu or LLVM. Regards, Jo

Hello! I'm trying LLVM for generating code and I found something I cannot figure out or find in the documentation: I'm doing something like this to call "sin" from std: std::vector<const Type*> params; params.push_back( Type::FloatTy ); FunctionType *FT = FunctionType::get( Type::FloatTy, params, false); Function *F = new Function( FT, Function::ExternalLinkage,

I got it to work by changing scheduled_end_date= nil to self.scheduled_end_date= nil. I believe these statements should be equivilent. I don''t know Ruby/Rails well enough to know when, if or why that should matter. Can someone educate me please? I want to believe this framework is ready for production work, but silently dropping data is a very big problem. I expect that I''ve

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 There doesn't appear to be a calling convention matching "thiscall"... on gcc targets, the standard calling convention is fine (just pass "this" as the first parameter), but with MSVC this won't work because it passes "this" in %ECX. Is there a workaround for this? Or how hard would it be to add a separate

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 It seems that LLVM bytecode doesn't have a builtin type for "pointer-sized integer". Is there a way to emulate a pointer-size integer type in bytecode without knowing the target processor word size? How hard would it be to add such a type? - --BDS - -- Benjamin Smedberg Platform Guru Mozilla Corporation benjamin at smedbergs.us

PLEASE help me - this is making me insane. Can anyone suggest any reasons for this behavior? It occured previously, then went away and is now back. In Task.rb (where task is an active record subclass) def clear_schedule puts "clearing schedule for task" scheduled_end_date= nil #reset the scheduled end date save halt.

Why? -eric On Sep 20, 2011, at 9:14 AM, Johannes Obermayr wrote: > --- > autoconf/configure.ac | 4 +++- > cmake/modules/LLVMConfig.cmake.in | 3 ++- > configure | 4 +++- > tools/llvm-config/Makefile | 6 ++++++ > tools/llvm-config/llvm-config.in.in | 3 ++- > 5 files changed, 16 insertions(+), 4 deletions(-)

--- autoconf/configure.ac | 4 +++- cmake/modules/LLVMConfig.cmake.in | 3 ++- configure | 4 +++- tools/llvm-config/Makefile | 6 ++++++ tools/llvm-config/llvm-config.in.in | 3 ++- 5 files changed, 16 insertions(+), 4 deletions(-) diff --git a/autoconf/configure.ac b/autoconf/configure.ac index f3e94e8..cd943cc 100644 ---

----- Mensaje original ---- De: Benjamin Smedberg <benjamin at smedbergs.us> Para: LLVM Developers Mailing List <llvmdev at cs.uiuc.edu> Enviado: viernes, 22 de febrero, 2008 22:36:40 Asunto: Re: [LLVMdev] Calling functions -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Álvaro Castro wrote: | Hello! | | | I'm trying LLVM for generating code and I found

On Tuesday 20 September 2011 11:23:05 Eric Christopher wrote: > Why? > > -eric > Because openSUSE and many other linux distributions put all things to /usr/lib (i586) and /usr/lib64 (x86_64) or /usr/lib32. So it is possible to install x86 and x86_64 versions at the same time ... (See also http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-July/042068.html and

Hi there I''m considering Rails for a new web-application planned to be built from scratch. The application has an extensive web front, and in fact most of the application is interfaced through that front and engaged through clients'' HTTP requests (i.e. the conventional way web-applications and "dynamic web-sites" work). However, some essential parts are supposed to

Does anyone know the best way to track time zone information. There doesn''t seem to be much documentation on this. So far it seems like a simple db field like create table accounts ( id int unsigned not null auto_increment, name varchar(50) not null, time_zone varchar(50) not null, ... primary key (id) ) and a class like class Account < AR ...

Hi, I''ve installed coverage (gem install coverage) and try to run tests with -rcoverage option on, but following error occurs: ruby: No such file to load -- coverage (LoadError) I try: ruby -rcoverage mytest.rb 1. I have rubygems in evn var RUBYOPT 2. Gem seems to be installed properly (I can read doc via gems, the files exist in the ../gem/... dirs) Does anyone know what is the