Re: CLANG vs GCC tests of fortran/f2c program
Jakub Lach jakub_lach at mailplus.pl writes: I am more concerned about an aspect of the language the clang tools are written in, namely the use of object-oriented paradigm of c++ (it is a phony paradigm, one that does not exist in nature or reality, which explains the failure rate of C++ OO projects historically and current usage decline). I sense that the relative slowness of generated code has to do with it. Perhaps some other attributes of that code's quality too, even if not now, then in the future. Yes, this is one thing really puzzled me. Maybe it's related to Apple's affinity to Objective-C? Well, let me add some more and important facts to this discussion. If it caused so much emotions and name calling, then at least everybody should know what this is all about. Clang is a compiler front-end for C, C++, Objective-C and Objective-C++ programming languages and it uses LLVM as its back-end. Both, clang and LLVM, are written in C++. LLVM provides middle layers of compilation process and is e.g. responsible for optimization of intermediate code, which next will be converted and linked into machine-dependent assembly code. Based on this source http://en.wikipedia.org/wiki/Objective-C the Objective-C was influenced by Smalltalk's object-oriented programming model, while C++ by Simula's. This has implications for characteristics and performance of Objective-C, for example: - there are quite few important language elements in C++ that are not in Objective-C, like namespaces, multiple inheritance, operator overloading, etc - ... Objective-C applications tend to be larger than similar C or C++ applications because Objective-C dynamic typing does not allow methods to be stripped or inlined. - ... Because Objective-C uses dynamic runtime typing and because all method calls are function calls (or, in some cases, syscalls), many common performance optimizations cannot be applied to Objective-C methods (for example: inlining, constant propagation, interprocedural optimizations, and scalar replacement of aggregates). This limits the performance of Objective-C abstractions relative to similar abstractions in languages such as C++ where such optimizations are possible. - ... Objective-C is decidedly geared toward run-time decisions while C++ is geared toward compile-time decisions. The tension between dynamic and static programming involves many of the classic trade-offs in programming: dynamic features add flexibility, static features add speed and type checking. My Note: please keep in mind we are talking about language used for writing clang, a compiler tool. So, Objective-C has disadvantage with regard to size od generated code, performance, and optimization as compared to C++. But both share OO (object-oriented) paradigm, which many pros consider synthetic, or pulled out of thin air if you prefer, with negative effects on devs mental health, design, and resulting code quality. I hope I got all facts right -:) It seems to me that switching to clang was a correct strategic decision for reasons linked to GPLv3 license as described in my prior post and by other thread posters. But there seems to be some price paid related to written in C++ facts described by me in both posts, which may make some people come to a conclusion that the decision was based more on a political factor (Apple) than on technical merits. Because I did not participate or followed FreeBSD's internal process, I can not express any opinion to what extend both factors were considered and discussed. OK. Judge for yourselves, and have fun. jb ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
programming involves many of the classic trade-offs in programming: dynamic features add flexibility, static features add speed and type checking. My Note: please keep in mind we are talking about language used for writing clang, a compiler tool. So, Objective-C has disadvantage with regard to size od generated code, performance, and optimization as compared to C++. But both share OO (object-oriented) paradigm, which many pros consider synthetic, or pulled out of thin air if you prefer, with negative effects on devs mental health, design, and resulting code quality. I hope I got all facts right -:) most probably, but what does it mean if clang have multiple layers, frontend, LLVM backend, etc. etc. for normal user who just needs C compiler. It doesn't matter how it do this but what are the results. It seems to me that switching to clang was a correct strategic decision for reasons linked to GPLv3 license as described in my prior post and by other thread posters. You didn't wrote anything new here. But there seems to be some price paid related to written in C++ facts described by me in both posts, which may make some people come to a conclusion that the decision was based more on a political factor (Apple) than on technical merits. It doesn't really care how clang is written but how it works. And it was political decision because compiler itself, on GPLv3 licence, does not block anyhow distributing it's output - binaries. C++ libraries can be limiting, but... wasn't replaced. If it would be truly about removing GPLv3 code that hurts, replacing libstdc++ would be first thing to do. For now we have removed GPL code that doesn't hurt ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On 25/06/2012 13:56, Wojciech Puchar wrote: C++ libraries can be limiting, but... wasn't replaced. If it would be truly about removing GPLv3 code that hurts, replacing libstdc++ would be first thing to do. I assume you mean like the new libc++? http://wiki.freebsd.org/NewC%2B%2BStack For now we have removed GPL code that doesn't hurt ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
