C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Necessity in comp.lang.c++.performance group?          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Alex Vinokur Guest Posted: Tue Sep 16, 2003 8:34 pm    Post subject: Necessity in comp.lang.c++.performance group? Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. ===================================== Alex Vinokur mailto:alexvn (AT) connect (DOT) to http://mathforum.org/library/view/10978.html ===================================== [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top White Wolf Guest Posted: Wed Sep 17, 2003 10:16 am    Post subject: Re: Necessity in comp.lang.c++.performance group? Alex Vinokur wrote: Quote: Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. It depends what topic fits into this "performance" word. If you talk about implementation specific issues - like how to make an optimizer or that memcpy is slower than std::copy on platform X - I guess that is off topic here. OTOH if you talk about language techniques (idioms) like the Duff's device or how to implement an efficient std::copy detecing POD types... that would be on topic here I guess. And not only that, but people would also be interested. -- WW aka Attila [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Michael Jørgensen Guest Posted: Wed Sep 17, 2003 7:48 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? "Alex Vinokur" <alexvn (AT) bigfoot (DOT) com> wrote Quote: Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. Good suggestion. It seems like a good idea, but I have a number of reservations, too. Many (well, some, at least) performance issues involve the STL. Now, since the STL was designed with performance in mind, one generally has the following equivalence: "Correct" use of STL <=> Optimal performance. In other words, performance discussions involving STL often end up explaining to newbies (and others like me) the correct way to use the STL. (E.g. use for_each() etc. instead of explicit loops.) Such discussions would certainly be of interest to the existing clcm group. On the other hand, it would be nice to have a separate place to discuss more general issues of performance, e.g. is it necessary, why is it necessary, when is it necessary, which tools to use, how to use the tools, common pitfalls, etc. Looking forward to hearing comments from the rest of you. -Michael. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Nils Petter Vaskinn Guest Posted: Wed Sep 17, 2003 7:56 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? On Tue, 16 Sep 2003 16:34:39 -0400, Alex Vinokur wrote: Quote: Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. So platform/compiler specific performance issues/solutions would be on-topic in such a group? Programming techniques to increase performance would obviously be on-topic, but how about usage of compilers? eg: "To improve that example rewrite the code like this and compile with mcppc-s (mythical c plus plus compiler) --do-nasty-things-to-speedup-code flag" How about non-standard C++? EG: "use mcppc-s #pragma speed and the non-c++ speed keyword in your function" Profiling? In line assembly? Is it just performance as in speed, or is other related issues (such as memory footprint) on-topic (even when there is a performance penalty)? -- NPV "Linux is to Lego as Windows is to Fisher Price." - Doctor J Frink [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Julian Smith Guest Posted: Wed Sep 17, 2003 8:12 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? On 16 Sep 2003 16:34:39 -0400 "Alex Vinokur" <alexvn (AT) bigfoot (DOT) com> wrote: Quote: Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. I haven't noticed many threads about performance issues, so it doesn't really see the point of this. But I'd like propose comp.lang.c++.moderated.interminable, for issues like SESE vs SEME. - Julian -- http://www.op59.net [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top White Wolf Guest Posted: Thu Sep 18, 2003 9:55 am    Post subject: Re: Necessity in comp.lang.c++.performance group? Julian Smith wrote: Quote: But I'd like propose comp.lang.c++.moderated.interminable, for issues like SESE vs SEME. I believe there is another solution. With a little tweak of the moderating rules (to stop certain rejections from occuring) we could get Godwin's Law working again. http://info.astrian.net/jargon/terms/g/Godwin_s_Law.html -- WW aka Attila [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Alex Vinokur Guest Posted: Fri Sep 19, 2003 10:13 am    Post subject: Re: Necessity in comp.lang.c++.performance group? "Nils Petter Vaskinn" <no (AT) spam (DOT) for.me.invalid> wrote Quote: On Tue, 16 Sep 2003 16:34:39 -0400, Alex Vinokur wrote: Is it worth creating separate comp.lang.c++.performance group? On the one hand, the C++ performance issue is part and parcel of C++ activity . "To know how fast it works" means "To understand how it works". On the other hand, specific character of the performance topic, its specific techniques and result representation manner quite often turn this issue info off-topic in comp.lang.c++.moderated and comp.lang.c++. So platform/compiler specific performance issues/solutions would