Salakhetdinov Shamil
mcp2004 at mail.ru
Thu May 23 01:48:27 CDT 2013
Hi Stuart -- The "multi-core scaling" is a logically following from "multi-core programming" - that is what I meant by posting the link for the article ( http://erratasec.blogspot.ru/2013/02/multi-core-scaling-its-not-multi.html#.UZ0l0bVplfB ) a few days ago and then reposting it yesterday. <<< Essentially that article just points out that to scale properly across multiple cores, threads blocking other threads is very inefficient>>> Yes - "just" but this "just" is available for C/C++ programmers only... <<< It says NOTHIING about what different languages/compilers can do and certainly doesn't sugggest that C/C++ has some magic capabilities not avaiable in other languages>>> Here is an excerpt from the article referred above: "You don’t want to mess around with assembly language, especially since you want your code to run on both x86 and ARM. Therefore, compilers let you access these instructions with built-in functions. On gcc, example functions are __sync_fetch_and_add() and __sync_bool_compare_and_swap(). They work just as well on x86 as ARM. Microsoft has similar intrinsics for their compilers." Please reread also the 'Conclusion' part of the article. Thank you. -- Shamil Четверг, 23 мая 2013, 8:12 +10:00 от "Stuart McLachlan" <stuart at lexacorp.com.pg>: >There's a difference between: > >"multi-core programming" which you were talking about: > >> <<< It's possible to do "true" multi-core programming without C/C++ >>> >> Please post some samples for general purpose programming languages and OSes. > >and 'multi-core scaling" which the article talks about. Essentially that article just points out >that to scale properly across multiple cores, threads blocking other threads is very inefficient >and you have to design your application properly. It says NOTHIING about what different >languages/compilers can do and certainly doesn't sugggest that C/C++ has some magic >capabilities not avaiable in other languages > > >-- >Stuart <<< skipped >>> >