If it would be truly about removing GPLv3 code that hurts, replacing libstdc++ would be first thing to do. I assume you mean like the new libc++? http://wiki.freebsd.org/NewC%2B%2BStack yes. this is actually GREAT MOVE! even if it's slower, object oriented languages are not about speed anyway. This should be done first, not compiler. Compiler - only after actually better would exist. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Chad Perrin perrin at apotheon.com writes: Anyway, switching from GCC to Clang has essentially nothing to do with the kinds of problems we increasingly see in the Linux world. In fact, one of the biggest problems in the Linux world is the fact that GNU projects have a tendency to degrade in quality over time and pretty thoroughly undermine the Unix philosophy in egregious ways, which means that the sooner we can divest ourselves of GNU tools (including GCC) the better off we will probably be (though I would still advocate a measured approach to replacing GNU tools, rather than a headlong rush without any forethought). Some Linux distros are starting to integrate clang into their environment. GPL had been introduced to address certain philosophical problems felt by devs, namely how to make sure that their products do not get hijacked by people who do not want to contribute back (because they are unscrupulous, or do not share with devs the same values). Fair enough. GPL has not been challenged in court of law perhaps for a good reason yet. One could be that its enemies are content with what it represents in their view and so why should they bother trying to protect its followers from hurting themselves ? They believe that GPL is viral, but not fascist yet and anybody of different philosophy can just ignore it and go their own licensing way. Another could be that its enemies want it to reach a critical mass of participation, so when they hit, it will be big time and hopefully deadly. Things like that need time, and perhaps some errors by GPL advocates. It is also possible (why not ?) that GPL will self-destruct - there are indications that they are going down, down, down as a choice of license. I believe GPL can be confusing in its interpretation - even its followers are confused, and that's the seeds of destructions as I see it (btw, I have participated in some discussion of GPL that made me and other participant feel its current interpretation is vulnerable, and so with potential significant effect on a business model dependent on it.). GPLv3 apparently introduced a very sensitive question: Does GPLv3 apply to gcc only, or it *may* apply to code generated by gcc as well ? and until it gets answered any responsible dev or business should stay away from it. If so, then it follows that the dev or the bussiness do not want to live in a LaLaLand according to GPL; they want to conduct activities based on sound rules - who in her right mind would blame them for that ? So, you cut the GPL cord now and go your own way, the sooner the better. This is the most important factor, in any business - clarity of contract, license law. I would strongly disagree with Wojciech's assertion that only performance of compiler generated code really counts. From the software purchaser's point of view, yes, she wants the app that was paid for fast, let's be realistic. But if you consider devs or software houses, that will be more subtle. I think the design and maintainability of compiler source code decides about quality of that tool in a significant way, equally or perhaps even little more than its speed or that of generated code alone - after all it is a tool on which software product depends now and in the future. And this translates into e.g. maintenance costs, a matter of interest to purchaser of software and others, in general the so called users. I am more concerned about an aspect of the language the clang tools are written in, namely the use of object-oriented paradigm of c++ (it is a phony paradigm, one that does not exist in nature or reality, which explains the failure rate of C++ OO projects historically and current usage decline). I sense that the relative slowness of generated code has to do with it. Perhaps some other attributes of that code's quality too, even if not now, then in the future. As one can see even by the test results done in 2010: http://www.phoronix.com/scan.php?page=articleitem=gcc_llvm_clangnum=1 clang looked good vice gcc. There was an improvement since then, e.g.: http://www.phoronix.com/scan.php?page=news_itempx=MTA5Nzc Once again, in a matter like choice of tools you make your living with, you should not gamble - the more clarity you have the better. Once you decide what your philosophy, license model, and business requiremnts are, you just go and do not look back. The choice was made (perhaps GPL helped make that choice ...) - it is clang compiler tools. jb ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
I am more concerned about an aspect of the language the clang tools are written in, namely the use of object-oriented paradigm of c++ (it is a phony paradigm, one that does not exist in nature or reality, which explains the failure rate of C++ OO projects historically and current usage decline). I sense that the relative slowness of generated code has to do with it. Perhaps some other attributes of that code's quality too, even if not now, then in the future. Yes, this is one thing really puzzled me. Maybe it's related to Apple's affinity to Objective-C? -- View this message in context: http://freebsd.1045724.n5.nabble.com/Why-Clang-tp5715861p5721495.html Sent from the freebsd-questions mailing list archive at Nabble.com. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Chad Perrin wrote: Someone in this extended discussion mentioned that there are efforts underway to make sure the base system will compile cleanly with both Clang and GCC 4.2+, so I think you're just making up complaints here. Someone (other than Wojciech Puchar, who would just be talking out of his ass) correct me if I'm mistaken. That was me. I don't have pure facts but I read svn logs daily. Today we have a bunch of: r237428 | eadler | 2012-06-22 08:48:53 +0300 (пт, 22 чер 2012) | 5 lines MFC r237253: Remove variables which are initialized but never used thereafter reported by gcc46 warning Approved by:cperciva (implicit) So at least there are some people working on polishing CURRENT/STABLE up to the point it will build with gcc46. -- Sphinx of black quartz judge my vow. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