be on-topic in such a group? I think it is advisable. Performance of C++program reveals itself as result of interaction between several aspects : * features of C++ language itself (linguistic aspect), * program design (algorithmic aspect), * compiler features, * specific character and features of platform/hardware, * application requirements (the best or acceptable performance) * etc. It would be interesting to know/understand how various approaches and features affect C++-based products. Perhaps, different platforms/compilers should involve writing different critical (from performance point of view) pieces of C++ code. Quote: Programming techniques to increase performance would obviously be on-topic, but how about usage of compilers? eg: "To improve that example rewrite the code like this and compile with mcppc-s (mythical c plus plus compiler) --do-nasty-things-to-speedup-code flag" How about non-standard C++? EG: "use mcppc-s #pragma speed and the non-c++ speed keyword in your function" Yes, it is non-standard, but it is natural environment (of C++) which a C++-sofware developer works in. Quote: Profiling? In line assembly? Is it just performance as in speed, or is other related issues (such as memory footprint) on-topic (even when there is a performance penalty)? Any resources used by a program might be of interest to members of the groups. For instance, resources from very useful 'struct rusage' (non-standard C/C++) which is used in UNIX and UNIX-like environments. ===================================== Alex Vinokur mailto:alexvn (AT) connect (DOT) to http://mathforum.org/library/view/10978.html ===================================== [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Nils Petter Vaskinn Guest Posted: Fri Sep 19, 2003 5:25 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? On Wed, 17 Sep 2003 06:16:38 -0400, White Wolf wrote: Quote: It depends what topic fits into this "performance" word. If you talk about implementation specific issues - like how to make an optimizer or that memcpy is slower than std::copy on platform X - I guess that is off topic here. OTOH if you talk about language techniques (idioms) like the Duff's device or how to implement an efficient std::copy detecing POD types... that would be on topic here I guess. And not only that, but people would also be interested. I think (hope) we'd see crossposts between clc++.performance and clc++ like: (within the ====s) ==== How can I speed up this code (that I have identified as the critical section)? [sample code] I use so and so compiler on this and that platform. NOTE: please remove clc++ from the crossposting if your suggestion is compiler/platform dependant ==== The last two lines beeing important. Would that be acceptable from the clc++ viewpoint? (I think so) regards -- NPV "Linux is to Lego as Windows is to Fisher Price." - Doctor J Frink [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top White Wolf Guest Posted: Sat Sep 20, 2003 10:05 am    Post subject: Re: Necessity in comp.lang.c++.performance group? Nils Petter Vaskinn wrote: Quote: I think (hope) we'd see crossposts between clc++.performance and clc++ like: (within the ====s) I see. Quote: NOTE: please remove clc++ from the crossposting if your suggestion is compiler/platform dependant ==== The last two lines beeing important. Would that be acceptable from the clc++ viewpoint? (I think so) I guess this is up to the moderators. What my feeling is that either that NG should also be moderated or you really really need to agree with the moderators of this one. I mean if the new NG will get a lot of posts and people would keep (accidentally) cross posting too much - it can end up in many many more posts in here, and elevated rejection ratio (compared to that 8 months when I was not posting ) -- WW aka Attila [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Sat Sep 20, 2003 3:19 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? In article <bkfnqd$r6j$1 (AT) phys-news1 (DOT) kolumbus.fi>, White Wolf <wolof (AT) freemail (DOT) hu> writes Quote: Nils Petter Vaskinn wrote: I think (hope) we'd see crossposts between clc++.performance and clc++ like: (within the ====s) I see. NOTE: please remove clc++ from the crossposting if your suggestion is compiler/platform dependant ==== The last two lines beeing important. Would that be acceptable from the clc++ viewpoint? (I think so) I guess this is up to the moderators. What my feeling is that either that NG should also be moderated or you really really need to agree with the moderators of this one. I mean if the new NG will get a lot of posts and people would keep (accidentally) cross posting too much - it can end up in many many more posts in here, and elevated rejection ratio (compared to that 8 months when I was not posting ) I am not sure what the problem is. If a performance issue concerns a single platform/implementation then it clearly ought to be discussed in a specialist forum dedicated to that environment. If a performance issue concerns porting across several platforms then why should it be off topic here? Please note that WG21 is in the process of releasing a substantial TR on performance (it is out for final voting by NBs) so at least WG21 considers performance an on topic issue for C++. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Agent Mulder Guest Posted: Wed Sep 24, 2003 11:30 am    Post subject: Re: Necessity in comp.lang.c++.performance group? <Alex Vinokur> Quote: Is it worth creating separate comp.lang.c++.performance group? / comp.lang.c++.performance will (by definition) be platform specific. That's just what many programmers need when they seek specific advice on a specific issue, but don't (dare to) ask in comp.lang.c++. A 'melting pot' newsgroup where you can post 'live' C++ examples, it would be great. When it is not about your system, you just skip it. A newsgroup called comp.lang.c++.code. It can be unmoderated, but flames will flare when there is no compilable content in the posting. -X int main(){return 0;} [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Wed Sep 24, 2003 10:29 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? In article <bkpus4$e8s$1 (AT) news4 (DOT) tilbu1.nb.home.nl>, Agent Mulder <mbmulder_remove_this_ (AT) home (DOT) nl> writes Quote: Alex Vinokur Is it worth creating separate comp.lang.c++.performance group? / comp.lang.c++.performance will (by definition) be platform specific. Actually it isn't or at least not entirely which is why WG21 has spent four years working on a Technical Report on the issue. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Attila Feher Guest Posted: Wed Sep 24, 2003 10:31 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? Agent Mulder wrote: Quote: Alex Vinokur Is it worth creating separate comp.lang.c++.performance group? / comp.lang.c++.performance will (by definition) be platform specific. That's just what many programmers need when they seek specific advice on a specific issue, but don't (dare to) ask in comp.lang.c++. A 'melting pot' newsgroup where you can post 'live' C++ examples, it would be great. When it is not about your system, you just skip it. A newsgroup called comp.lang.c++.code. It can be unmoderated, but flames will flare when there is no compilable content in the posting. Khm. *Only* platform specific? AFAIK there is a newsgroup nearly for every compiler. Why would we want to make one, which - by definition - would only contain absolutely unrelated issues? Everyone reading it should ignore 90% of the posts. IMHO if such a group will exist, its purpose should be to have threads which *mix* platforms. John Doe Endoe comes up with The Big Idea, knowing platform X and Y. Then compiler implementers of other platforms join in and there is a lively debate if TBI is generally applicable, if yes how, or not, if not why not. IMHO it would not be beneficial to create a newsgroup which mixes platform issues unless there is a common theme, topic and serious crosstalk. Since it would not be about separate platforms and how to go deep in assembly in one or the other, but about the C++ language, and for example how to make a (possibly future) standard library template rock on every platform. And what the platforms can or cannot do will reflect back on the implementation of the template. Quote: int main(){return 0;} SET VERBOSE OFF int main(){ } SET SATISFY LINT ON int main(){ ; } -- Attila aka WW [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top llewelly.at@xmission.dot. Guest Posted: Thu Sep 25, 2003 1:28 pm    Post subject: Re: Necessity in comp.lang.c++.performance group? Francis Glassborow <francis (AT) robinton (DOT) demon.co.uk> writes: Quote: In article <bkpus4$e8s$1 (AT) news4 (DOT) tilbu1.nb.home.nl>, Agent Mulder [email]mbmulder_remove_this_ (AT) home (DOT) nl[/email]> writes Alex Vinokur Is it worth creating separate comp.lang.c++.performance group? / comp.lang.c++.performance will (by definition) be platform specific. Actually it isn't or at least not entirely which is why WG21 has spent four years working on a Technical Report on the issue. There are many useful performance measurements which cannot be reasonably compared unless they are conducted on the same machine. There are some important platform-specific performance issues. I think the above two facts make certain kinds of performance-related discussion unnecessarily cumbersome when platform-specific references are out-lawed. I do believe that most performance issues are common to all platforms, or to a wide range of platforms, but even for these it is desirable to be specific about the conditions under which measurements were taken, including such platform-specific details such as cpu and memory architecture. I think a c++ performance forum should prefer platform-generic performance issues, but not rule out platform-specific issues, and expect the mention of platform-specific details of test environments and / or results. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         C++Talk.NET Forum Index -> C++ Language (Moderated) All times are GMT Page 1 of 1   Jump to: Select a forum C/C++----------------C LanguageC Language (Moderated)C++ Language (Moderated)C++ language (comp.lang.c++)C++ language, library and standards Non-English C/C++ Newsgroups----------------C++ (German)C++ (French)  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.