underway to make sure the base system will compile cleanly with both Clang and GCC 4.2+, so I think you're just making up complaints here. Someone (other than Wojciech Puchar, who would just be talking out of his once again personal attacks from unhappy childs. ass) correct me if I'm mistaken. reported by gcc46 warning Approved by:cperciva (implicit) So at least there are some people working on polishing CURRENT/STABLE up to the point it will build with gcc46. sounds good. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Thu, Jun 21, 2012 at 12:26 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: i would recommend you to take more care about yourself, and not me. You are not in the right position to give advice, young man. -- chs, ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Snippet from Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: I successfully predicted the fall of linux (in quality point of view) years ago, then netbsd - after this and my prediction were good. Now i predict FreeBSD will fall within 2015 time frame. What i mean fall - that it would be better to use older version as long as possible because newer are worse. For now - FreeBSD 6 was an improvement - FreeBSD 7 was an improvement, except first releases but that's normal - FreeBSD 8 was a big improvement in performance and quality. FreeBSD 9 as for now: - have similar performance at most - have some improvement and important functionality like TRIM support. - have some useful functionality like softdep journalling, but risky. Still - forcing full check reveals some inconsistencies now and then. FreeBSD 10 will unlikely be better, but for sure slower unless you will force gcc build that MAYBE will work. possibly not. My experience with NetBSD suggests you may be right there, but Linux? I'll have to build a new Linux installation and see for myself! I'm still inclined to say FreeBSD 9.0 is an improvement over 8.2; I never got to 8.3. Tom ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
force gcc build that MAYBE will work. possibly not. My experience with NetBSD suggests you may be right there, but Linux? After commercial support got too much about directing decisions, NetBSD got very quickly useless. I'll have to build a new Linux installation and see for myself! Warning: You may not go through it healthy. I'm still inclined to say FreeBSD 9.0 is an improvement over 8.2; I never got to 8.3. There are some new functionality. rctl may be very very very useful. But as for speed - i don't see it to be better, and at high disk I/O load it seems to get somehow longer stalls but it is subjective, no precise test done. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Thu, Jun 21, 2012 at 05:50:24AM -0400, Thomas Mueller wrote: Snippet from Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: I successfully predicted the fall of linux (in quality point of view) years ago, then netbsd - after this and my prediction were good. Now i predict FreeBSD will fall within 2015 time frame. What i mean fall - that it would be better to use older version as long as possible because newer are worse. For now - FreeBSD 6 was an improvement - FreeBSD 7 was an improvement, except first releases but that's normal - FreeBSD 8 was a big improvement in performance and quality. FreeBSD 9 as for now: - have similar performance at most - have some improvement and important functionality like TRIM support. - have some useful functionality like softdep journalling, but risky. Still - forcing full check reveals some inconsistencies now and then. FreeBSD 10 will unlikely be better, but for sure slower unless you will force gcc build that MAYBE will work. possibly not. My experience with NetBSD suggests you may be right there, but Linux? I'll have to build a new Linux installation and see for myself! I'm still inclined to say FreeBSD 9.0 is an improvement over 8.2; I never got to 8.3. I can definitely vouch for his estimate of the quality of Linux-based OSes, at least in the majority of cases. I primarily used Debian for a while, then went through a transitional period where I gradually phased out Debian, until about half a dozen years was spent entirely Linux-free (apart from the Linux kernel on a couple of embedded consumer devices), during which time I used FreeBSD for everything. Over the course of the last -- well, more than a year, less than 1.5 years -- I have been forced to use a Linux-based system again to get halfway decent graphics support on a laptop I bought without checking hardware compatibility carefully enough. In the meantime, however, I have provided some support for other people using Linux-based systems. During that time, I had occasion to see a Slackware installer hose an entire system (luckily with backups) that was initially intended to be set up as a multi-boot with FreeBSD and MS Windows; Ubuntu get cursed at great length with words like If I wanted to deal with this crap, I'd use Windows!; and similar issues crop up. Even so, installing Debian on my new laptop early last year (and trying to install Arch Linux on it -- which didn't hose anything up, but did fail to detect the free space on the hard drive, and thus failed to install, before I decided it was easier to skip Arch) and using it since then on a regular basis has been an eye-opener. Myriad little stupidities have crept into the system, including such wonders of engineering brilliance as some documentation to the effect that basic system network management tools were no longer guaranteed to work. I have some pretty strong opinions about the way things are getting broken in the Linux world, and some of the reasons this sort of problem is growing, but they're increasingly off-topic for this venue. Suffice to say that I could write a short book about the subject, and still leave a lot of problems unaddressed. Anyway, switching from GCC to Clang has essentially nothing to do with the kinds of problems we increasingly see in the Linux world. In fact, one of the biggest problems in the Linux world is the fact that GNU projects have a tendency to degrade in quality over time and pretty thoroughly undermine the Unix philosophy in egregious ways, which means that the sooner we can divest ourselves of GNU tools (including GCC) the better off we will probably be (though I would still advocate a measured approach to replacing GNU tools, rather than a headlong rush without any forethought). -- Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 7:18 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: Yes, Clang in general produces slower binaries than gcc. Is that in dispute or something? Or is this just repetition in case we didn't hear you the first time? just yesterday i've heard lots of otherwise claim. Try thinking of the transition as a step back to take many steps forward. What exactly step forward it means? For now i see ONLY politics and aggression after pointing out facts. This doesn't look like serious behaviour of serious people. I think that this is a more complicated decision than just choosing the 'fastest' compiler. There are many other variables involved, and of course the decision has a political dimension. Most things do. Diversity and competition are nice attributes to have in a system. Having alternatives allows users choose a compiler based on what criteria they think are important. Users also benefit from the experience, but more importantly, for such non-trivial projects as LLVM, different designs are interesting in themselves. I personally, am looking forward to seeing what the lldb debugger can do. Historically, some of the most important software projects have been themselves disasters, but they've lead people to change the way they think about a problem and lead to later better solutions-- for example MULTICS ;) This is part of the development process. And this can't just happen in a laboratory. LLVM needs projects like FreeBSD to test it and simply be involved. I notice that bitrig, which recently forked from OpenBSD, and which want to be a more progressive operating system will also be swapping to LLVM and Clang. We don't know what possible benefits there will be from the LLVM project. But there will be some. I was a bit frustrated about being stuck with gcc4.2 for a while, and was trying to compile as many ports as possible using gcc4.6 (FreeBSD 8.2). There seemed to be some improvement in performance, but now I don't bother, world is compiled with Clang and the ports are compiled with gcc4.2 and everything works (most of the time.) I'm satisfied with performance. I don't really understand your concerns. I mean unless you're a fairly radical environmentalist and are really concerned about saving every clock-cycle, running a bit slower really isn't that much of a problem most of the time. Or just change your compiler. Will i be able to compile FreeBSD base system with gcc after some time? not sure. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org -- joe gain jacob-burckhardt-str. 16 78464 konstanz germany +49 (0)7531 60389 (...otherwise in ???) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 12:18 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: Yes, Clang in general produces slower binaries than gcc. Is that in dispute or something? Or is this just repetition in case we didn't hear you the first time? just yesterday i've heard lots of otherwise claim. Try thinking of the transition as a step back to take many steps forward. What exactly step forward it means? These are a few: http://clang.llvm.org/comparison.html#gcc And the performance overall in clang is gaining more rapidly than gcc. At it's present rate, it won't be long until your are complaining for clang to be the default if that is your primary objection. Other factors have pushed this change into motion sooner than perhaps desirable for some. However, it is inevitable given the licensing barriers and the project's long term goals. Eliminating, or at least not being dependent on a GNU toolchain. GPL v3 brings with it a whole host problems such as: http://www.tech-faq.com/linux-licensing-in-conflict-with-secure-boot-support.html Those licensing issues may not be an issue for you, but they are for many of the targets FreeBSD wishes to serve so keeping the base system as unpolluted as possible is important. -- Adam Vande More ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
long term goals. Eliminating, or at least not being dependent on a GNU toolchain. GPL v3 brings with it a whole host problems such as: As you already know i don't like GPL very much. As i already said for me GNU is computer communism. But like or not like, i don't prefer my likeness above facts and FreeBSD performance. And the facts are against clang. BUT PLEASE stop offtopic explaining about secure boot problems and answer one clear question: What exactly GPLv3 have wrong that we can use gcc in longer term for FreeBSD system? OK? Can you just answer that simple question clearly? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 1:59 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: OK? Can you just answer that simple question clearly? Yes Wojciech, I can attempt an answer for you. Pay attention, this gets very complex. The decision to move to Clang was motivated by what is best for the project, and not what is best for Wojciech. -- Adam Vande More ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Yes Wojciech, I can attempt an answer for you. Pay attention, this gets very complex. The decision to move to Clang was motivated by what is best for the project, and not what is best for Wojciech. still not stopped personal attacks (last part of last sentence) but lets forget. So please give an answer - not summary. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
The answer is: 1. gcc will still be available through the ports system. 2. The move to clang/llvm as a default compiler will reduce the amount of GPL code in the base system, eventually reducing distribution issues (especially for 3rd parties). 3. clang/llvm provides better error and warning messages, as well as good static code analysis, which helps reduce some classes of bugs and eventually will result in a more reliable FreeBSD system. 4. clang/llvm is improving quickly. 5. clang/llvm is more modular than gcc, although there are plans for gcc to become as modular, it will take time. 6. gcc produces faster code, but clang/llvm will eventually (soon enough) get there. 7. From the reasons above, it makes sense to complete a task sooner rather than later, especially that clang/llvm isn't showing any signs of weakness (lack of development power, etc). 8. There might be more reasons for or against, but I couldn't think of any. On Wed, Jun 20, 2012 at 10:25 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: Yes Wojciech, I can attempt an answer for you. Pay attention, this gets very complex. The decision to move to Clang was motivated by what is best for the project, and not what is best for Wojciech. still not stopped personal attacks (last part of last sentence) but lets forget. So please give an answer - not summary. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
1. gcc will still be available through the ports system. As well as clang is available in ports. not an argument. 2. The move to clang/llvm as a default compiler will reduce the amount of GPL code in the base system, eventually reducing distribution issues (especially for 3rd parties). true. But JUST reducing GPL code should be a target per se. FreeBSD is about performance and quality not politics or religion. We don't want GPL code because it prevents doing binary only distributions and closed source derivatives, which means reducing FREEDOM. But - if you do closed source derivatives you don't need to include C compiler to run it. And you may allow C compiler separately with source included. No problem. 3. clang/llvm provides better error and warning messages, as well as good static code analysis, which helps reduce some classes of bugs and eventually will result in a more reliable FreeBSD system. as with 1 - you may use clang when developing. 4. clang/llvm is improving quickly. When/If it WILL actually improve to be better than gcc it should be imported to FreeBSD. not sooner. 5. clang/llvm is more modular than gcc, although there are plans for gcc to become as modular, it will take time. Doesn't matter how it is written, but how it performs. 6. gcc produces faster code, but clang/llvm will eventually (soon enough) get there. This is your prediction. Not definite fact, mine and other predictions are different. 7. From the reasons above, it makes sense to complete a task sooner rather than later, especially that clang/llvm isn't showing any signs of weakness (lack of development power, etc). NOBODY prevent you and FreeBSD developers to fix things already - so FreeBSD and ports would compile with both compilers. Actually it is good to fix it already, as making programs compile with both means usually fixing non-portability bug which would help using third compiler that may possibly emerge. But this DO NOT require clang to be a part of main system. As well as making it default. 8. There might be more reasons for or against, but I couldn't think of any. Against: All for arguments assumes clang WILL be better. This is a change as less than 2 days before it was stated to already be better. As a comparision - DragonFly BSD is stated to have better ideas that would result in better performing system in a future. But FOR NOW it is much worse performer than FreeBSD that's why i (and lots of other people) does not switch to DragonFly but of course will when(and if) DragonFly BSD will actually be better. And i don't really think this IF will happen at all, but i wish to DragonFly BSD i am wrong. Same should be used for clang. AS LONG as it is not better it should not be imported into base system or worse - used as default. Making decision based on wishes and personal likes instead of technical facts isn't something that anyone should be proud about. As for your words about doing decision good for FreeBSD, not me - it is pure nonsense attacks because anything that is good for FreeBSD quality is good for me, and the reverse. The decision of switching to clang now it shows that ideologic and religious arguments won over technical arguments. The agressive and data-manipulationg reactions of many people on that list shows that above sentence is true. This IMHO much change. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Wojciech Puchar wrote: 5. clang/llvm is more modular than gcc, although there are plans for gcc to become as modular, it will take time. Doesn't matter how it is written, but how it performs. That's a hard one. I remember an error in gcc loop optimizer which makes gcc produce SSE2 opcodes for pre-SSE2 athlon chips. Due to gcc internal design such errors are often seen and almost never patched as you should have eternal knowledge of gcc code. gcc's bugtraq is just a cemetery. Opposing to this ones most fixes to clang touch minimal source lines and minimal set of files. Same should be used for clang. AS LONG as it is not better it should not be imported into base system or worse - used as default. And why you think it's not better then gcc? With gcc I can result in code that will hang locking some parts of system forever, yet with clang the code will break predictably yielding a core and a point on where the debugging should start. That was long ago and I can't correctly remember the PR's are I noted this but that was long ago and helped me to debug ZFS issues a lot. The code that runs faster is not the best one. The code that is predictable and runs as fast as possible is. -- Sphinx of black quartz judge my vow. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
And why you think it's not better then gcc? because - as you already should know - test shows otherwise. As well as FreeBSD running predictable with gcc anyway. Still theory and ideology. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
[ Semi-apologies to all for being blunt, and possibly somewhat offensive. ] [ More tactful approaches have been shown to be ineffective, and Wojceich ] [ has a demonstrated propensity to blather on as though he knows more ] [ about everything than anyone else. ] From: Wojciech Puchar woj...@wojtek.tensor.gdynia.pl Yes Wojciech, I can attempt an answer for you. Pay attention, this gets very complex. The decision to move to Clang was motivated by what is best for the project, and not what is best for Wojciech. still not stopped personal attacks (last part of last sentence) but lets forget. Fact; that was NOT a personal attack. Your entire line of reasoning so far has been about -your- preferences, and things as you see them, for _your_ use. The Project does not make decisions based on what is best for any particular user -- be it 'Wojceich' or _anyone_ else. You admit you are 'not a developer'. That *you* don't see problems, is irrelevant to whether those who _are_ developers do. Your perceptions of problems, or the lack thereof, is similarly immaterial. Those who _do_ do the work have a number of valid issues with GCC, of -long- standing. *Major* users of FreeBSD have serious 'issues' with the GPLv3, based on the opinions received from their professional legal counsel -- your legally uninformed opinion not withstanding. So please give an answer - not summary. What would be the use of *repeating* the _multiple_ valid reasons that, in combination, compelled the Project to make the change? They were already provided, once, far earlier in this thread. You dismissed them, and dragged in 'strawman' reasoning, based on arrogant personal bias and flawed reasoning/analysis. The facts; 1) Your opinion about the choice of the standard compiler doesn't matter. 2) The decision _has_ been made. The only question at this point is when. 3) Nobody 'owes' you an explanation for why the decision was made. Nonetheless, you _were_ given an outline of the multiple factors that went into the decision. 4) In your personal view, you didn't find those reasons compelling. Too bad for you. But -irrelevant- to the decision process. 5) You _are_ 'free' to use GCC for anything you want, now or in the future. Nobody is under any obligation to make it particularly 'easy' for you. 6) In the future, to use GCC you may have to do lots of code fix-ups on base-system components -- to work around situations where GCC generates *BAD*CODE* from standards-compliant source, and/or where GNU has introduced 'extensions' that are incompatible with standards-compliant code. That _is_ your choice, and your problem. The Project has chosen not to spend any more time working around those _growing_ deficiencies in GCC. You have stated that you are 'not a developer' -- that means that you are _not_competent_ to have an opinion with regard to the magnitude of problems the 'non standards compliant' behavior of all even remotely- recent versions of GCC causes. 7) *Regardless* of your non-professional opinion of the GPLv3, it is a undisputed fact that it is 'unacceptable' to many large-scale users (and paying supporters of FreeBSD), based on the opinions of their PAID, PROFESSIONAL, legal counsel. 8) Keeping Wojciech happy, at the 'cost' of *all* the problems that using newer versions of GCC brings to the Project, it's staff, and it's _major_ users, is simply 'not worth it.' Live with it. Your ongoing bitching and moaning about the already-made decision is *NOT* going to change anything. And is getting tiresome to listen to. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
still not stopped personal attacks (last part of last sentence) but lets forget. Fact; that was NOT a personal attack. Your entire line of reasoning so far has been about -your- preferences, and things as you see them, for _your_ What is specifically my preference? 1) Your opinion about the choice of the standard compiler doesn't matter. Once more - messing with my words and you know this. I am saying that it doesn't matter others than performance. Clang performance is just bad. 2) The decision _has_ been made. The only question at this point is when. And can be reversed because it is faulty. I successfully predicted the fall of linux (in quality point of view) years ago, then netbsd - after this and my prediction were good. Now i predict FreeBSD will fall within 2015 time frame. What i mean fall - that it would be better to use older version as long as possible because newer are worse. For now - FreeBSD 6 was an improvement - FreeBSD 7 was an improvement, except first releases but that's normal - FreeBSD 8 was a big improvement in performance and quality. FreeBSD 9 as for now: - have similar performance at most - have some improvement and important functionality like TRIM support. - have some useful functionality like softdep journalling, but risky. Still - forcing full check reveals some inconsistencies now and then. FreeBSD 10 will unlikely be better, but for sure slower unless you will force gcc build that MAYBE will work. possibly not. So now there will be more and more backports done by users just for new drivers until something that replace FreeBSD will be available. Assuming there will at all. Wish i am wrong. Twice i wasn't ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Wojciech Puchar wrote: And why you think it's not better then gcc? because - as you already should know - test shows otherwise. Test show only that clang-compiled binaries are still subject for improvement. It doesn't show how strict and clear this binary is. As well as FreeBSD running predictable with gcc anyway. You mileage may vary. I'm using clang-compiled world ports on production servers since clang was added to the ports. And nothing bad happens to me. Still theory and ideology. That what you do too. You are stating clang is less potent only by counting speed estimates. You leave aside things like standard compliance, ease of use and healthy ecosystem. Besides, NetBSD and OpenBSD has already selected and using pcc now. And they are fine with that one. -- Sphinx of black quartz judge my vow. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
speed estimates. there are a difference between speed estimate and actual speed - and i talk about the latter only. Besides, NetBSD and OpenBSD has already selected and using pcc now. And they are fine with that one. their problem. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Wojciech, Why not make FreeBSD better for everyone by cooperating with the CLANG project? 1. Find simple programs with severe performance issues 2. Report to the CLANG developers 3. They fix, tweak, and tune the compiler 4. FreeBSD imports latest release 5. Everybody wins ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Why not make FreeBSD better for everyone by cooperating with the CLANG project? because we already have great compiler - GCC. In spite of using GPL licence. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 12:14:09PM +0200, Wojciech Puchar wrote: And why you think it's not better then gcc? because - as you already should know - test shows otherwise. You just ignored everything Volodymyr Kostyrko said about the other factors that are also important for a compiler being considered better. Good job. I have a hint to share with you, though: Ignoring an argument does not make it wrong. As well as FreeBSD running predictable with gcc anyway. . . . for some use cases, evidently including yours. In my case, Clang's stability and predictability is better than GCC's, and in some other cases it may be *much* better. In the cases where it isn't, that's a case of standards-noncompliant code in a port causing problems, and it is a problem that is being fixed prior to FreeBSD 10 release with Clang as the sole compiler in the base system (last I heard). This is what happens when you use a more standards-compliant compiler: you get more stable and predictable behavior. -- Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 02:16:43PM +0200, Wojciech Puchar wrote: speed estimates. there are a difference between speed estimate and actual speed - and i talk about the latter only. You're talking about poorly managed benchmarks that are imprecise and prone to fluctuation, applying only to very specific cases that are not necessarily very broadly representative, but you are talking about them as though they are perfectly representative of all cases. That is, at best, speed estimates. Besides, NetBSD and OpenBSD has already selected and using pcc now. And they are fine with that one. their problem. No -- it's their solution. It would be a problem only if the previous statement said and they are *not* fine with that one. -- Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Besides, NetBSD and OpenBSD has already selected and using pcc now. And they are fine with that one. I wish that or something like that were true, but pcc is dead even in OpenBSD packages/ports. There was just some discussion on misc@ I am hoping for the day gcc is only used on Linux and many free compilers are used everywhere else. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
I wish that or something like that were true, but pcc is dead even in OpenBSD packages/ports. There was just some discussion on misc@ I am hoping for the day gcc is only used on Linux and many free compilers are used everywhere else. me too. but first we need to have Free compiler that would be at least comparable with gcc in resulting code. Actually i would like to see that even linux migrates out of GNU communism. For now - as i've read in many places, less than 50% of newly developed open source software use GPL licence. It was 95% not long time ago. Good. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 7:18 AM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: Will i be able to compile FreeBSD base system with gcc after some time? not sure. Why is that so important for you? -- chs, ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
woj...@wojtek.tensor.gdynia.pl wrote: Will i be able to compile FreeBSD base system with gcc after some time? not sure. Why is that so important for you? if you would read even less than carefully the topic you will get the answer. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 10:07 PM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: woj...@wojtek.tensor.gdynia.pl wrote: Will i be able to compile FreeBSD base system with gcc after some time? not sure. Why is that so important for you? if you would read even less than carefully the topic you will get the answer. No, I don't. And don't patronize me that way. You'll loose. -- chs, ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 10:07:09PM +0200, Wojciech Puchar wrote: email elided for purposes of courtesy wrote: Will i be able to compile FreeBSD base system with gcc after some time? not sure. Why is that so important for you? if you would read even less than carefully the topic you will get the answer. I'll try to help out, here. Christer Solskogen: I think the reason that is so very important to Wojciech Puchar is the fact that he is incapable of imagining: 1. other concerns that might apply 2. that things appear highly likely to change 3. that a negligible performance difference is . . . negligible I'm pretty sure he's not running compute clusters on FreeBSD, after all. -- Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Wed, Jun 20, 2012 at 08:40:56PM +0400, Евгений Лактанов wrote: 20.06.2012 18:47, Mark Felder пишет: On Wed, 20 Jun 2012 09:43:14 -0500, Wojciech Puchar email address elided for purposes of courtesy wrote: [attribution lost by Wojciech Puchar and I'm too lazy to check] Why not make FreeBSD better for everyone by cooperating with the CLANG project? because we already have great compiler - GCC. In spite of using GPL licence. GCC performs well, but it is a very messy undocumented codebase which makes maintaining it a nightmare. Just ask Google -- you'll find many others saying the same thing. It would take MORE work to get FreeBSD devs up to speed on the GCC codebase to add the features we want than it is to cooperate with the CLANG community and help them make their compiler better than GCC in every test case. It is the classic developer/user argument. It is also stupid. The user side simply doesn't have the same needs, it can't understand how freaking hard it is sometimes to debug a large and complex program in a badly documented environment or worse with undocumented features. If it works faster ergo it is better - that is the only criteria to really have a meaning to a user. It's bikeshed painting. Someone who doesn't understand the many factors that apply, and doesn't even *want* to know, picks one thing he thinks he understands and argues about it in an attempt to make the entire project change course. Well, dammit, I *like* blue, and he can take his bucket of red paint home with him to paint his *own* bikeshed. -- Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ] ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
21.06.2012 01:14, Chad Perrin пишет: On Wed, Jun 20, 2012 at 08:40:56PM +0400, Евгений Лактанов wrote: 20.06.2012 18:47, Mark Felder пишет: On Wed, 20 Jun 2012 09:43:14 -0500, Wojciech Puchar email address elided for purposes of courtesy wrote: [attribution lost by Wojciech Puchar and I'm too lazy to check] Why not make FreeBSD better for everyone by cooperating with the CLANG project? because we already have great compiler - GCC. In spite of using GPL licence. GCC performs well, but it is a very messy undocumented codebase which makes maintaining it a nightmare. Just ask Google -- you'll find many others saying the same thing. It would take MORE work to get FreeBSD devs up to speed on the GCC codebase to add the features we want than it is to cooperate with the CLANG community and help them make their compiler better than GCC in every test case. It is the classic developer/user argument. It is also stupid. The user side simply doesn't have the same needs, it can't understand how freaking hard it is sometimes to debug a large and complex program in a badly documented environment or worse with undocumented features. If it works faster ergo it is better - that is the only criteria to really have a meaning to a user. It's bikeshed painting. Someone who doesn't understand the many factors that apply, and doesn't even *want* to know, picks one thing he thinks he understands and argues about it in an attempt to make the entire project change course. Well, dammit, I *like* blue, and he can take his bucket of red paint home with him to paint his *own* bikeshed. Haven't heard it described like this, but appropriate. Also the Danth's law applies always) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
the answer. I'll try to help out, here. Christer Solskogen: I think the reason that is so very important to Wojciech Puchar is the fact that he is incapable of imagining: 1. other concerns that might apply 2. that things appear highly likely to change 3. that a negligible performance difference is . . . negligible I'm pretty sure he's not running compute clusters on FreeBSD, after all. i would recommend you to take more care about yourself, and not me. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
21.06.2012 02:26, Wojciech Puchar пишет: the answer. I'll try to help out, here. Christer Solskogen: I think the reason that is so very important to Wojciech Puchar is the fact that he is incapable of imagining: 1. other concerns that might apply 2. that things appear highly likely to change 3. that a negligible performance difference is . . . negligible I'm pretty sure he's not running compute clusters on FreeBSD, after all. i would recommend you to take more care about yourself, and not me. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org Watch out, we got a badass over here ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
CLANG vs GCC tests of fortran/f2c program
i tested your test program, and in that case, contrary to testing common unix programs, difference is far higher showing gcc superiority. i did this test with FreeBSD 9 supplied clang and FreeBSD 9 supplied gcc. clearly shows that clang actually cannot do more agressive optimization (that trades space) at all, and at -O2 is far slower. produced: -rwxr-xr-x 1 tmp tmp 11168 20 cze 06:18 test.cc.O2 -rwxr-xr-x 1 tmp tmp 17024 20 cze 06:18 test.cc.O3 -rwxr-xr-x 1 tmp tmp 17024 20 cze 06:18 test.cc.O9 -rwxr-xr-x 1 tmp tmp 11096 20 cze 06:18 test.clang.O2 -rwxr-xr-x 1 tmp tmp 11096 20 cze 06:18 test.clang.O3 cc.O2: real0m2.877s user0m2.829s sys 0m0.030s cc.O3: real0m2.142s user0m2.131s sys 0m0.000s cc.09: real0m2.071s user0m2.054s sys 0m0.008s clang.O2: real0m3.440s user0m3.405s sys 0m0.018s clang.O3: real0m3.217s user0m3.205s sys 0m0.001s How about leaving politics and getting back to technical grounds? From what i know now GPLv3 isn't really a problem for us, your may freely distribute binary only software compiled by latest gcc. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Hi, On Wednesday 20 June 2012 11:26:13 Wojciech Puchar wrote: How about leaving politics and getting back to technical grounds? what is the problem as long as gcc is in the ports tree? Erich ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
How about leaving politics and getting back to technical grounds? what is the problem as long as gcc is in the ports tree? what is a problem as clang is in the ports tree? the problem is that these compilers are not 100% compatible and soon if clang will be default it will be not just easy to build freebsd with gcc. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
On Tue, Jun 19, 2012 at 11:26 PM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: i tested your test program, and in that case, contrary to testing common unix programs, difference is far higher showing gcc superiority. i did this test with FreeBSD 9 supplied clang and FreeBSD 9 supplied gcc. clearly shows that clang actually cannot do more agressive optimization (that trades space) at all, and at -O2 is far slower. Yes, Clang in general produces slower binaries than gcc. Is that in dispute or something? Or is this just repetition in case we didn't hear you the first time? Try thinking of the transition as a step back to take many steps forward. Or just change your compiler. Complaining on this list is definitely the wrong place though. Those who have offended your sensibilities by moving to Clang don't live here. People have already done nice work on the benchmarks: http://blog.vx.sk/archives/25-FreeBSD-Compiler-Benchmark-gcc-base-vs-gcc-ports-vs-clang.html -- Adam Vande More ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Yes, Clang in general produces slower binaries than gcc. Is that in dispute or something? Or is this just repetition in case we didn't hear you the first time? just yesterday i've heard lots of otherwise claim. Try thinking of the transition as a step back to take many steps forward. What exactly step forward it means? For now i see ONLY politics and aggression after pointing out facts. This doesn't look like serious behaviour of serious people. Or just change your compiler. Will i be able to compile FreeBSD base system with gcc after some time? not sure. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: CLANG vs GCC tests of fortran/f2c program
Hi, On Wednesday 20 June 2012 11:46:20 Wojciech Puchar wrote: How about leaving politics and getting back to technical grounds? what is the problem as long as gcc is in the ports tree? what is a problem as clang is in the ports tree? for the port? It does not make a difference. the problem is that these compilers are not 100% compatible and soon if clang will be default it will be not just easy to build freebsd with gcc. For the kernel? How old is the gcc which comes with the kernel? Why are newer versions not in the base system? Erich ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org